Double Versus Single Stenting for Coronary Bifurcation LesionsCLINICAL PERSPECTIVE
Background— Several trials have addressed whether bifurcation lesions require stenting of both the main vessel and side branch, but uncertainty remains on the benefits of such double versus single stenting of the main vessel only.
Methods and Results— We have conducted a meta-analysis of randomized trials including patients with coronary bifurcation lesions who were randomly selected to undergo percutaneous coronary intervention by either double or single stenting. Six studies (n=1642 patients) were eligible. There was increased risk of myocardial infarction with double stenting (risk ratio, 1.78; P=0.001 by fixed effects; risk ratio, 1.49 with Bayesian meta-analysis). The summary point estimate suggested also an increased risk of stent thrombosis with double stenting, but the difference was not nominally significant given the sparse data (risk ratio, 1.85; P=0.19). No obvious difference was seen for death (risk ratio, 0.81; P=0.66) and target lesion revascularization (risk ratio, 1.09; P=0.67).
Conclusions— Stenting of both the main vessel and side branch in bifurcation lesions may increase myocardial infarction and stent thrombosis risk compared with stenting of the main vessel only.
Received March 24, 2009; accepted July 22, 2009.
Percutaneous coronary intervention (PCI) of bifurcation lesions remains a subject of debate in terms of optimum treatment technique. Balloon angioplasty and bare metal stents have gradually been replaced by drug-eluting stents.1–6 When stenting is used, a major question is whether both the main vessel and the side branch should be stented. Such double stenting is appealing because it produces attractive postprocedural angiographic results. However, there have been concerns about long-term major adverse cardiac events and, in particular, stent thrombosis with adjacent stents. Among several randomized trials published so far,7–13 the deployment of an additional stent on the side branch of a bifurcation lesion has never been shown to result in significantly improved clinical outcomes as initially hypothesized. These trials, however, have been of too limited a sample size to allow robust conclusions, when seen in isolation. A synopsis of 3 randomized and 3 observational comparisons failed to detect any significant differences between double and single stenting in a bifurcation.14 However, several other trials have appeared recently. We have, therefore, performed a meta-analysis of randomized trials to examine the relative outcomes of the double versus single stenting strategies for bifurcation lesions in terms of mortality, myocardial infarction, stent thrombosis, and target lesion revascularization.
Clinical Perspective on p 409
Study Eligibility and Selection
We have conducted a meta-analysis of randomized trials including patients with coronary bifurcation lesions who were randomly selected to undergo PCI by either single stenting of the main vessel only (simple strategy) or double stenting of both the main vessel and the side branch (complex strategy) with any PCI technique. For study selection, we searched PubMed, using the terms bifurcation, coronary, and stent (last update, February 26, 2009). In addition, we used the same terms to locate potentially eligible studies in the Cochrane Central Register of Controlled Trials and the Clinical Trials Registry (www.clinicaltrials.gov), through which data available in presentation and/or abstract form of conferences were also considered for eligibility screening. Two investigators (D.G.K. and G.C.M.S.) independently evaluated search results for eligible studies, which were then scrutinized in full text. A third investigator (J.P.A.I.) commented on their discrepancies. We included trials regardless of the location of the bifurcation lesions in the coronary artery tree and regardless of the clinical presentation (acute or stable). Randomized studies that compared percutaneous coronary balloon angioplasty only versus PCI of the main vessel with provisional stenting of the side branch, as well as studies that compared different techniques of stenting of both bifurcation branches were excluded. Nonrandomized and non-English language studies were excluded.
All data were extracted independently by 2 evaluators (D.G.K. and G.C.M.S.) and discrepancies were resolved by a third evaluator (J.P.A.I.). From each eligible study, we recorded information regarding the number of patients that were randomly assigned to 1 of the 2 stenting strategies based on the intention-to-treat analysis and patient demographics (mean age, proportion of male patients, major cardiovascular risk factors, history of myocardial infarction, prior PCI or coronary artery bypass graft surgery, prevalence of angina symptoms, and mean left ventricular ejection fraction). Furthermore, we extracted data on crossover rates; percentages of true bifurcations (when not given by the authors, we defined true bifurcations as Medina 1,1,1 or 1,0,1 or 0,1,1 and Lefevre 1 or 4)15; procedural parameters (type of complex stenting technique, type of drug-eluting stent, use of glycoprotein IIb/IIIa inhibitors, implementation of final “kissing balloon” inflation); measurements that derived from the quantitative coronary angiography analysis (minimum lumen diameter, diameter stenosis, lesion length) and the location of the bifurcation lesions.
The following major outcomes were extracted: death from any cause, myocardial infarction (Q wave and non-Q wave), stent thrombosis, and target lesion revascularization.
We considered the longest available follow-up period for all events in each eligible study. For stent thromobosis, the criteria used in each individual eligible study were adopted. Myocardial infarction was defined by ECG changes and elevation of myocardial enzymes. In cases that the events in each intervention group were analyzed based on the finally implemented treatment and not according to the intention-to-treat approach, we contacted the corresponding authors to provide us the intention-to-treat data.
Categorical data are summarized as frequencies and percentages, whereas summary statistics for continuous variables are presented as means and SDs. For meta-analyses, the risk ratio was used as the metric of choice, and we also performed analyses using the risk difference metric. Between-study heterogeneity was evaluated with the χ2 test-based Q statistic and was considered statistically significant at a level of <0.10. We further quantified the effect of the heterogeneity across studies using the I2 statistic, which is independent of the number of studies16,17 and obtained its 95% CIs.18 Fixed effects models with studies weighted by the inverse of their variance and random effects models using the DerSimonian and Laird method were used to combine the data across studies.19 When there is no detectable between-study heterogeneity, the 2 models give identical results. In the presence of detectable between-study heterogeneity, random effects give wider CIs.
For meta-analysis results that were nominally statistically significant by both fixed and random effects, we also performed a sensitivity analysis using a Bayesian meta-analysis model for the risk ratio. This approach generates a posterior distribution of the effect size and 95% credibility, as opposed to CIs generated by traditional frequentist meta-analysis. We used the WinBUGS software20 (Imperial College and MRC, London) and estimated the posterior of the effect size using 20 000 iterations after a 10 000 iteration initial burn-in. Bayesian inference is more appropriate for making probabilistic inferences about the magnitude of the treatment effect and the variability that it may have in different populations that are similar to the ones studied in the randomized trials whose data are combined.
Statistical analyses were conducted in Stata 10.0 (Stata Corp, College Station, Tex) and WinBUGS software. P values are 2 tailed.
Study Selection and Characteristics
Our search yielded 518 items (459 in PubMed, 33 in the Cochrane Central Register of Controlled Trials, and 26 in the Clinical Trials Registry), which were evaluated for eligibility. Of those, 457 were excluded based on their title, and 47 were excluded based on their abstract. Fourteen items on randomized trials comparing different treatment interventions in patients with coronary bifurcation lesions were assessed in full text. Of those, we excluded 4 that compared different types of complex strategies and 3 that compared percutaneous transluminal coronary angioplasty versus PCI/percutaneous transluminal coronary angioplasty. Two of the remaining articles pertained to the same study in different follow-up periods.11,12 Finally, 6 randomized trials fulfilled our inclusion criteria.7–13
Patients, Demographic, and Procedural Characteristics
Demographic characteristics of the included patients and procedural details of the implemented PCI technique are shown in Tables 1 through 3⇓⇓. A total of 1642 patients were included (821 in each intervention arm). Patient populations’ mean age ranged from 60 to 67 years, whereas women comprised 20% to 23% of the study populations. The proportion of diabetic patients ranged from 12% to 41% (Table 1). In 4 studies,8–11 patients that presented with myocardial infarction in the last 24 hours were excluded. Moreover, reduced left ventricular function (ejection fraction <35%) was an exclusion criterion in CACTUS8 and Colombo et al9 Some patients with myocardial infarction were included only in BBC ONE (21% of the study population).7 The definition of myocardial infarction (Q wave and non-Q wave) differs among the included studies, as shown in Table 2.
One hundred twelve (14% of the simple strategy arm) patients that were initially randomly assigned to the single-stenting strategy arm crossed over to have >1 stent based on the operator’s decision. Different stenting techniques (crush, culotte, T-stenting, and other) were used for double stenting and most operators preferred the crush technique (Table 3). Paclitaxel- and sirolimus-eluting stents were used across the studies. No bare-metal stents were used. Intravenous inhibitors of the platelet glycoprotein IIb/IIIa receptor were administered in 243 and 307 cases of the simple and complex strategy arm, respectively. Final “kissing balloon” inflation was preferred in most cases in both intervention arms. Assessment of the treated vessels with intravascular ultrasound after stenting and at follow-up was performed per protocol only in 1 study9 and also in a small proportion of patients (n=59) in 2 other studies.8,13
In 5 of the 6 studies, information regarding quantitative coronary measurements (minimum lumen diameter, diameter stenosis, and the length of lesion) pre-PCI, post-PCI, and at follow-up were available for each group as shown in supplemental Table I. The most common location of the treated bifurcation lesions across the studies were the left anterior descending/diagonal artery (in 76% of all cases in both arms), whereas bifurcation lesions that included the left main stem were stented only in 13 cases in 2 of the studies (supplemental Table I).
Outcome Events and Data Synthesis
The incidence of the main outcomes across the eligible studies is shown in Table 4. Death occurred in 18 patients (simple strategy n=10 versus complex strategy n=8). In total, 144 patients were reported to have myocardial infarction (simple strategy n=51 versus complex strategy n=93), and 21 events of definite stent thrombosis were recorded (simple strategy n=7 versus complex strategy n=14). Target lesion revascularization was performed in 88 cases (simple strategy n=42 versus complex strategy n=46).
Table 5 and the Figure show the results of the meta-analyses for each outcome. In the risk ratio scale, the simple and complex strategies did not differ significantly in the risk of death, although the 95% CIs were quite large and could not totally exclude even considerable differences. Target lesion revascularization rates were also similar between strategies. There was almost a doubling of the risk of myocardial infarction with the use of 2 stents instead of 1, and this difference was statistically significant (P=0.008 by random effects, P=0.001 by fixed effects). Moreover, the summary point estimates suggested an estimated doubling of the risk of stent thrombosis with the complex strategy, but the difference was not nominally significant given the sparse data (P=0.19 by both fixed and random effects). There was no statistically significant heterogeneity for any of the 3 outcomes, and the I2 point estimate was 0% to 20%, although there was still considerable uncertainty about the extent of possible heterogeneity, given the relatively limited number of events. The largest estimate of I2 was noted for myocardial infarction, where the BBC ONE trial had found a large risk ratio, corresponding to more than tripling of the risk of infarction with double versus single stenting. BBC and NORDIC had each found statistically significantly increased risk of myocardial infarction (P<0.05) with double stenting, and BBC also had the largest point estimate for stent thrombosis.
Bayesian meta-analysis for the risk of myocardial infarction yielded a median risk ratio estimate of 1.49, which is slightly less than the estimate obtained by the traditional fixed or random effects meta-analyses described earlier. The 95% credibility interval was considerably large (0.54 to 2.43) because of the relatively small number of events in several of the combined trials.
Inferences were similar in the risk difference scale, with the exception that there was statistically significant between-study heterogeneity for the myocardial infarction outcome. Point estimates suggested approximately a 3% excess risk of myocardial infarction and 1% excess risk of stent thrombosis with the double-stent versus single-stent approach.
The current meta-analysis shows that the management of bifurcation lesions with double stenting of the main vessel and the side branch rather than using a single stent in the main vessel substantially increases the risk of myocardial infarction during follow-up, although the magnitude and even the presence of the excess risk has substantial uncertainty in Bayesian calculations, given the limited number of events in these studies. The risk of stent thrombosis may also be almost doubled, although this outcome was even more rare, and the meta-analysis result is not formally statistically significant or conclusive. Double stenting does not seem to result in different mortality versus single stenting and target lesion revascularization rates also seem comparable with the 2 strategies.
The introduction of drug-eluting stents in the treatment of bifurcation lesions has been thought to improve operational outcomes and decrease the need for restenosis and revascularization. In the Arterial Revascularization Therapies Study II, event rates were similar in bifurcation and nonbifurcation lesions.21 Erglis et al22 from the NORDIC group also showed that stenting of both bifurcation branches with either the cullote or the crush techniques yielded relatively low restenosis rates. These results, however, were drawn following 6 months clinical and 8 months angiographic follow-up, which is rather inadequate for definitive conclusions about the long-term clinical outcomes of this approach. A previous systematic review14 that did not include 3 recent randomized trials had found no demonstrable differences between the 2 strategies, but the 3 latest studies show an inferiority of double stenting in terms of the risk of myocardial infarction and possibly also stent thrombosis in particular. All studies used drug-eluting stents and modern techniques. The absolute excess risk (in the range of 3 for myocardial infarction) is considerable.
Our findings indicate that double stenting of a bifurcation lesion is not advantageous and seems also to have a detrimental impact on major clinical outcomes. They also suggest that current practices for bifurcation lesions need careful reexamination. For example, in a recent statement by the European Bifurcation Club, complex treatment strategies are discussed and the evidence in favor of the simple strategy is ignored.23 Of course, in clinical practice, complex stenting is often inevitable, especially in the case of a true bifurcated lesion with a massive plaque burden and a side branch of considerable size. However, the angiographic appearance of ostial residual stenosis of a side branch does not correspond to its functional significance, and in most cases, kissing balloon dilatation may suffice.24 The prognosis of jailed side branches is favorable. Dissected or even occluded side branches are usually clinically silent and probably do not affect long-term clinical event-free survival.13,25,26 The majority of side branches (up to 90%) reappear at follow-up.25 Furthermore, complex stenting strategies at bifurcation sites have been identified as predictors of stent thrombosis. Overall, PCI with stenting does not seem to improve death or myocardial infarction rates as compared with optimized medical therapy alone in chronic stable coronary artery disease.27,28 For bifurcation lesions, using 2 stents rather than 1 seems to result even in worse outcomes.
Some caveats should be discussed. First, some of the studies had considerable crossover rates. In particular, Colombo et al9 have reported a crossover rate as high as 51.2% in their study evaluating sirolimus-eluting stents implanted in coronary bifurcation lesions. A high crossover rate may be due to an angiographically unsatisfactory postangioplasty result. Therefore, what the meta-analysis evaluates is the original intention: whether it is better to try to place a single stent or 2. Second, the design of analyzed trials, background profiles of the studied populations, rates of use of adjunctive IIB/IIIA glycoprotein inhibitors, and techniques used to perform double stenting were different between studies. This may need to be considered in the generalization of our conclusions, but we found no strong evidence for between-study heterogeneity in the treatment effects in the risk ratio scale. These sources of diversity should not preclude synthesizing the results using meta-analysis and drawing useful inferences from the meta-analysis.29 Finally, we did not address in this meta-analysis the impact of double versus single stenting on angiographic outcomes. The reported success rates of double stenting in trials are high, and this approach results in good immediate angiographic outcomes. This, however, does not translate in reduced mortality, myocardial infarction, or revascularization rates. The financial burden may also be considerable.
In conclusion, acknowledging these caveats, the current meta-analysis indicates that stenting of both the main and side branches of a bifurcation lesion may be associated with considerable excess risk of myocardial infarction and possible excess risk of stent thrombosis. Our study suggests that double stenting of both the main vessel and the side branch vessel in bifurcation lesions should be avoided. Perhaps double stenting could be considered in cases of a very big bifurcation branch, such as main stem lesions or lesions involving bifurcation branches with diameter almost as big as that of the main branch. However, no convincing data exist in this context. Further studies are needed to define the optimum treatment of bifurcation lesions in the era of changing indications for coronary intervention.
We thank Drs Antonio Colombo, Leif Thuesen, and David Hildick-Smith for offering additional data and clarifications from their studies.
Dr Katritsis is an interventional cardiologist and receives research grants from Boston Scientific and Johnson & Johnson.
Thuesen L, Kelbaek H, Kløvgaard L, Helqvist S, Jørgensen E, Aljabbari S, Krusell LR, Jensen GV, Bøtker HE, Saunamäki K, Lassen JF, van Weert A; SCANDSTENT Investigators. Comparison of sirolimus-eluting and bare metal stents in coronary bifurcation lesions: subgroup analysis of the Stenting Coronary Arteries in Non-Stress/Benestent Disease Trial (SCANDSTENT). Am Heart J. 2006; 152: 1140–1145.
Palmerini T, Marzocchi A, Tamburino C, Sheiban I, Margheri M, Vecchi G, Sangiorgi G, Santarelli A, Bartorelli A, Briguori C, Vignali L, Di Pede F, Ramondo A, Inglese L, De Carlo M, Falsini G, Benassi A, Palmieri C, Filippone V, Sangiorgi D, Barlocco F, De Servi S. Impact of bifurcation technique on 2-year clinical outcomes in 773 patients with distal unprotected left main coronary artery stenosis treated with drug-eluting stents. Circ Cardiovasc Intervent. 2008; 1: 185–192.
Hildick-Smith D, BBC ONE British Bifurcation Coronary study: Old, New and Evolving strategies trial. Presented at TCT, Washington, 2008.
Colombo A, Bramucci E, Saccà S, Violini R, Lettieri C, Zanini R, Sheiban I, Paloscia L, Grube E, Schofer J, Bolognese L, Orlandi M, Niccoli G, Latib A, Airoldi F. Randomized study of the crush technique versus provisional side-branch stenting in true coronary bifurcations. The CACTUS (coronary bifurcations: application of the crushing technique using sirolimus-eluting stents) study. Circulation. 2009; 119: 71–78.
Colombo A, Moses JW, Morice MC, Ludwig J, Holmes DR Jr, Spanos V, Louvard Y, Desmedt B, Di Mario C, Leon MB. Randomized study to evaluate sirolimus-eluting stents implanted at coronary bifurcation lesions. Circulation. 2004; 109: 1244–1249.
Ferenc M, Gick M, Kienzle RP, Bestehorn HP, Werner KD, Comberg T, Kuebler P, Büttner HJ, Neumann FJ. Randomized trial on routine vs. provisional T-stenting in the treatment of de novo coronary bifurcation lesions. Eur Heart J. 2008; 29: 2859–2867.
Steigen TK, Maeng M, Wiseth R, Erglis A, Kumsars I, Narbute I, Gunnes P, Mannsverk J, Meyerdierks O, Rotevatn S, Niemelä M, Kervinen K, Jensen JS, Galløe A, Nikus K, Vikman S, Ravkilde J, James S, Aarøe J, Ylitalo A, Helqvist S, Sjögren I, Thayssen P, Virtanen K, Puhakka M, Airaksinen J, Lassen JF, Thuesen L; Nordic PCI Study Group. Randomized study on simple versus complex stenting of coronary artery bifurcation lesions: the Nordic bifurcation study. Circulation. 2006; 114: 1955–1961.
Jensen JS, Galløe A, Lassen JF, Erglis A, Kumsars I, Steigen TK, Wiseth R, Narbute I, Gunnes P, Mannsverk J, Meyerdierks O, Rotevatn S, Niemelä M, Kervinen K, Nikus K, Vikman S, Ravkilde J, James S, Aarøe J, Ylitalo A, Helqvist S, Sjögren I, Thayssen P, Virtanen K, Puhakka M, Airaksinen J, Thuesen L; Nordic-Baltic PCI Study Group. Safety in simple versus complex stenting of coronary artery bifurcation lesions. The Nordic bifurcation study 14-month follow-up results. EuroIntervention. 2008; 4: 229–233.
Pan M, de Lezo JS, Medina A, Romero M, Segura J, Pavlovic D, Delgado A, Ojeda S, Melián F, Herrador J, Ureña I, Burgos L. Rapamycin-eluting stents for the treatment of bifurcated coronary lesions: a randomized comparison of a simple versus complex strategy. Am Heart J. 2004; 148: 857–864.
Niccoli G, Ferrante G, Porto I, Burzotta F, Leone AM, Mongiardo R, Mazzari MA, Trani C, Rebuzzi AG, Crea F. Coronary bifurcation lesions: to stent one branch or both? A meta-analysis of patients treated with drug eluting stents [published online ahead of print November 21, 2008]. Int J Cardiol.
Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003; 327: 557–560.
Ioannidis JP, Patsopoulos NA, Evangelou E. Uncertainty in heterogeneity estimates in meta-analyses. BMJ. 2007; 335: 914–916.
WinBUGS With DoodleBUGS Version 1.4.3, August 3, 2007. London, United Kingdom: Imperial College and MRC; 2000–2007.
Tsuchida K, Colombo A, Lefevre T, Oldroyd KG, Guetta V, Guagliumi G, von Scheidt W, Ruzyllo W, Hamm CW, Bressers M, Stoll HP, Wittebols K, Donohoe DJ, Serruys PW. The clinical outcome of percutaneous treatment of bifurcation lesions in multivessel coronary artery disease with the sirolimus-eluting stent: insights from the arterial revascularization therapies study part II (ARTS II). Eur Heart J. 2007; 28: 433–442.
Erglis A, Kumsars I, Niemelä M, Kervinen K, Maeng M, Lassen JF, Gunnes P, Stavnes S, Jensen JS, Galløe A, Narbute I, Sondore D, Mäkikallio T, Ylitalo K, Christiansen EH, Ravkilde J, Steigen TK, Mannsverk J, Thayssen P, Nørregaard Hansen K, Syvänne M, Helqvist S, Kjell N, Wiseth R, Aarøe J, Puhakka M, Thuesen L; for the Nordic PCI Study Group. Randomized comparison of coronary bifurcation stenting with the crush versus the culotte technique using sirolimus eluting stents. Circ Cardiovasc Intervent. 2009; 2: 27–34.
Louvard Y, Thomas M, Dzavik V, Hildick-Smith D, Galassi AR, Pan M, Burzotta F, Zelizko M, Dudek D, Ludman P, Sheiban I, Lassen JF, Darremont O, Kastrati A, Ludwig J, Iakovou I, Brunel P, Lansky A, Meerkin D, Legrand V, Medina A, Lefèvre T. Classification of coronary artery bifurcation lesions and treatments: time for a consensus! Catheter Cardiovasc Interv. 2008; 71: 175–183.
Koo BK, Park KW, Kang HJ, Cho YS, Chung WY, Youn TJ, Chae IH, Choi DJ, Tahk SJ, Oh BH, Park YB, Kim HS. Physiological evaluation of the provisional side-branch intervention strategy for bifurcation lesions using fractional flow reserve. Eur Heart J. 2008; 29: 726–732.
Alfonso F, Hernández C, Pérez-Vizcayno MJ, Hernández R, Fernández-Ortíz A, Escaned J, Bañuelos C, Sabaté M, Sanmartín M, Fernández C, Macaya C. Fate of stent-related side branches after coronary intervention in patients with in-stent restenosis. J Am Coll Cardiol. 2000; 36: 1549–1556.
Katritsis DG, Ioannidis JP. Percutaneous coronary intervention versus conservative therapy in nonacute coronary artery disease: a meta-analysis. Circulation. 2005; 111: 2906–2912.
Ioannidis JP, Patsopoulos NA, Rothstein HR. Reasons or excuses for avoiding meta-analysis in forest plots. BMJ. 2008; 336: 1413–1415.
The ideal approach to percutaneous coronary intervention of bifurcation lesions has not been established. Double stenting of both main and side branches produces attractive postprocedural angiographic results and several techniques for successful deployment have been described. Randomized trials published so far have not detected any clinical benefit with the deployment of an additional stent on the side branch, but they were of limited sample size to allow robust conclusions. A meta-analysis of 6 studies involving 1642 patients did not detect any difference in mortality and target lesion revascularization between the 2 approaches. However, there was an increased risk of myocardial infarction and probably stent thrombosis with double stenting. This indicates that a procedure that entails extra costs and irradiation time for both physicians and patients does not confer any real benefit; on the contrary, it can increase the risk of potential complications. Occluded side branches are usually clinically silent, and the majority of them reappears at follow-up and probably do not affect long-term clinical event-free survival. In this context, the results of the meta-analysis clearly suggest that, in most cases, stenting of the side branch should be avoided.
The online-only Data Supplement is available at http://circinterventions.ahajournals.org/cgi/content/full/CIRCINTERVENTIONS.109.868091/DC1.