Clinical research Interventional cardiology

Transcription

1 European Heart Journal (2006) 27, doi: /eurheartj/ehi736 Clinical research Interventional cardiology Platelet glycoprotein receptor inhibition as adjunctive treatment during saphenous vein graft stenting: differential effects after randomization to occlusion or filter-based embolic protection Michael Jonas 1, Gregg W. Stone 2, Roxana Mehran 2, James Hermiller 3, Robert Feldman 4, Howard C. Herrmann 5, David A. Cox 6, Richard E. Kuntz 1, Jeffrey J. Popma 1, and Campbell Rogers 1 * for the FilterWire EX Randomized Evaluation (FIRE) Investigators 1 Cardiovascular Division, Brigham and Women s Hospital, 75 Francis Street, Boston, MA 02115, USA; 2 New York Presbyterian Hospital/Cardiovascular Research Foundation, New York, NY, USA; 3 St Vincent Hospital, Indianapolis, IN, USA; 4 Evanston Hospital, Evanston, IL, USA; 5 Hospital of the University of Pennsylvania, Philadelphia, PA, USA; and 6 Mid Carolina Cardiology, Charlotte, NC, USA Received 7 March 2005; revised 23 November 2005; accepted 22 December 2005; online publish-ahead-of-print 16 January 2006 See page 891 for the editorial comment on this article (doi: /eurheartj/ehi797) KEYWORDS Saphenous vein graft; Stent; Glycoprotein antagonists; Embolic protection devices Aims Although embolic protection devices reduce complications during saphenous vein graft (SVG) stenting, adverse events still occur in 10% of patients. antagonists have not been proven effective during SVG intervention. We hypothesized that adjunctive use of these agents might enhance the efficacy of embolic protection devices. Methods and results In the prospective, multicentre FilterWire EX Randomized Evaluation trial, 651 patients undergoing SVG stenting were randomized to either filter-based FilterWire EX or balloon occlusion/aspiration GuardWire embolic protection devices. inhibitor use was at the discretion of the investigator, with randomization stratified by intended use. Patients pre-selected for inhibitor use (n ¼ 345) had higher baseline risk, with increased 30-day major adverse cardiac events (MACE, 13.0 vs. 8.0%, P ¼ 0.03). GuardWire assigned patients treated with inhibitors had higher 30-day MACE compared with those not treated with inhibitors (16.0 vs. 6.3%, P ¼ 0.007). In contrast, MACE in high-risk FilterWire patients treated with inhibitors were similar to their lower risk, untreated counterparts (9.9 vs. 9.5%, P ¼ 0.89). Multivariable analysis detected a borderline significant (P ¼ 0.056) interaction for lower MACE between FilterWire and IIb/ IIIa inhibitor use. Adjustment by the propensity to use inhibitors resulted in a significant (P ¼ 0.023) interaction for lower MACE rates. inhibition in conjunction with FilterWire was associated with less abrupt closure, no reflow, or distal embolization. Conclusion antagonists may improve procedural outcome during SVG stenting in high risk patients, utilizing filter-based embolic protection devices. Introduction Saphenous vein graft (SVG) intervention carries high potential for peri-procedural complications. 1,2 The efficacy of distal embolic protection devices in reducing clinical and angiographic complications in this patient cohort has underscored the pathogenic role of atheromatous and thrombotic particulate debris liberated during SVG stenting. 3,4 In contrast, although the adjunctive use of glycoprotein (GP) IIb/ IIIa antagonists during percutaneous coronary interventions (PCIs) has consistently demonstrated clinical benefit in * Corresponding author. address: crogers@partners.org native coronary arteries, two pooled analyses of randomized clinical trials have failed to reveal any benefit from GP IIb/ IIIa inhibitor use during SVG stenting. 5 7 However, whether GP antagonists have clinical utility in patients undergoing SVG stenting procedures in which distal embolic protection devices are used is unknown. In the SAFER trial [801 patients undergoing SVG stenting randomized to embolic protection using the balloon occlusion and aspiration GuardWire device (Medtronic Inc., Santa Rosa, CA, USA) vs. control], 4 adverse events were more common in patients treated with GP inhibitors, and distal protection with the GuardWire resulted in protection from adverse events whether or not GP inhibitors were administered. When using filter-based embolic protection & The European Society of Cardiology All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 Platelet GP blockade and embolic protection devices 921 devices, filter occlusion during the procedure may produce diminished or absent flow, which may result in ischaemia and peri-procedural complications. 8 Whether GP inhibitors, by preserving filter patency throughout the procedure, may improve outcomes in patients undergoing PCI with filter-based distal protection is unknown. The multicentre, prospective, randomized FilterWire EX Randomized Evaluation (FIRE) trial was the pivotal study in which distal protection with the filter-based FilterWire EX device (Boston Scientific Corp., Natick, MA, USA) was established as non-inferior to the balloon occlusion and aspiration GuardWire in preventing peri-procedural complications during SVG stenting. 9 We therefore studied the FIRE trial database to examine whether GP antagonists would interact differentially with the FilterWire and the Guard- Wire. We hypothesized that adjunctive therapy with IIb/ IIIa inhibitors would improve clinical outcomes with filterbased embolic protection, by reducing intra-procedural complications and enhancing device-related procedural success, in a manner not seen with balloon occlusion devices. Methods Patient population and protocol The present study represents the results of a pre-specified subset analysis of the FIRE trial. Details of the FIRE protocol and results have been previously reported. 9 In summary, patients undergoing planned stenting of one or more eligible de novo SVG lesions were randomized 1:1 to distal protection with either the FilterWire EX or the GuardWire. Major exclusion criteria included acute or recent MI, uncontrollable allergy or contraindication to contrast or any of the study medications, recent cerebrovascular event, serum creatinine.2.5 mg/dl, SVG age,6 months, true aortoostial lesion,10 mm in length, reference vessel diameter,3.5 or.5.5 mm, TIMI-0 flow, lesion within 2.5 cm of the distal anastomosis or.2 cm of relatively straight vessel distal to the lesion not present, unprotected Y-limb branch vessel proximal to the study device, planned use of atherothrombectomy devices, or left ventricular ejection fraction (LVEF),25%. The protocol was approved by the Investigational Review Board at each site, and all patients provided informed written consent. The decision whether to use a GP inhibitor, as the choice of a specific drug, was left to the discretion of the investigator. Randomization was stratified after the operator declared the intention to use or not to use GP inhibitors to assure balanced distribution of the two devices in patients treated with these drugs. Intravenous heparin was administered to maintain the procedural activated clotting time to 300 s (200 s if a inhibitor was used). Aspirin 325 mg was administered to all patients before the procedure and indefinitely thereafter. A 300 mg clopidogrel or 500 mg ticlopidine load was recommended before the procedure but was administered in all cases within 4 h after the procedure and for 4 weeks after stent implantation. CPK-MB levels were measured at least three times before the procedure within the first 24 h, and a 12-lead electrocardiogram was obtained immediately and after discharge. Clinical follow-up after hospital discharge was performed at 30 days. Definitions and endpoints The primary endpoint of the study was the 30-day composite occurrence of major adverse cardiac events (MACE) including death, target vessel revascularization (TVR), or myocardial infarction (MI). An MI was defined as any post-procedural CK-MB elevation to three or more times normal. The development of new, pathological Q-waves in two or more contiguous leads (assessed by a blinded ECG core laboratory) was required to diagnose a Q-wave MI. Device-related adverse events were tracked, including problems with device deployment or removal, retrieval sheath issues, device failure requiring a second system, and inability to establish or maintain distal occlusion or perform aspiration. All primary endpoint events were adjudicated by an independent committee blinded to treatment allocation. Angiographic films were analysed in a central core laboratory using previously validated techniques. 9 Angiographic parameters including TIMI flow, noreflow/abrupt closure/distal embolization, and dissection events were determined by core laboratory analysis. The state of graft disease outside the target lesion was estimated by the SVG degeneration score, defined as the axial length of lumen irregularities or ectasia (.20% lumen irregularity) expressed as a per cent of the entire SVG length. The amount of plaque at risk for embolization at the time of stent placement was projected using the estimated lesion plaque volume, as described by Giugliano et al. 10 This was defined as the volume of a cylinder, whose diameter was equal to the reference vessel diameter and whose length was the core-labmeasured shoulder-to-shoulder lesion length, minus the volume of a cylinder of the same length and a diameter equal to the measured minimal lumen diameter. Plaque volume was determined by the equation: p(lesion length)((rvd/2) 2 2 (MLD/2) 2 ), where p ¼ 3.14, RVD is the reference vessel diameter in millimetres, and MLD is the minimal lumen diameter of the lesion in millimetres. Statistical analysis Categorical variables were compared by Fisher s exact test. Continuous variables are presented as mean + standard deviation and were compared using unpaired t-test. A multivariable stepwise logistic regression analysis was performed to determine the potential influence of inhibitors on outcome of FilterWire and GuardWire patients. Multivariate predictors were chosen with entry/stay criteria of 0.1/0.15. Candidate predictors were gender, age, previous MI, previous CVA/TIA, diabetes, hypertension, renal insufficiency, prior CABG, smoker, stable angina, CCS: 3/4; lesion characteristics: aneurysm, eccentricity, bend.45, calcification, tortuosity present, ulcerated; coronary artery: LCX, LAD, RCA, estimated plaque burden, and SVG degeneration score. inhibitor use was forced into the model. Interaction between randomization arm and GP inhibitors use was examined to determine the presence of a modified treatment effect by GP inhibitors use. To further adjust for the choice to use a GP blocker, a propensity analysis was performed with the use of a saturated multivariable logistic regression model to predict the use of GP blockers. From this model, a continuous propensity score was generated and added to the final stepdown multivariable interaction model of GP inhibitor use and treatment assignment, modified from similar methods used for observational studies. 11,12 All analyses were by intention to treat, and all P-values are two-sided. Statistical significance was determined at the P, 0.05 level. Results Patient population Of 651 patients randomized in the FIRE trial, complete data regarding GP inhibitor use were available for 646 patients (99.2%), who comprise the study population. GP inhibitors were administered to 171 patients (51.5%) in the FilterWire EX arm and 174 patients (53.3%) in the GuardWire arm (P ¼ 0.65). Among patients receiving inhibitors, eptifibatide was used in 56.8% of cases, abciximab in 33.9%, and tirofiban in 9.3%, with no difference in MACE outcomes or bleeding complications. Table 1 displays the clinical and angiographic characteristics

3 922 M. Jonas et al. Table 1 Baseline clinical and angiographic characteristics stratified by inhibitor use (n ¼ 345) No (n ¼ 301) P-value Clinical features Age (years) Male (%) Diabetes (%) Hypertension (%) Hyperlipidaemia (%) Smoking during past year (%) Prior MI (%) Cerebrovascular disease (%) Prior renal failure (%) Revascularization for angina or recent MI (%) Ischaemic ST-T changes on baseline ECG (%) Canadian Heart Association Angina class LVEF (%) SVG age (months) Angiographic features SVG distribution (%) Left anterior descending Left circumflex Right SVG lesion location (%) Proximal Ostial involved Mid Lesion length (mm) Reference vessel diameter (mm) Minimal luminal diameter (mm) Diameter stenosis (%) Baseline TIMI-3 flow (%) Second stent required for lesion (%) Stents per lesion Total stent length (mm) Degeneration score Plaque volume (mm 3 ) of patients treated with against without GP antagonists. The receptor blocker treated patients were younger, but otherwise had significantly higher risk clinical and lesion morphology profiles, including more frequent angina or recent MI, lower rates of baseline TIMI-3 flow, more severe diameter stenosis, and longer lesions. SVG degeneration scores tended to be higher, while lesion plaque volume was significantly increased in antagonists-treated patients. Although patients treated with antagonists were younger, the age of the target vein grafts was similar in the two groups. Baseline clinical and angiographic characteristics in the randomized groups, according to inhibitor use, are shown in Table 2. Clopidogrel pre-treatment.6 h prior to stenting was administered to approximately a quarter of the patients, with no difference among those treated with or without inhibitors. Clinical outcomes and inhibitors FilterWire interaction In the entire study cohort, patients pre-selected to receive GP inhibitors had a higher 30-days MACE rate than those not treated with these agents (13.0 vs. 8.0%, P ¼ 0.039). As seen in Table 3 and Figure 1, patients preselected by the operator for GP inhibitor use and then randomly assigned to the GuardWire had a significantly higher incidence of MACE than in GuardWire patients in whom antagonists were not used (16.0 vs. 6.3%, P ¼ 0.007). In contrast, patients pre-selected for GP IIb/ IIIa inhibitor use who were then randomly assigned to the FilterWire had similar rates of MACE compared with Filter- Wire patients not treated with GP inhibitors (9.9 vs. 9.5% respectively, P ¼ 0.89). Thus, patients with highrisk lesions pre-selected for GP inhibitor tended to have lower 30-day adverse event rates if randomized to the FilterWire rather than the GuardWire (9.9 vs. 16.0% respectively, P ¼ 0.09), whereas lower-risk patients not pre-selected for use had similar MACE rates with both devices (9.5 vs. 6.3% respectively, P ¼ 0.31). The extent of peri-procedural myonecrosis was also markedly increased among patients pre-treated with inhibitors in the GuardWire group, but remained similar between highrisk FilterWire patients treated with inhibitors and their lower-risk counterparts not treated with inhibitors (Table 3).

4 Platelet GP blockade and embolic protection devices 923 Table 2 Baseline clinical and angiographic characteristics in the randomized groups, stratified by inhibitor use FilterWire P-value GuardWire P-value (n ¼ 171) No (n ¼ 158) (n ¼ 174) No (n ¼ 143) Clinical features Age (years) Male (%) Diabetes (%) Hypertension (%) Hyperlipidaemia (%) Smoking during past year (%) Prior MI (%) Cerebrovascular disease (%) Prior renal failure (%) Revascularization for angina or recent MI (%) Baseline ECG ischaemic ST-T changes (%) CCS Angina class LVEF (%) SVG age (months) Clopidogrel treatment.6 h prior to stenting (%) Angiographic features SVG distribution (%) Left anterior descending Left circumflex Right SVG lesion location (%) Proximal Ostial involved Mid Lesion length (mm) Reference vessel diameter (mm) Minimal luminal diameter (mm) Diameter stenosis (%) Second stent required for lesion (%) Stents per lesion Total stent length (mm) Degeneration score Plaque volume (mm 3 ) Table 3 Thirty-day clinical outcomes FilterWire P-value GuardWire P-value (n ¼ 171) No (n ¼ 158) (n ¼ 174) No (n ¼ 143) Composite MACE (%) Death (%) MI (%) Q-wave (%) Non-Q-wave ( %) Post-procedural peak CPK Post-procedural peak CPK-MB TVR (%) Stroke (%) Blood transfusion (%) Vascular complications Haematoma.5 cm (%) False aneurysm (%) AV fistula (%) NA Retroperitoneal bleed (%)

5 924 M. Jonas et al. Figure 1 Procedural outcomes and 30-day MACE rates in patients randomized to GuardWire (upper graph) or FilterWire (lower graph) according to pre-treatment with (black bars) or without (white bars) inhibitors. Outcomes in the GuardWire population were uniformly worse in the group treated with antagonists, with 30-day MACE significantly elevated. In contrast to GuardWire, in the FilterWire group, patients receiving inhibitors did better than their counterparts treated with the FilterWire but without inhibitors, with similar 30-day MACE rates. smoking, lesion length, plaque volume, angiographic morphological characteristics (such as eccentric lesion, bend), thrombus appearance, calcification, SVG score, reference vessel size, and minimum lumen diameter, as well as several squared terms: plaque volume, vein graft age, lesion length, and reference vessel diameter. This saturated propensity model achieved a C-statistic of From this model, we generated a propensity score, and when added to the interaction model of GP inhibitors and treatment assignment, the positive interaction term P-value (FilterWire:GP inhibitor interaction) was improved from to The marginally significant interaction between GP blocker use and FilterWire assignment on 30-day MACE, improved by adjustment for propensity, should be regarded as a secondary, hypothesis generating result of the FIRE trial. Requirements for blood transfusion during hospitalization and up to 30-day follow-up were higher in patients treated with GP inhibitors (7.5 vs. 3.7%, P ¼ 0.02). Vascular complications were similar among patients treated with or without GP inhibitors (Table 3). After 6 months, MACE rates (Table 5) were equivalent between the two groups, with or without GP inhibitors, and were driven in large part by TVR, thus diluting the early periprocedural ischaemic endpoints (more influenced by inhibitor therapy) which had driven 30-day MACE rates. However, interestingly, the 6-month mortality in the FilterWire plus GP inhibitor group was 30% lower than that in the FilterWire alone arm, whereas, in contrast, the GuardWire plus GP inhibitor group had approximately three-fold higher mortality than the GuardWire alone arm. Final stepwise multivariate predictors of 30-day MACE for each treatment assignment (Filterwire or Guardwire) are shown in Table 4. GP inhibitor therapy predicted higher MACE in the GuardWire arm, in accordance with this therapy being administered to high-risk patients. In the FilterWire group, therapy did not predict higher MACE, possibly marking a beneficial effect of GP antagonist therapy in conjunction with FilterWire use. The interaction between GP inhibitors use and the embolic protection device best describes the differential MACE effects of GP inhibitors use on device choice. The final stepwise multivariable model of MACE for the entire randomized cohort consisted of the two main effects: degenerative SVG score and plaque volume (Table 4). This is in accordance with previous modelling from the SAFER (GuardWire) trial, 10 where estimated plaque volume and degenerative SVG score were independent predictors of 30-day MACE. A GP Filterwire interaction term was added which demonstrated a borderline significant positive interaction (P ¼ 0.056) between the FilterWire and GP inhibitor use and lower MACE rates. As the use of GP was not randomly assigned, we adjusted for the propensity to use this therapy by applying a continuous propensity score to the interaction model. The propensity score was derived from a saturated model that included age, sex, vein graft age, diabetes, hyperlipidaemia, hypertension, prior MI, recent cigarette Procedural results As intended by protocol, patients treated with GP inhibitors had lower peak ACTs than their counterparts ( vs sec, P, ), with no difference between those assigned to FilterWire or GuardWire. The angiographic core lab protocol mandated intracoronary administration of vasodilators prior to and after PCI. Nitroglycerin was used in the majority of cases, whereas intracoronary nitroprusside was given in,10% of the cases. Vasodilator use was similar in patients treated with or without GP inhibitors. When compared with patients not treated with GP inhibitors, those treated with inhibitors were more likely to require a second stent for the treated lesion, required a greater number of stents per lesion, and overall greater stent length (Table 1). With both devices, procedural duration was increased and embolic material capture was more frequent in patients treated with GP inhibitors (Table 6). In patients randomized to the FilterWire, GP inhibitor use was associated with a higher rate of successful stent deployment, fewer device-related adverse events (determined by the operator during SVG stenting), and reduced rates of procedural complications (abrupt closure, no reflow, or distal embolization determined by angiographic core lab analysis) than in FilterWire patients not treated with GP inhibitors (Figure 1 and Table 6). No such benefits were noted in GuardWire patients. In patients randomized to FilterWire, every procedural and

6 Platelet GP blockade and embolic protection devices 925 Table 4 Multivariable predictors of 30-day MACE outcomes according to FilterWire or GuardWire embolic protection Multivarable predictors Coefficient Standard error P-value Odds ratio (95% CI) FilterWire a IIB/IIIA use ( ) Estimated plaque burden ( ) Calcification ( ) Tortuosity ( ) GuardWire a IIB/IIIA use ( ) SVG degeneration score ( ) Full randomized cohort (unadjusted) b SVG degeneration score , ( ) Estimated plaque burden ( ) FW GP interaction ( ) Propensity adjustment b SVG degeneration score ( ) Estimated plaque burden ( ) FW GP interaction ( ) a Final stepdown models for the FilterWire and GuardWire assigned cohorts. b Final stepdown model for the entire randomized cohort plus the FiterWire GP inhibitor interaction for the unadjusted and GP inhibitor propensity adjusted models. Table 5 Clinical outcomes at 6 months FilterWire P-value GuardWire P-value (n ¼ 171) No (n ¼ 158) (n ¼ 174) No (n ¼ 143) Composite MACE (%) Death, % MI (%) Q-wave (%) non-q-wave (%) TVR (%) Stroke (%) Table 6 Procedural outcomes and angiographic results FilterWire P-value GuardWire P-value (n ¼ 171) No (n ¼ 158) (n ¼ 174) No (n ¼ 143) Procedure duration (min) Successful delivery of stent (%) Embolic material capture (%) Device-related adverse events (%) Final result Thrombus (%) Dissection (%) Distal embolus (%) No reflow (%) Abrupt closure TIMI 0 2 flow (%)

7 926 M. Jonas et al. angiographic parameter studied was similar to or directionally better in GP inhibitor-treated patients. In contrast, in patients randomized to GuardWire, each parameter was directionally worse in GP inhibitortreated patients (Table 6). Despite the more severe baseline characteristics and higher usage of a second stent among patients receiving GP inhibitors, the frequency of any degree of reduced antegrade flow while the FilterWire was deployed was similar in both patients treated with (10.8%) and without (10.0%) GP receptor blockers. Moreover, angiographic core lab analysis demonstrated that, in Filter- Wire patients in whom reduced antegrade flow did develop, the degree of flow reduction was less severe in the GP inhibitor-treated patients: only 26.3% of GP inhibitor-treated patients with reduced antegrade flow developed TIMI 0 1 flow (with 73.7% having TIMI-2 flow), compared with 76.5% TIMI 0 1 flow and 23.5% TIMI 2 flow in patients without inhibitors (P ¼ 0.002). Despite the lower-risk lesion characteristics, slow flow and ischaemia occurring after FilterWire deployment, just before stent deployment, were more than twice as common in patients not receiving GP inhibitors (9.3 vs. 3.8%, P ¼ 0.049). Discussion The principal findings of this report are (i) in the FIRE trial, patients pre-selected for GP inhibitor therapy manifested higher risk baseline characteristics, greater procedural complexity, and correspondingly higher overall 30-day MACE rates. They also had higher bleeding risk and required more transfusions. (ii) Among patients randomized to distal embolic protection with the GuardWire, MACE rates were higher with GP inhibitors than without. In contrast, among patients randomized to the FilterWire, MACE rates were not higher with GP antagonists than without. (iii) GP inhibitor therapy was associated with superior FilterWire (but not GuardWire) performance, including better preservation of flow through the filter, reduced procedural ischaemia, and reduced occurrence of abrupt closure, no reflow, or distal embolization. GP inhibitors in SVG interventions without distal embolic protection The long-term efficacy of bypass surgery is limited by SVG degeneration, with.50% of such grafts developing significant narrowing or total occlusion within 10 years. 13 Treatment of SVGs with stenting is fraught with high complication rates, including peri-procedural MI, repeat revascularization, and long-term mortality. 2,3 Despite early enthusiasm, potent antiplatelet therapy with GP inhibitors has consistently failed to show benefit during SVG interventions. Two pooled analyses of SVG-stenting patients culled from randomized trials of GP inhibition did not demonstrate improved MACE rates (but did note increased major and minor bleeding rates) in patients treated with GP inhibition. 5 7 This finding provides striking contrast to native coronary artery PCI, where adjunctive inhibition yields well-documented reductions in periprocedural MACE. 14,15 GP inhibitors with distal embolic protection: SAFER and FIRE Friable, soft, lipid-rich atheroma is the hallmark of SVG degeneration. Embolization of this atherothrombotic debris initiates a complex cascade involving physical capillary obstruction, distal spasm, oedema, and local thrombosis, all resulting in decreased or absent microvascular flow. 7,14,16 Mechanical embolic protection devices utilizing filters or balloon occlusion/aspiration systems are designed to recover the atherothrombotic debris dislodged during angioplasty prior to its reaching the distal microcirculation. The SAFER and FIRE trials established the safety and efficacy of balloon occlusion/aspiration and filter-based protection devices as useful adjuncts during SVG intervention. 4,9 In both trials, the use of GP inhibitors was at the discretion of the investigator, with randomization stratified by intention to use GP receptor blockade so that roughly equal numbers of patients in each arm would be treated with GP inhibitors. In the SAFER trial, patients pre-selected for treatment in both the GuardWire assigned and control (unprotected) arms had higher rates of MACE than those not treated with GP inhibitors. As a frankly adverse effect of GP inhibitors on outcomes after SVG intervention has never been identified, this observation is most likely due to selection of a higher-risk cohort to receive antagonists. Historically, defining risk factors for MACE after SVG PCI has been difficult. Recently, Giugliano et al 10 sought to identify determinants of 30-day MACE post-svg PCI using the SAFER trial population. Two novel angiographic markers, extent of graft degeneration and estimated plaque volume in the target lesion, were the only independent correlates of increased 30-day MACE with or without embolic protection. The results of the present analysis of the FIRE trial support the hypothesis that patients preselected for GP inhibitor use were at higher risk for peri-procedural ischaemic complications. They had more frequent angina or recent MI, lower rates of baseline TIMI-3 flow, more severe diameter stenosis, longer lesions, a greater number and length of stents implanted, and more frequent retrieval of embolic debris. SVG degeneration score tended to be increased and lesion plaque volume was significantly higher in the GP antagonisttreated patients. Confirming the findings in SAFER trial, the FIRE trial 30- day MACE rates were higher among patients pre-selected for GP inhibitor use and assigned to GuardWire embolic protection. In contrast, patients pre-selected for GP inhibitor treatment and randomized to the Filter- Wire had 30-day MACE rates comparable to their lower-risk counterparts not pre-selected to receive GP receptor blockade. Antiplatelet therapy with GP inhibitors seemed to improve filter patency and decrease peri-procedural myonecrosis among high-risk FilterWire patients. Despite the sample size limitations of subset analysis, interaction testing demonstrated a trend for a differential effect of GP inhibitors on 30-day MACE in patients treated with the FilterWire vs. the GuardWire. Moreover,

8 Platelet GP blockade and embolic protection devices 927 procedural success rates were higher, stent deployment more often successful, device-related events less common, and intra-procedural embolic complications reduced in FilterWire patients pre-treated with GP inhibitors. These procedural benefits were not realized in the Guard- Wire arm. The primary endpoint of the FIRE trial was 30-day MACE, to capture ischaemic data, whereas the 6-month MACE data are driven in large part by TVR with some ischaemic events likely to be a consequence of disease progression in non-treated SVG segments. The FIRE study was not powered to capture any late difference in MACE derived from early beneficial reduction in ischaemic events. Nonetheless, it is of interest to note the increased mortality trend in high-risk patients treated with GP inhibitors in the GuardWire embolic protection arm, not seen in the equally high-risk inhibitor-treated patients in the FilterWire arm (Table 6). Interaction of GP inhibitors and FilterWire: possible mechanisms Several possible explanations exist for why GP inhibition beneficially affects FilterWire but not GuardWire use. Balloon occlusion devices facilitate embolic retrieval through complete vascular occlusion (requiring several minutes of obligate ischaemia) and aspiration. Adverse events may still occur if wall apposition of the occluding balloon or aspiration are incomplete or if aspiration efficacy is suboptimal. These mechanical conditions are unlikely to be affected by GP inhibitors. In contrast, distal filters permit antegrade perfusion during PCI while collecting liberated particulate debris. Indeed, antegrade flow is necessary to carry the embolic load into the filter. Slow flow and resultant MACE may occur when filters become overloaded with debris or occluded with platelets and fibrin during use. GP inhibitors may reduce platelet aggregation and deposition on the filter surface, thereby preventing filter occlusion. In support of this hypothesis, when reduced antegrade flow did occur during stenting with FilterWire protection in FIRE, the flow in the GP inhibitor-treated group was mostly TIMI-2 flow (which would still allow debris transport into the filter). In contrast, in the absence of GP inhibitors when slow flow did occur, it was more likely to be TIMI 0 1, presumably causing ischaemia and at the same time limiting delivery of emboli into the filter, resulting in debris embolization after filter removal. Among patients not treated with GP inhibition, the lower MACE rate in patients randomly assigned to GuardWire rather than FilterWire (P ¼ 0.31) suggests that not using GP inhibitors may paradoxically prevent the FilterWire from attaining optimal outcomes even in low-risk patients. Limitations The FIRE trial included non-randomized allocation of GP inhibitor use. However, patients were randomized to FilterWire or GuardWire distal protection after stratification for pre-treatment GP inhibitor therapy. Thus, because the effect of GP inhibitor therapy was compared among patients within each device group, any selection bias ought to have had a similar effect on both device cohorts. Our present analysis is a secondary analysis of the randomized FIRE trial. The secondary hypothesis tested was the differential effect of GP use on device performance in terms of avoidance of MACE. The present observation of a differential effect of GP inhibitors on the outcomes of SVG stenting with the two distal protection devices in FIRE should be regarded as hypothesis generating findings, requiring prospective validation by an independent study. The directional concordance of the procedural and clinical results (Figure 1, Tables 3 and 4) is strongly supportive of a synergy between drug and device, although had more patients been enrolled in FIRE, the robustness of many of our observations would have increased. Mechanistic understanding of the potential synergy between FilterWire and GP blocker therapy would be aided by core laboratory examination of the debris retrieved and the filters themselves from patients in each group, but these materials are not available for analysis. Requirements for blood transfusion during hospitalization were higher in patients treated with GP inhibitors. However, lack of GP blocker dosing data does not permit assessment of their impact on bleeding complications, especially in the setting of renal failure. Additionally, the relatively high ACT levels may indicate higher-than-customary use of heparin in combination with GP inhibitors, likely contributing further to increased bleeding. Conclusions and clinical implications In the randomized FIRE trial, operators choose GP inhibitors during SVG stenting in high-risk lesions. The combination of GP inhibitors with FilterWire distal embolic protection did reduce ischaemic complications, whereas the combination of GP inhibitors with GuardWire protection did not. Although transfusion rates were indeed higher in those patients treated with GP IIb/ IIIa inhibitors, higher-than-customary use of heparin and overdosing of GP inhibitors cannot be excluded as contributing to bleeding complications. Even in low-risk patients treated with the FilterWire, MACE rates may be further reduced by utilizing GP inhibitors. Thus, the decision to use GP inhibitors in conjunction with distal protection devices during SVG stenting should be individualized: the benefits of improved procedural and short-term outcomes will require consideration of embolic protection choice, financial costs, and possible haemorrhagic risks. The optimal treatment for SVG stenting may lie in combining FilterWire use and GP inhibitors with attention to dosing regimens aimed at reducing bleeding risk, a hypothesis that requires prospective testing. Conflict of interest: Financial disclosure: J.J.P. and H.C.H. received research funding from Boston Scientific Corp. (BSC); G.W.S., D.A.C., and J.H. serve as consultants to BSC; and C.R. is on the speaker s bureau for BSC. References 1. Hong MK, Mehran R, Dangas G, Mintz GS, Lansky AJ, Pichard AD, Kent KM, Satler LF, Stone GW, Leon MB. Creatine kinase-mb enzyme elevation following successful saphenous vein graft intervention is associated with late mortality. Circulation 1999;100:

9 928 M. Jonas et al. 2. Hong MK, Mehran R, Dangas G, Mintz GS, Lansky A, Kent KM, Pichard AD, Satler LF, Stone GW, Leon MB. Are we making progress with percutaneous saphenous vein graft treatment? A comparison of 1990 to 1994 and 1995 to 1998 results. J Am Coll Cardiol 2001;38: Piana RN, Paik GY, Moscucci M, Cohen DJ, Gibson CM, Kugelmass AD, Carrozza JP Jr, Kuntz RE, Baim DS. Incidence and treatment of noreflow after percutaneous coronary intervention. Circulation 1994;89: Baim DS, Wahr D, George B, Leon MB, Greenberg J, Cutlip DE, Kaya U, Popma JJ, Ho KK, Kuntz RE. Randomized trial of a distal embolic protection device during percutaneous intervention of saphenous vein aorto-coronary bypass grafts. Circulation 2002;105: Ellis SG, Lincoff AM, Miller D, Tcheng JE, Kleiman NS, Kereiakes D, Califf R, Topol EJ. Reduction in complications of angioplasty with abciximab occurs largely independently of baseline lesion morphology. EPIC and EPILOG Investigators. Evaluation of 7E3 for the Prevention of Ischemic Complications. Evaluation of PTCA To Improve Long-term Outcome with abciximab GP Receptor Blockade. J Am Coll Cardiol 1998;32: Roffi M, Mukherjee D, Chew DP, Bhatt DL, Cho L, Robbins MA, Ziada KM, Brennan DM, Ellis SG, Topol EJ. Lack of benefit from intravenous platelet glycoprotein receptor inhibition as adjunctive treatment for percutaneous interventions of aortocoronary bypass grafts: a pooled analysis of five randomized clinical trials. Circulation 2002;106: Kereiakes DJ. Platelet glycoprotein inhibition and atheroembolism during bypass graft angioplasty: a cup half full. Circulation 2002;106: Stone GW, Rogers C, Ramee S, White C, Kuntz RE, Popma JJ, George J, Almany S, Bailey S. Distal filter protection during saphenous vein graft stenting: technical and clinical correlates of efficacy. J Am Coll Cardiol 2002;40: Stone GW, Rogers C, Hermiller J, Feldman R, Hall P, Haber R, Masud A, Cambier P, Caputo RP, Turco M, Kovach R, Brodie B, Herrmann HC, Kuntz RE, Popma JJ, Ramee S, Cox DA. Randomized comparison of distal protection with a filter-based catheter and a balloon occlusion and aspiration system during percutaneous intervention of diseased saphenous vein aorto-coronary bypass grafts. Circulation 2003;108: Giugliano GR, Kuntz RE, Popma JJ, Cutlip DE, Baim DS. Determinants of 30-day adverse events following saphenous vein graft intervention with and without a distal occlusion embolic protection device. Am J Cardiol 2005;95: Rosenbaum P, Rubin DB. Reducing bias in observational studies using subclassification on the propensity score. J Am Statist Assoc 1984;79: D Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17: de Feyter PJ, van Suylen RJ, de Jaegere PP, Topol EJ, Serruys PW. Balloon angioplasty for the treatment of lesions in saphenous vein bypass grafts. J Am Coll Cardiol 1993;21: Kong DF, Califf RM, Miller DP, Moliterno DJ, White HD, Harrington RA, Tcheng JE, Lincoff AM, Hasselblad V, Topol EJ. Clinical outcomes of therapeutic agents that block the platelet glycoprotein integrin in ischemic heart disease. Circulation 1998;98: Kereiakes DJ, Lincoff AM, Anderson KM, Achenbach R, Patel K, Barnathan E, Califf RM, Topol EJ. Abciximab survival advantage following percutaneous coronary intervention is predicted by clinical risk profile. Am J Cardiol 2002;90: Webb JG, Carere RG, Virmani R, Baim D, Teirstein PS, Whitlow P, McQueen C, Kolodgie FD, Buller E, Dodek A, Mancini GB, Oesterle S. - Retrieval and analysis of particulate debris after saphenous vein graft intervention. J Am Coll Cardiol 1999;34: