There are no practical criteria for the use of triple antiplatelet therapy after drug-eluting stent (DES) implantation. In our present report, pooled analysis of 3 randomized studies in patients with diabetes mellitus (Drug-Eluting Stenting Followed by Cilostazol treatment reduces LAte Restenosis in patients with diabetes mellitus trial) and long coronary narrowings (Drug-Eluting Stenting Followed by Cilostazol Treatment Reduces Late Restenosis in Patients with Long Coronary Lesions trials I and II) compared triple (aspirin, clopidogrel, and cilostazol; triple group, n = 700) and dual antiplatelet therapies (aspirin and clopidogrel; dual group, n = 699) after DES implantation. Among pooled population (n = 1,399 patients), 1,173 patients with follow-up angiography were divided into 3 stent length categories (≤20, 20 to 40, and >40 mm). There was no statistical significance of in-stent restenosis (ISR) in ≤20- and 20- to 40-mm categories between 2 groups. However, ISR rate was significantly reduced in triple versus dual group in >40-mm stent length category (12.4% vs 22.1%, p = 0.008). In diabetic patients, triple group also showed significant reduction in the ISR rate in >40-mm stent length category (15.4% vs 32.3%, p = 0.003). According to postprocedural minimal lumen diameter, triple group showed a trend toward a lower ISR than that of the dual group in all categories (p = 0.033 for ≤2.5 mm, p = 0.087 for 2.5 to 3.0 mm, and p = 0.119 for >3.0 mm). In conclusion, the triple group had a significantly reduced ISR in patients with >40-mm stent length after DES implantation compared with the dual group. Therefore, this suggestion for use of triple antiplatelet therapy could be easily applied after DES implantation in routine clinical practice.
Cilostazol, a phosphodiesterase III inhibitor, has antiproliferative effects, as shown by its ability to reduce angiographic restenosis rate after bare-metal and drug-eluting stent (DES) implantations. We previously performed a randomized, multicenter, prospective study in which we reported that triple antiplatelet therapy (aspirin, clopidogrel, and cilostazol) reduced restenosis and subsequent repeat revascularization after first generation (sirolimus- [SES] and paclitaxel-eluting stents) or newer generation DES (Endeavor zotarolimus-eluting stent, Medtronic Vascular, Santa Rosa, California) implantation in diabetics (Drug-Eluting Stenting Followed by Cilostazol treatment reduces LAte Restenosis in patients with diabetes mellitus [DECLARE-DIABETES] trial) or patients with long lesion (Drug-Eluting Stenting Followed by Cilostazol Treatment Reduces Late Restenosis in Patients with Long Coronary Lesions [DECLARE-LONG] I and II trials). Accordingly, we suggested that triple antiplatelet therapy is a valuable strategy in patients at high risk of restenosis after DES implantation. However, there remains doubt regarding which patients benefit most from triple antiplatelet therapy in terms of restenosis. Interventional cardiologists still hesitate to use cilostazol in daily clinical practice because there are no practical criteria for its use. From a practical point of view, implanted stent length is an easily applicable criteria for cilostazol use immediately after stenting. Therefore, to find appropriate use criteria of triple antiplatelet therapy based on the association between the length of the stented segment and the risk of angiographic restenosis after DES implantation, we analyzed follow-up angiographic outcomes of patients included in the DECLARE-DIABETES, -LONG I, and -LONG II trials.
Methods
The design, exclusion and inclusion criteria, and data collection of the DECLARE-DIABETES, -LONG I, and -LONG II trials have been previously described. In brief, the 3 randomized studies included 1,399 patients (400 patients of the DECLARE-DIABETES trial, 500 of the DECLARE-LONG I trial, and 499 of the DECLARE-LONG II trial) aged ≥18 years with angina pectoris and/or a positive result of a stress test and a native coronary lesion. The DECLARE-DIABETES and -LONG I studies involved 5 cardiac centers in Korea from August 2004 to March 2006. The DECLARE-LONG II study involved 10 cardiac centers in Korea from December 2007 to December 2008. Patients were considered eligible if they had diabetes mellitus (DM) or a long lesion (lesion length of ≥25 mm), presented with angina pectoris and/or had a positive result of a stress test, and had clinically significant angiographic stenosis in a native coronary vessel with a diameter stenosis of ≥50% and visual reference diameter of ≥2.5 mm. Patients were excluded if aspirin, clopidogrel, or cilostazol use was contraindicated or if they had left main disease (diameter stenosis of ≥50% by visual estimate), graft vessel disease, a left ventricular ejection fraction of <30%, a recent history of hematologic disease or leukocyte count of <3,000/mm 3 and/or platelet count of <100,000/mm 3 , hepatic dysfunction with aspartate or alanine aminotransferase of ≥3× the upper normal reference limit, a history of renal dysfunction or serum creatinine level of ≥2.0 mg/dl, a serious noncardiac co-morbid disease with a life expectancy of <1 year, planned bifurcation stenting in the side branch, primary angioplasty for acute myocardial infarction within 24 hours, or an inability to follow the protocol. The same type of allocated stent was used for all lesions in patients with multiple lesions. The institutional review board at each participating center approved the protocol. All patients provided written informed consent.
Once the guidewire had crossed the target lesion, patients were randomly assigned in a 1:1 ratio to SES or paclitaxel-eluting stent implantation. After DES randomization, patients were randomly allocated in a 1:1 ratio to the triple (200 patients of the DECLARE-DIABETES trial and 250 of the DECLARE-LONG I trial; aspirin, clopidogrel, and cilostazol; triple group, n = 450) or the dual antiplatelet therapy group (200 patients of the DECLARE-DIABETES trial and 250 of the DECLARE-LONG trial; aspirin and clopidogrel; dual group, n = 450) on the basis of a 2 × 2 factorial design with a computer-generated randomization sequence. In the DECLARE-LONG II trial, after successful Endeavor zotarolimus-eluting stent implantation, patients were randomly allocated in a 1:1 ratio to the triple (aspirin, clopidogrel, and cilostazol; triple group, n = 250) or the dual antiplatelet therapy group (aspirin, clopidogrel, and placebo; dual group, n = 249) with an interactive web response system. From at least 24 hours before the procedure and thereafter, all patients received aspirin (200 mg/day) and clopidogrel (loading dose of 300 mg, followed by 75 mg/day for at least 6 months in the DECLARE-DIABETES and -LONG I trials or 8 months in the DECLARE-LONG II trial). Patients in the triple group received a loading dose of 200 mg cilostazol immediately after the procedure and 100 mg twice a day for 6 or 8 months. Coronary stenting was performed with the standard technique. The decision of predilation or direct stenting was made by the operator. The use of intravenous glycoprotein IIb/IIIa inhibitors was at the operators’ discretion. A 12-lead electrocardiogram was obtained after the procedure and before discharge. Serum levels of creatine kinase and its MB isoenzyme were assessed 8, 12, and 24 hours after the procedure and thereafter if considered necessary.
Primary end point was to find the relation between implanted stent length and angiographic in-stent restenosis (ISR) to find an optimal indicator of cilostazol addition after DES implantation. The relation between the postprocedural minimal lumen diameter (MLD) and angiographic ISR was also assessed.
Angiographic restenosis was defined as a diameter stenosis of >50% on follow-up angiography. Repeat coronary angiography was mandatory at 6 months in the DECLARE-DIABETES and -LONG I trials, or 8 months in the DECLARE-LONG II trial after stenting, or earlier if indicated by clinical symptoms or evidence of myocardial ischemia. Coronary angiograms were obtained after intracoronary nitroglycerin administration. Preprocedure (baseline), postprocedure, and follow-up angiograms were submitted to the angiographic core analysis center (Asan Medical Center, Seoul, Korea) for analysis by independent angiographers. Digital angiograms were analyzed using an automated edge detection system (CASS II; Pie Medical, Maastricht, The Netherlands). Angiographic variables included lesion length, stent length, reference vessel diameter, MLD, percent diameter stenosis, restenosis rate, acute gain, and late loss. Quantitative coronary angiographic measurements of target lesions were obtained for both the stented segment only (in-stent) and the region that included the stented segment and the margins 5 mm proximal and distal to the stent (in-segment).
Continuous variables are presented as mean ± SD and were compared using Student unpaired t or Mann-Whitney U test according to the restenosis. Categorical variables are presented as numbers or percentages and were compared using chi-square or Fisher’s exact test. Because we combined the 3 trials, we performed a test of heterogeneity (Q statistics and I 2 index) across the trials. Based on the heterogeneity test, the 3 trials could be combined (Q statistics and I 2 index; p = 0.787 and 0% [0% to 56.5%] for ISR and p = 0.903 and 0% [0% to 0%] for in-segment restenosis, respectively). Multivariate analyses involved a backward elimination technique, and variables with a p value of <0.20 were used to find risk factors associated with the restenosis. The relative risk for the restenosis and its 95% confidence interval (CI) were computed and receiver operating characteristic curve analysis was conducted to find a stent length cutoff for predicting ISR in the dual group. Then, we computed the restenosis rates according to the stent length and the postprocedural MLD between the triple and dual groups. A p value of <0.05 was considered to indicate a significant difference. In the subgroup analysis, we conducted several testing together for 2 outcome variables, respectively. Therefore, we used the Bonferroni-corrected significance level α to adjust for the multiplicity (e.g., 0.05/3 = 0.017). Multivariate logistic regression analysis was used to obtain the adjusted odds ratio. This model was adjusted for type of DES, DM, bifurcation lesion, stent length, and postprocedural MLD, which were identified as independent predictors of restenosis.
All p values were 2-sided and a probability value of p <0.05 was considered significant. Statistical analysis was performed using SAS software, version 9.1 (SAS Institute, Cary, North Carolina).
Results
In the pooled population (n = 1,399), there were no significant differences between the 2 groups in baseline clinical characteristics and risk factors except left ventricular ejection fraction between triple and dual groups (not shown). Among them, follow-up angiography was performed in 1,173 patients (83.8%). In follow-up angiographic analysis, triple group significantly reduced ISR (8.2% vs 13.6%; relative risk 0.60; 95% CI 0.53 to 0.84; p = 0.003) and in-segment restenosis (9.0% vs 15.7%; relative risk 0.58, 95% CI 0.53 to 0.65; p = 0.001) compared with dual group. Tables 1 and 2 list the angiographic characteristics and procedural results with or without restenosis. The restenosis group had a longer stent length and more number of stents at target lesion. Baseline and postprocedural quantitative coronary angiographic outcomes for the 2 groups are listed in Table 3 . The restenosis group had a longer lesion length, smaller reference vessel diameter, smaller postprocedural and follow-up MLDs, and smaller postprocedural and follow-up diameter stenoses than no restenosis group. Restenosis group had a higher late loss than no restenosis group.
| Variable | Restenosis | p | |
|---|---|---|---|
| Yes, n = 145 (%) | No, n = 1,028 (%) | ||
| Triple antiplatelet therapy | 53 (36.6) | 533 (51.8) | 0.001 |
| Age (yrs) | 61 ± 8 | 60 ± 9 | 0.271 |
| Men | 99 (68.3) | 679 (66.1) | 0.596 |
| Hypertension | 86 (59.3) | 599 (58.3) | 0.812 |
| DM | 83 (57.2) | 534 (51.9) | 0.232 |
| Total cholesterol level of ≥200 mg/dl | 46 (31.7) | 355 (34.5) | 0.504 |
| Current smoker | 49 (33.8) | 330 (32.1) | 0.683 |
| Previous percutaneous coronary intervention | 17 (11.7) | 95 (9.2) | 0.341 |
| Previous coronary artery bypass surgery | 0 (0) | 11 (1.1) | 0.378 |
| Clinical diagnosis | 0.484 | ||
| Stable angina pectoris | 65 (44.8) | 473 (46.0) | |
| Unstable angina pectoris | 61 (42.1) | 388 (37.7) | |
| Acute myocardial infarction | 19 (13.1) | 167 (16.2) | |
| Left ventricular ejection fraction (%) | 60 ± 9 | 60 ± 9 | 0.947 |
| Multivessel coronary disease | 70 (48.3) | 560 (54.5) | 0.161 |
| Variable | Restenosis | p | |
|---|---|---|---|
| Yes, n = 145 (%) | No, n = 1,028 (%) | ||
| DESs | <0.001 | ||
| SES | 14 (9.7) | 372 (36.2) | |
| Paclitaxel-eluting stent | 62 (42.8) | 297 (28.9) | |
| Zotarolimus-eluting stent | 69 (47.6) | 359 (34.9) | |
| Target coronary artery | 0.270 | ||
| Left anterior descending | 94 (64.8) | 642 (62.5) | |
| Left circumflex | 12 (8.3) | 133 (12.9) | |
| Right coronary | 39 (26.9) | 253 (24.6) | |
| Maximal inflation pressure (atm) | 15.4 ± 3.8 | 15.7 ± 3.6 | 0.372 |
| Use of intravascular ultrasound | 77 (53.1) | 554 (53.9) | 0.859 |
| Use of glycoprotein IIb/IIIa inhibitor | 2 (1.4) | 20 (1.9) | 0.781 |
| Total stent length at the target lesion (mm) | 44.1 ± 16.1 | 37.1 ± 13.5 | <0.001 |
| Multivessel coronary stenting | 53 (36.6) | 392 (38.1) | 0.713 |
| Number of used stents at the target lesion | 1.69 ± 0.68 | 1.43 ± 0.58 | <0.001 |
| Procedure-related non–Q-myocardial infarction | 9 (6.2) | 98 (9.5) | 0.220 |
| Variable | Restenosis | p | |
|---|---|---|---|
| Yes (n = 145) | No (n = 1,028) | ||
| Reference diameter (mm) | 2.82 ± 0.49 | 2.92 ± 0.43 | 0.038 |
| Lesion length (mm) | 38.6 ± 15.9 | 31.7 ± 12.9 | <0.001 |
| MLD (mm) | |||
| In-segment | |||
| Before procedure | 0.76 ± 0.49 | 0.81 ± 0.48 | 0.200 |
| After procedure | 2.16 ± 0.53 | 2.26 ± 0.48 | 0.018 |
| At follow-up | 1.05 ± 0.49 | 2.15 ± 0.44 | <0.001 |
| In-stent | |||
| After procedure | 2.48 ± 0.44 | 2.59 ± 0.42 | 0.002 |
| At follow-up | 1.11 ± 0.57 | 2.31 ± 0.48 | <0.001 |
| Diameter stenosis (%) | |||
| In-segment | |||
| Before procedure | 70.4 ± 14.6 | 70.4 ± 15.8 | 0.958 |
| After procedure | 19.2 ± 11.7 | 16.9 ± 11.1 | 0.033 |
| At follow-up | 62.8 ± 16.8 | 23.1 ± 12.1 | <0.001 |
| In-stent | |||
| After procedure | 9.4 ± 12.8 | 7.3 ± 12.8 | 0.089 |
| At follow-up | 59.8 ± 20.5 | 18.4 ± 15.4 | <0.001 |
| Acute gain (mm) | |||
| In-segment | 1.39 ± 0.63 | 1.45 ± 0.64 | 0.368 |
| In-stent | 1.72 ± 0.59 | 1.78 ± 0.59 | 0.225 |
| Late loss (mm) | |||
| In-segment | 1.09 ± 0.58 | 0.12 ± 0.39 | <0.001 |
| In-stent | 1.35 ± 0.59 | 0.28 ± 0.41 | <0.001 |
On multivariate analysis, all clinical and angiographic variables with p <0.2 in univariate analysis were tested. Independent predictors of angiographic restenosis were use of cilostazol, SES, DM, stent length, bifurcation lesion, and postprocedural MLD ( Table 4 ).
| Variable | Odds Ratio | 95% CI | p |
|---|---|---|---|
| Cilostazol | 0.49 | 0.35–0.67 | <0.001 |
| SES | 0.12 | 0.09–0.14 | <0.001 |
| DM | 1.99 | 1.37–2.89 | <0.001 |
| Bifurcation lesion | 1.33 | 1.17–1.52 | <0.001 |
| Stent length | 1.03 | 1.02–1.04 | <0.001 |
| Postprocedural MLD | 0.38 | 0.17–0.87 | 0.021 |
The threshold of stent length for predicting ISR in dual group was 39.5 mm (area under the receiver operating characteristic curve, 0.628; 95% CI 0.588 to 0.668, p <0.001), which had a sensitivity and specificity of 63.7% and 65.3%, respectively ( Figure 1 ). Therefore, the patients with follow-up angiography (n = 1,173) were divided into 3 stent length categories (≤20, 20 to 40, and >40 mm). The ISR (12.4% vs 22.1%, p = 0.008) and in-segment restenosis rates (13.8% vs 23.9%, p = 0.008) were significantly different between the triple and dual groups in the >40-mm stent length group ( Figure 2 ). If patients were divided into 3 categories (≤2.5, 2.5 to 3.0, and >3.0 mm) according to postprocedural MLD, triple group showed lower ISR (p = 0.022 for ≤2.5 mm, p = 0.087 for 2.5 to 3.0 mm, and p = 0.119 for >3.0 mm) and in-segment restenosis (p = 0.006 for ≤2.5 mm, p = 0.073 for 2.5 to 3.0 mm, and p = 0.088 for >3.0 mm) compared with dual group in all categories ( Figure 2 ). After adjustment, triple group showed consistent findings in >40-mm group ( Table 5 ). In patients with >40-mm stent length, triple group showed similar trend toward a lower ISR (14.5% vs 22.2% for ≤2.5 mm, 10.2% vs 23.9% for 2.5 to 3.0 mm, and 8.8% vs 16% for >3.0 mm) according to postprocedural MLD compared with dual group in all categories. In addition, each stent type also showed similar trend according to stent length between triple and dual groups ( Figure 3 ).
