Author + information
- Received April 2, 2007
- Revision received May 18, 2007
- Accepted May 21, 2007
- Published online July 17, 2007.
- Paul Vermeersch, MD1,
- Pierfrancesco Agostoni, MD1,⁎ (, )
- Stefan Verheye, MD, PhD,
- Paul Van den Heuvel, MD,
- Carl Convens, MD,
- Frank Van den Branden, MD,
- Glenn Van Langenhove, MD, PhD,
- DELAYED RRISC (Death and Events at Long-term follow-up AnalYsis: Extended Duration of the Reduction of Restenosis In Saphenous vein grafts with Cypher stent) Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Pierfrancesco Agostoni, Antwerp Cardiovascular Institute Middelheim, AZ Middelheim, Lindendreef 1, 2020 Antwerp, Belgium.
Objectives We sought to provide long-term follow-up data of sirolimus-eluting stents (SES) versus bare-metal stents (BMS) in saphenous vein grafts (SVG) from the RRISC (Reduction of Restenosis In Saphenous vein grafts with Cypher) trial.
Background We have previously shown that, in SVG, the use of SES reduces 6-month restenosis and repeated revascularization procedures versus the use of BMS. These data are consistent with trials in native coronary arteries. However, recently published long-term follow-up data of these trials have revealed an increased risk of adverse events (particularly very late stent thrombosis) after SES.
Methods A total of 75 patients with 96 SVG lesions were randomized to SES versus BMS. All patients underwent clinical follow-up up to 3 years. Specific outcomes assessed in this secondary post-hoc analysis were all-cause mortality, myocardial infarction, and target vessel revascularization.
Results Thirty-eight patients received 60 SES for 47 lesions, whereas 37 patients received 54 BMS for 49 lesions. At a median follow-up time of 32 months (interquartile range 26.5 to 36 months), 11 deaths (7 cardiac, of which 1 was caused by very late stent thrombosis and, 3 were sudden) occurred after SES (29% [95% confidence interval (CI) 17% to 45%]) versus 0 after BMS (0% [95% CI 0% to 9%]) with an absolute difference of 29% ([95% CI 14% to 45%], p < 0.001). The rates of myocardial infarction and target vessel revascularization were not different: 18% and 34% after SES, respectively, versus 5% and 38% after BMS, respectively (p = 0.15 and p = 0.74, respectively).
Conclusions In this secondary post-hoc analysis, BMS were associated with lower long-term mortality than SES for SVG disease. Also, the 6-month reduction in repeated revascularization procedures with SES was lost at longer-term follow-up. (RRISC Study: Reduction of Restenosis In Saphenous Vein Grafts With Cypher Sirolimus-Eluting Stent; http://clinicaltrials.gov/ct/show/NCT00263263?order=1; NCT00263263)
Sirolimus-eluting stents (SES) have been shown to reduce 6-month to 1-year major adverse cardiac events as compared with bare-metal stents (BMS) in native coronary arteries, mainly by considerably decreasing the rate of repeated revascularization procedures caused by the restenotic process in the stent (1–5). Conversely, recent data on long-term (>2 years) clinical follow-up have raised the issue of an increased rate of “hard” end points, such as death and myocardial infarction (MI), after SES, potentially due to an augmented risk of late stent thrombosis (6–10) and to an increase in noncardiac-related deaths (11).
Despite this large body of evidence, lesions in saphenous vein grafts (SVG) have been poorly represented if not totally excluded in pivotal drug-eluting stent trials. However, this lesion subset represents a consistent proportion of lesions in which percutaneous procedures are performed, up to 10% to 15% in most centers (12), and effective and satisfactory therapies are still lacking, as periprocedural complications and long-term events remain elevated (12,13).
We have recently shown that the use of SES is effective in reducing 6-month angiographic and clinical parameters of restenosis with respect to BMS in SVG (14). However, because of the lack of long-term data in patients with this type of lesion, and to offer additional information to the current debate on the safety of SES, we performed a clinical follow-up evaluation of the patients enrolled in the RRISC (Reduction of Restenosis In Saphenous vein grafts with Cypher sirolimus-eluting stent) trial up to 3 years, focusing specifically on all-cause mortality.
The RRISC trial is a randomized study approved by the local ethics committee, and its design and main results have been previously described (14). The study was double blind until the moment of primary analysis (14). Blinding was maintained in the follow-up period for patients and referring clinical cardiologists and general practitioners but not for the interventional cardiologists performing the index procedure and analyzing the data. Thus, the current analysis is single blind.
Between September 2003 and November 2004, 75 patients with a history of previous coronary artery bypass surgery and with 96 “de novo” target lesions localized in 80 diseased SVG with a reference vessel diameter >2.5 and <4.0 mm were enrolled. All patients provided written informed consent before the index procedure.
Patients were allocated randomly in a 1:1 ratio to treatment with Cypher SES or BX-Velocity BMS (both from Cordis, Johnson & Johnson, Miami Lakes, Florida). Details on the randomization process have been described (14). Acetylsalicylic acid (100 to 300 mg/day) was given daily and clopidogrel (loading dose of 300 mg, 6 to 48 h before the procedure and 75 mg/day thereafter) was administered for at least 2 months in all patients. Prolonged therapy with clopidogrel or ticlopidine (250 mg twice per day) instead of clopidogrel was left to the decision of the physician taking care of the patient (who was also blinded to the type of stent implanted, as previously described). Clinical follow-up was scheduled at 1 and 6 months. Coronary angiography was repeated at 6 months.
In September 2006, we obtained a new approval from the Ethics Committee at Antwerp Cardiovascular Institute Middelheim to prolong the follow-up duration of the RRISC trial in order to directly query patients about their status and the occurrence of further events after the initial 6 months. We mailed an additional informed consent to all patients (or to the relatives) to obtain permission to use their data. Once we received all of the signed informed consents, beginning in November 2006 we contacted every patient by phone and investigated the occurrence of events such as death, MI, repeated revascularizations, new hospitalizations, and additional ambulatory visits using a prespecified questionnaire. If a positive result for any of the aforementioned events was obtained, we asked the patient (or relatives) to send us the records related to each event. If the patient was hospitalized in our institution, we reviewed our internal clinical records. In case of hospitalization in other hospitals, we formally requested the clinical records of interest. To ascertain survival status, the National Civil Registry was further contacted for final confirmation of the events.
End points and definitions
The new post-hoc main end point of this secondary long-term follow-up analysis was all-cause mortality. Additional end points were the rate of major adverse cardiac events, which included death, all nonfatal major MI (also periprocedural), and target vessel revascularization (TVR). All deaths were considered cardiac unless a clear noncardiac cause could be established. Specifically, any unexpected or unwitnessed death was considered of cardiac origin. Periprocedural MI was defined as an elevation of creatine kinase-MB activity >3 times above the upper limit of normal (16 U/l in our institution). Nonperiprocedural MI was defined as a new ischemic event with creatine kinase-MB >2 times the upper limit of normal, or the electrocardiographic presence of new pathological Q waves. Target lesion revascularization (TLR) was defined as a repeated revascularization procedure (either percutaneous coronary intervention or coronary bypass surgery), as the result of restenosis in the stented segment. We defined TVR as a new revascularization procedure in the target vessel, also including TLR. Stent thrombosis was defined as early (<30 days), late (<1 year), and very late (>1 year) and as definite, probable, or possible. Definite stent thrombosis required an acute coronary ischemic event associated to angiographic or autopsy documentation of partial or total stent occlusion or thrombosis. Probable stent thrombosis included any unexplained death within 30 days after the index procedure and any target vessel MI (also without angiographic documentation) any time during follow-up. Possible stent thrombosis included any unexplained death occurring >30 days after the index procedure (15). All the clinical events were adjudicated by an independent clinical events committee unaware of the patients’ treatment assignment.
Continuous data are expressed as means ± standard deviation or as medians (interquartile range) as appropriate, whereas dichotomous data are summarized as frequencies. Parametric Student tor nonparametric Mann-Whitney Utest (as appropriate) and chi-square or the Fisher exact test (as appropriate) have been used, respectively, for unpaired continuous and for categorical variables, to analyze differences between the 2 study arms. Additionally, survival analysis was graphically represented with the Kaplan-Meier technique, and groups were compared with the log rank test. A 2-sided p value <0.05 was considered significant for all tests. We also calculated the 95% confidence limits for the single rates found and the 95% confidence intervals (CIs) for the absolute differences computed, using the Confidence Interval Analysis software (CIA, version 2.0.0) (16). The number needed to treat/harm with 95% CI (extrapolated from the absolute difference) was computed for the occurrence of dichotomous events. Dr. Agostoni had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Among the 75 patients (with 96 lesions localized in 80 SVG) enrolled in the RRISC trial, 38 patients received 60 SES for 47 lesions, whereas 37 patients received 54 BMS for 49 lesions. The baseline clinical characteristics of the 2 groups are shown in Table 1and are not different.
All patients were contacted between November and December 2006. The median follow-up was not different between the 2 groups: 30.5 months (range 24 to 36 months) for patients treated with SES versus 32 months (range 29 to 35.5 months) for patients treated with BMS (p = 0.24).
Death occurred in 11 patients after SES (29% [95% confidence limits 17% to 45%]) versus 0 after BMS (0% [0% to 9%]) with an absolute difference of 29% (95% CI 14% to 45%, p < 0.001) (Fig. 1).According to the number needed to harm analysis, in our trial the treatment of 3.4 patients (95% CI 2.2 to 7.3) with SES resulted in 1 additional death with respect to treatment with BMS. Specifically, there were 7 cardiac deaths, of which 3 were sudden, possibly related to late or very late stent thrombosis according to our definition, and 1 was caused by an angiographically documented very late stent thrombosis 13 months after stent implantation, which caused a large MI leading to death. Other details on the deceased patients are presented in Tables 2 and 3.⇓⇓Another additional very late angiographically documented SES thrombosis caused a nonfatal MI at 30 months after the index procedure, and it was successfully treated percutaneously. The patient was on acetylsalicylic acid, whereas he stopped clopidogrel 6 months after the index procedure. Thus, the overall rate of definite angiographically documented stent thrombosis was 5% in the SES group (2 of 38, both very late) versus 0% in the BMS group (p = 0.49), whereas the rate of any possible stent thrombosis was 13% (5 of 38, 2 late and 3 very late) after SES versus 0% after BMS (Fisher exact test 2-sided p value = 0.054; log rank test = 0.022).
Table 4shows the overall rate of events that occurred in the 2 groups at long-term follow-up and the medical therapy taken by the patients at the moment of last follow-up. As evident, a nonsignificant trend toward an increased rate of lesion- and vessel-related revascularization procedures is evident in the SES group after 6-month follow-up, neutralizing the initial 6-month benefit of SES in terms of TLR and TVR (Fig. 2).Potential confounding factors, such as additional revascularization procedures not TLR/TVR-related or the antiplatelet/hypocholesterolemic therapy, were not significantly different and were well balanced between the 2 groups.
This extended analysis of the randomized double-blind RRISC trial evaluates the long-term clinical effects of SES versus BMS in patients with diseased SVG. The main findings are: 1) patients treated with SES showed a significant increase in total mortality; and 2) the benefit of SES in terms of reduced revascularization procedures shown at 6 months (14) was no longer evident up to 3 years.
Great focus has recently been put on the evaluation of long-term follow-up of drug-eluting stents in native coronary arteries, mainly after publication of original and meta-analytical studies showing a possible increase in “hard” adverse events, specifically very late stent thrombosis, after drug-eluting stent deployment with respect to BMS (6–11). In light of these data, in December 2006, the U.S. Food and Drug Administration set up a panel to thoroughly assess the long-term values and drawbacks of drug-eluting stents. A final statement was made, in which the U.S. Food and Drug Administration deemed the on-label use of drug-eluting stents safe but warned against an increased risk of adverse events with off-label use (17,18). As percutaneous SVG treatment with SES is currently off-label, and also in light of the results of our trial, potential advantages and drawbacks of the implantation of SES (or other drug-eluting stents) versus BMS in SVG should be carefully evaluated before the procedure and discussed personally with every patient.
Our trial also suggests that in SVG there can be a potential late “catch-up” phenomenon, as a nonsignificant trend toward more repeated revascularization procedures occurred in the SES group after 6 months with respect to the BMS group, thus leading to a lack of benefit of SES over BMS in reduction of clinical restenosis. A percutaneous treatment with a clear additional benefit over BMS in SVG lesions, which account for up to 10% to 15% of lesions treated percutaneously in the majority of the catheterization laboratories (12), is still missing. Of interest, other recent studies assessing the clinical impact of drug-eluting stent versus BMS in diseased SVG failed to show benefits of the former over the latter (19,20).
A clear pathophysiological explanation for these findings is lacking. It is well established that the long-term prognosis of patients with diseased SVG is mainly impacted by progression of disease in nonintervened SVG segments (21,22). However, the specific anatomic and pathological response of diseased SVG to stent implantation also can lead to processes that conduct to late/very late thrombosis and late restenosis in a more pronounced way than what occurs in native coronary arteries. Indeed, SVG lesions, which are more lipid-rich, softer, and more prone to rupture than plaques in native coronary arteries, can induce progression of atherosclerosis and may lead to an enhanced inflammatory and thrombotic reaction after deployment of stents; this can potentially be more pronounced with devices coated with drugs and polymers (23–25).
Thus far, the use of BMS should still be considered the reference percutaneous treatment of SVG and the correct control arm for future SVG randomized trials, which should also be designed bearing in mind the potential occurrence of very late (>1 year) thrombosis and late (>6 months) restenosis.
There are major limitations in our study. First, the sample size of this cohort of patients is small, thus the results can be underpowered to appropriately address specific questions and can be prone to type I and type II statistical error. Second, the recommended duration of double antiplatelet therapy was only mandatory for at least 2 months in our study (although one-half of the population was still taking acetylsalicylic acid and a thienopyridine up to 3 years, as evident in Table 4). Recent evidence has shown that double antiplatelet therapy should be recommended in all patients receiving drug-eluting stents for at least 12 months (26). On the other hand, the RRISC trial was started in 2003 and, during that period, patients in whom SES were implanted, routinely received 2- to 3-month double antiplatelet therapy, according also to other trials (1–4). However, we cannot exclude that some of the events described in the current study could be explained by “premature” discontinuation of dual antiplatelet therapy. Third, this study presents a secondary post-hoc analysis; thus, the current main end point (death) was not prespecified at the moment of the beginning of the trial (which was powered for a 6-month difference in angiographic late loss analysis). Death was selected as the end point after the recent concerns related to an increase in long-term mortality after SES treatment in native coronary arteries (6–10).
However, similar to the BASKET-LATE (Basel Stent Kosten Effektivitats—LAte Thrombotic Events) trial (6), the use of a previously randomized study cohort for a long-term observational study can preserve some of the internal validity of a randomized comparison (27). Because most of the effort and resources go into the organization and conduct of the original trial, adding a second question and extending follow-up can be logistically efficient and scientifically helpful (27). Moreover, no drug-eluting stent trial to date has been powered to assess long-term mortality (1–4), thus preliminary data on long-term results of the currently available trials can be useful for future directions.
Our study has shown that the use of BMS was associated with lower long-term mortality than the use of SES for SVG disease. Also, the 6-month reduction in repeated revascularization procedures shown with the use of SES was lost at longer-term follow-up. Because the observations seen in this secondary post-hoc analysis may have arisen from the play of chance or other clinical factors unrelated to stent type, further studies are required before conclusions can be made about the safety or harm of using SES for SVG lesions.
↵1 Drs. Vermeersch and Agostoni contributed equally to this work.
- Abbreviations and Acronyms
- bare-metal stent(s)
- confidence interval
- myocardial infarction
- sirolimus-eluting stent(s)
- saphenous vein graft(s)
- target lesion revascularization
- target vessel revascularization
- Received April 2, 2007.
- Revision received May 18, 2007.
- Accepted May 21, 2007.
- American College of Cardiology Foundation
- Schampaert E.,
- Cohen E.A.,
- Schluter M.,
- et al.,
- C-SIRIUS Investigators
- Kaiser C.,
- Brunner-La Rocca H.P.,
- Buser P.T.,
- et al.,
- BASKET Investigators
- Pfisterer M.,
- Brunner-La Rocca H.P.,
- Buser P.T.,
- et al.,
- BASKET-LATE Investigators
- Camenzind E.,
- Steg P.G.,
- Wijns W.
- Nordmann A.J.,
- Briel M.,
- Bucher H.C.
- Baim D.
- De Feyter P.
- Vermeersch P.,
- Agostoni P.,
- Verheye S.,
- et al.
- Altman D.G.,
- Machin D.,
- Bryant T.N.,
- Gardner M.J.
- Mayor S.
- Wilson B.H.,
- Humphrey A.D.,
- Cedarholm J.C.,
- et al.
- Keeley E.C.,
- Velez C.A.,
- O’Neill W.W.,
- Safian R.D.
- Depre C.,
- Havaux X.,
- Wijns W.
- Van Beusekom H.,
- Van Der Giessen W.,
- Van Suylen R.,
- Bos E.,
- Bosman F.T.,
- Serruys P.W.
- Grines C.L.,
- Bonow R.O.,
- Casey D.E. Jr..,
- et al.
- Harrington R.A.,
- Califf R.M.