Columbia University Medical Center Columbia University Medical Center The Cardiovascular Research Foundation The Cardiovascular Research Foundation Comprehensive Meta-Analysis Comprehensive Meta-Analysis of DES vs. BMS of DES vs. BMS Randomized Trials and Randomized Trials and Registries Registries Ajay J. Kirtane, M.D., S.M. Ajay J. Kirtane, M.D., S.M. Gregg W. Stone, M.D. Gregg W. Stone, M.D.
36
Embed
Comprehensive Meta-Analysis of DES vs. BMS Randomized Trials and Registries
Comprehensive Meta-Analysis of DES vs. BMS Randomized Trials and Registries. Ajay J. Kirtane, M.D., S.M. Gregg W. Stone, M.D. Conflict of Interest Disclosure. Ajay J. Kirtane Past honorarium from Boston Scientific Corporation (modest) - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Columbia University Medical CenterColumbia University Medical CenterThe Cardiovascular Research FoundationThe Cardiovascular Research Foundation
Comprehensive Meta-Analysis Comprehensive Meta-Analysis of DES vs. BMS Randomized of DES vs. BMS Randomized
Trials and RegistriesTrials and RegistriesAjay J. Kirtane, M.D., S.M.Ajay J. Kirtane, M.D., S.M.
Gregg W. Stone, M.D.Gregg W. Stone, M.D.
Conflict of Interest Disclosure
• Ajay J. Kirtane Past honorarium from Boston Scientific
Abbott Vascular (modest)• Gregg W. StoneGregg W. Stone
Research grants from Boston Scientific Research grants from Boston Scientific and Abbott Vascularand Abbott Vascular
Background: DES vs. BMS RCTs• In most individual RCTs, DES have reduced the rates of TLR
and TVR compared to BMS, with no significant differences in death or MI
• However, individual RCTs are underpowered to assess low frequency endpoints
• RCTs, particularly the pivotal RCTs leading to regulatory approval, have been criticized for not reflecting “real-world” DES use
• RCT outcomes may vary based upon differences in enrollment criteria (e.g. “on-label” vs. “off-label”), the amount of routine angiographic FU, and with the duration of clinical FU
Background: DES vs. BMS Registries• In order to address issues of both sample size as
well as generalizability to the “real-world”, numerous observational and registry comparisons of DES vs. BMS have been undertaken
• The outcomes from these studies have varied• While more generalizable than RCTs, the DES vs.
BMS registries are heterogeneous, with differences in design and analysis methodology (e.g. adjusted vs. unadjusted, type of adjustment)
• Registry outcomes may also vary based upon the types of patients enrolled (e.g. all comers vs. just ACS or high risk), and the duration of clinical FU
Persistent Questions: DES vs. BMS• While some of the alarm generated after ESC 2006 has
been mitigated by analyses of patient-level data from the “on-label” RCTs*, there remains concern regarding DES outcomes in “off-label” patients and lesions, and with uncontrolled use Are DES safe in higher risk off-label pts and in the
unregulated environment of real-world use? Are the benefits of DES in reducing TVR as robust in the
real-world as in the RCTs, given the impact of routine angio FU and the oculostenotic reflex in many RCTs?
*Stone et al, Kastrati et al, Spaulding et al, Mauri et alN Engl J Med 2007; 356(10).
Methods: Goals and Objectives (1)
• We therefore sought to perform a systematic review and meta-analysis of DES vs. BMS studies To derive summary estimates of all-cause
mortality, MI, and TVR in studies with ≥1 year of follow-up
To specifically assess differences between RCT and registry safety and effiacy with regard to these endpoints
Methods: Goals and Objectives (2)• Randomized Trials
To assess differences between RCTs according to “on-label” vs. “off-label” use, duration of FU, and baseline risk
• Registries / Observational AnalysesTo assess differences in the estimates derived
from registries using unadjusted and adjusted analyses (and according to the types of adjustment)
To assess differences between registries according to duration of FU, and baseline risk
• To assess differences in effect size estimates between the RCTs and registries for each endpoint
Methods: Inclusion Criteria• English language RCTs or registries which English language RCTs or registries which
reported a reported a direct comparison direct comparison of DES of DES (commercialized formulations of SES and PES (commercialized formulations of SES and PES only) vs. BMS. only) vs. BMS.
• Criteria for each study: Criteria for each study: ≥≥100 patients total100 patients totalMortality reported (± MI and/or TVR)Mortality reported (± MI and/or TVR)≥≥1 year of 1 year of cumulativecumulative follow-up reported, with the follow-up reported, with the
outcome assessed at the same time point in both outcome assessed at the same time point in both comparator armscomparator arms
Methods: Exclusion Criteria
• ““DES era” vs. pure “BMS era” studies in which DES era” vs. pure “BMS era” studies in which the DES era group did not exclude BMS ptsthe DES era group did not exclude BMS pts Excepting intent-to-treat RCTsExcepting intent-to-treat RCTs
• Study used a control group from another study Study used a control group from another study already in the meta-analysisalready in the meta-analysis
• Study was itself a meta-analysis (although data Study was itself a meta-analysis (although data abstracted from individual studies in a abstracted from individual studies in a published meta-analysis could be used)published meta-analysis could be used)
Focused: Focused: (eluting stent OR DES OR drug-eluting stent) AND (eluting stent OR DES OR drug-eluting stent) AND (bare OR uncoated OR standard OR BMS) AND (("2002"[PDat] (bare OR uncoated OR standard OR BMS) AND (("2002"[PDat] : "2008"[PDat]) AND (Humans[Mesh]) AND (English[lang])) : "2008"[PDat]) AND (Humans[Mesh]) AND (English[lang])) AND (coronary) NOT (cost-effectiveness) NOT review[pt] NOT AND (coronary) NOT (cost-effectiveness) NOT review[pt] NOT case reports[pt] NOT editorial[pt] NOT comment[pt]case reports[pt] NOT editorial[pt] NOT comment[pt]
Broad: Broad: stent AND bare AND (eluting OR sirolimus OR stent AND bare AND (eluting OR sirolimus OR paclitaxel)paclitaxel)
• Data requested from study PI’s for large registriesData requested from study PI’s for large registries For most updated dataFor most updated data Where not publicly available or to clarify methodologyWhere not publicly available or to clarify methodology
Methods: Analysis• Pre-specified separate analysis of RCTs and Pre-specified separate analysis of RCTs and
Registries performed given clinical heterogeneityRegistries performed given clinical heterogeneity• RCTs: Direct randomization to DES vs. BMSRCTs: Direct randomization to DES vs. BMS• Registries: Non-rand comparison of DES vs. BMS Registries: Non-rand comparison of DES vs. BMS
(including non-rand comparisons within a RCT)(including non-rand comparisons within a RCT)• All analyses cumulativeAll analyses cumulative
• No landmarksNo landmarks• Single time point estimate for each study assuming Single time point estimate for each study assuming
constant hazard of DES vs. BMS through study period constant hazard of DES vs. BMS through study period (hence use of HR or RR as the estimate)(hence use of HR or RR as the estimate)
• Higher quality estimate picked for primary analyses Higher quality estimate picked for primary analyses (adjusted > unadjusted)(adjusted > unadjusted)
Methods: Statistical Analysis• All analyses were performed at The Cardiovascular All analyses were performed at The Cardiovascular
Research Foundation/Columbia UniversityResearch Foundation/Columbia University• Models (both reported):Models (both reported):
• Fixed effects (Inverse-Variance weighted)Fixed effects (Inverse-Variance weighted)• Random effects (DerSimonian and Laird)*Random effects (DerSimonian and Laird)*• Fixed effects model was considered the primary Fixed effects model was considered the primary
model if significant heterogeneity was not model if significant heterogeneity was not present; otherwise random effects was present; otherwise random effects was considered primaryconsidered primary
• Formal heterogeneity testing was performed using Formal heterogeneity testing was performed using the Ithe I22 statistic; heterogeneity was defined as I statistic; heterogeneity was defined as I22 ≥ 25% ≥ 25%
*Weights displayed in figures are based upon the primary model used
• GRACEGRACE Non-landmark data not available (PI contacted as well)Non-landmark data not available (PI contacted as well) Unequal follow-up in comparator arms (data not Unequal follow-up in comparator arms (data not
presented at fixed timepoint)presented at fixed timepoint)• RRISCRRISC
Less than 100 patientsLess than 100 patients• Medicare DataMedicare Data
Comparison of pre-DES era with post-DES rather Comparison of pre-DES era with post-DES rather than DES vs. BMSthan DES vs. BMS
- Fixed effectsFixed effects- Random effectsRandom effects
0.830.830.80*0.80*
0.960.960.89*0.89*
0.570.570.53*0.53*
<1.0 <1.0 DES better DES better
Study LimitationsStudy Limitations• Randomized trial analyses are still underpowered
to assess these clinical endpoints• Registry analyses are based upon observational,
non-randomized analyses• Potential for residual confounding• Significant heterogeneity, despite attempts to
address this through random effects models, meta-regression and sensitivity analyses
• Analysis was primarily of summary-level data and included unpublished studies
• Use of hazard ratio / relative risk assumes constant hazards throughout the FU period
Conclusions (1)Conclusions (1)In 22 RCTs in which 9,470 pts were randomized to DES or BMS and followed for ≥1 yr, DES resulted in:
• Non significant 3% and 6% reductions in mortality and MI respectively• A highly significant 55% reduction in TVR
In 30 registries in which 174,302 pts were treated with either DES or BMS (non-randomized) and followed for ≥1 yr, DES was associated with:
• A highly significant 20% reduction in mortality• A significant 11% reduction in MI• A highly significant 47% reduction in TVR
Conclusions (2)Conclusions (2)The favorable results of DES from the RCT and registry analysis populations were robust and consistent for both on-label and off-label use, and for clinical f/u extending to 3-4 years
These findings, derived from more than 180,000 pts treated in 52 studies, strongly suggest that DES are safe for both on-label and off-label use, and have comparable efficacy in both RCTs and in the “real-world”