The investigators performed a network meta-analysis of randomized trials comparing the effectiveness of currently available strategies for the treatment of drug-eluting stent (DES) restenosis. Despite the widespread use of DES in patients who undergo percutaneous coronary intervention, the optimal treatment for DES restenosis remains poorly defined. A systematic search of electronic resources was performed. The primary end point was diameter stenosis at follow-up angiography. Seven trials were included, enrolling a total of 1,586 patients with 1,728 restenotic lesions. The following treatment options were found: balloon angioplasty (BA) in 343 patients (19.3%), iopromide-based paclitaxel-eluting balloons (PEB) in 343 (21.6%), sirolimus-eluting stents in 441 (27.8%), paclitaxel-eluting stents in 462 (29.1%), and everolimus-eluting stents in 34 (2.2%). Compared with BA, PEB (−17.74%, 95% credible interval [CI] −25.17% to −11.31%), everolimus-eluting stents (−14.93%, 95% CI −33.47% to 1.16%), paclitaxel-eluting stents (−15.3%, 95% CI −22.96% to −8.35%), and sirolimus-eluting stents (−11.08%, 95% CI −17.89% to −3.4%) had similar reductions in diameter stenosis at follow-up angiography. PEB (85%) and everolimus-eluting stents (68%) had the greatest probabilities for being the best treatment option. Furthermore, PEB were the best treatment in terms of late luminal loss (85%) and binary restenosis (85%). BA had the lowest efficacy with respect to all study end points. In conclusion, in patients with DES restenosis, repeat DES implantation and iopromide-based PEB are valid alternatives. However, PEB had greater angiographic efficacy and therefore should be considered the new benchmark comparator in the treatment of DES restenosis. The use of BA should be discouraged in patients with DES restenosis.
Highlights
- •
Although several strategies have been proposed, the optimal treatment for DES restenosis remains poorly defined.
- •
To address this issue, we performed a network meta-analysis of randomized trials to evaluate the effectiveness of different treatment options in patients with DES restenosis.
- •
We found that in patients with DES restenosis, repeat DES implantation and iopromide-based PEBs are valid alternatives. However, PEBs had greater angiographic efficacy and therefore should be considered the new benchmark comparator in the treatment of DES restenosis.
- •
The use of balloon angioplasty should be discouraged in patients with DES restenosis.
Drug-eluting stents (DES) represented a breakthrough technology in the treatment of patients who underwent percutaneous coronary intervention because of a dramatic reduction in the need for repeat revascularization compared with bare-metal stents (BMS). However, in-stent restenosis after DES implantation still occurs, and its prevalence is not negligible as a result of the widespread use of DES in increasing complex subsets of lesions and patients. Although several strategies have been proposed, the optimal treatment for DES restenosis remains poorly defined. Recently, initial data from randomized studies have been reported, but direct evidence from head-to-head trials is still limited. Network meta-analysis is an established research method in which direct evidence of different therapy comparisons is combined with indirect evidence that is derived from studies sharing a common comparator within the network frame. Therefore, we sought to perform a network meta-analysis of randomized trials to evaluate the effectiveness of different treatment options in patients with DES restenosis.
Methods
We searched MEDLINE, the Cochrane Library, Scopus, scientific session abstracts (published in Circulation , the Journal of the American College of Cardiology , the European Heart Journal , and The American Journal of Cardiology ), and Web sites ( www.acc.org , www.americanheart.org , www.escardio.org , www.europcronline.com , www.clinicaltrialresults.org , and www.tctmd.com ). The reference lists of relevant studies were additionally scanned. No language, publication date, or publication status restrictions were imposed. The last search was run on April 20, 2014. The following search terms were matched: “restenosis,” “drug-eluting stent,” “percutaneous coronary intervention,” and “randomized.” To be included, a citation had to meet the following criteria: random treatment allocation and inclusion of patients with DES restenosis. Exclusion criteria were ongoing studies and irretrievable data. Two investigators (R.P. and G.G.) independently assessed reports for eligibility at the title and/or abstract level, with divergences resolved by consensus; studies that met the inclusion criteria were selected for further analysis. The study validity was evaluated by the same 2 reviewers, according to the risk-for-bias tool recommended by the Cochrane Collaboration. The primary end point of this study was diameter stenosis at follow-up angiography. Secondary end points were late luminal loss (LLL), defined as the change in minimal luminal diameter from final to follow-up angiography on quantitative coronary angiography; binary restenosis, defined as percentage diameter stenosis ≥50% on follow-up angiography; and target-lesion revascularization (TLR), defined as any clinically driven revascularization procedure involving the target lesion. Statistical analysis was performed by using the Aggregate Data Drug Information System, version 1.16.3, software package. First, we performed standard pairwise meta-analysis for trials that directly compared different treatment arms. Than we performed Bayesian network meta-analysis to compare different therapies for DES restenosis. The following 5 treatment arms were identified: (1) balloon angioplasty (BA), defined as plain-old balloon or cutting BA; (2) paclitaxel-eluting balloon (PEB); (3) sirolimus-eluting stent (SES); (4) paclitaxel-eluting stent (PES); and (5) everolimus-eluting stent (EES). The κ statistic was used to assess agreement between reviewers for study selection. Mean differences or odds ratios (OR) with 95% confidence intervals (CIs) were used as summary statistics for traditional pairwise analyses, by applying the random-effects DerSimonian and Laird model. Heterogeneity of treatment effects across studies was assessed by I 2 statistics and the Cochran Q test. Network meta-analysis was performed by using the Bayesian hierarchical random-effects model proposed by Lu and Ades. The pooled estimates were obtained using the Markov-chain Monte Carlo method. For each model, we generated 100,000 simulations for each of 2 sets of different initial values, and we discarded the first 20,000 simulations as the burn-in period. The achievement of convergence was assessed with the Brooks-Gelman-Rubin method, which compares within-chain and between-chain variance. When a loop connected 3 treatments, it was possible to evaluate the inconsistency between direct and indirect evidence. We used the node-splitting method to calculate the inconsistency of the model, which separated evidence on a particular comparison into direct and indirect evidence. We then evaluated the agreement between the direct and indirect evidence and reported its Bayesian p value. The treatments were ranked for the 4 outcomes in each simulation on the basis of their posterior probabilities. We assessed the probability that each treatment was the most efficacious therapy, the second best, and so on. Probability values were summarized as surface under the cumulative ranking percentages (0% to 100%), with larger surface under the cumulative rankings denoting more effective interventions. All Bayesian results are reported as mean differences or ORs with corresponding 95% credible intervals, as well as the probabilities of ranking by treatment.
Results
As reported in Figure 1 , we screened the titles and/or the abstracts of 3,901 potentially eligible publications. Of these, 3,803 citations were excluded because they were not relevant to this study or were duplicated publications. Ninety-eight studies were thus assessed for eligibility, and 91 records were discarded because the inclusion criteria were not met. Finally, 7 trials with 8 data sets were included in this meta-analysis, enrolling a total of 1,586 patients with 1,728 restenotic lesions. Figure 2 shows the network for the comparisons of available treatments. The interobserver agreement for study selection was very good, with κ = 0.95. The main characteristics of the included studies are listed in Table 1 . The risk for bias among studies is reported in the supplementary data ( Supplementary Table 1 ). SES (Cypher or Cypher Select; Cordis, Johnson & Johnson, Warren, New Jersey) were tested in 3 trials, PES (Taxus Express or Taxus Libertè; Boston Scientific Corporation, Natick, Massachusetts) in 3 studies, and EES (Xience V; Abbott Vascular, Santa Clara, California) in 1 study. PEB were tested in 4 trials, with iopromide-paclitaxel-coated balloons (SeQuent Please; B. Braun Melsungen AG, Berlin, Germany) used in all cases. Plain BA was the comparator in 4 trials, while 1 study performed cutting BA (Boston Scientific Corporation). Sample sizes ranged from 50 to 450 patients. Treatments categories were SES in 441 patients (27.8%), PES in 462 (29.1%), PEB in 343 (21.6%), EES in 34 (2.2%), and BA in 306 (19.3%). Diabetes prevalence among included trials ranged from 30.3% to 62%. Angiographic follow-up was performed in all studies after 6 to 12 months. The pattern of restenosis according to the Mehran classification was reported in all but 1 study. More than half of restenotic lesions had a focal pattern, with the exception of the trial of Song et al, which stratified randomization according to focal or diffuse restenosis characteristics. The mean reference vessel diameter was 2.78 ± 0.30 mm, while the mean lesion length was 13.6 ± 4.3 mm. Dual-antiplatelet therapy was prescribed for a period ranging from 1 to 12 months. Compared with BA, PEB (−17.74%, 95% CI −25.17% to −11.31%), EES (−14.93%, 95% CI −33.47% to 1.16%), PES (−15.3%, 95% CI −22.96% to −8.35%), and SES (−11.08%, 95% CI −17.89% to −3.4%) had similar reductions in diameter stenosis at follow-up angiography ( Table 2 ). However, PEBs (85%) and EESs (68%) had the greatest probabilities for being the best treatment option ( Figure 3 ). No other differences were observed among different treatments. Compared with BA, PEB (−0.43 mm, 95% CI −0.68 to −0.22) and PES (−0.35 mm, 95% CI −0.62 to −0.10) showed a significant reduction in LLL ( Table 2 ). No significant differences were observed among different treatments. As shown in Figure 3 , PEB had the greatest probability (85%) for being the best treatment option for reducing LLL, followed by PES (65%) and EES (54%). PEB (OR 0.18, 95% CI 0.05 to 0.45), PES (OR 0.24, 95% CI 0.06 to 0.75), and SES (OR 0.29, 95% CI 0.07 to 0.99) showed statistical significance in reducing the incidence of angiographic restenosis compared with BA ( Table 2 ), with PEB having the greatest probability (85%) of being the best treatment ( Figure 3 ). PES (OR 0.20, 95% CI 0.05 to 0.45), SES (OR 0.25, 95% CI 0.06 to 0.68), and PEB (OR 0.29, 95% CI 0.09 to 0.57) significantly reduced the incidence of TLR compared with BA. No significant differences were observed among other treatments ( Table 2 ). PES (85%) had the greatest probability of being the best treatment for preventing TLR ( Figure 3 ). As reported in Figure 3 , BA had the lowest surface under the cumulative ranking probabilities for all study end points. Moreover, all data were confirmed by splitting BA treatment into plain and cutting balloon groups (data not shown). The results of traditional pairwise meta-analysis are listed in Table 3 . Although the point estimates showed small differences, the CIs from traditional pairwise meta-analysis and the credible intervals from Bayesian network meta-analysis in general overlapped. The node-splitting method showed no significant inconsistency within the networks for any of the 4 study end points ( Supplementary Table 2 ).
| Trial | CRISTAL | Habara et al. | ISAR-DESIRE 2 | ISAR-DESIRE 3 | PEPCAD-DES | PEPCAD China ISR | Song et al. (focal ISR) | Song et al. (diffuse ISR) |
|---|---|---|---|---|---|---|---|---|
| Study design | SES vs. BA | PEB vs. BA | SES vs. PES | PEB vs. PES vs. BA | PEB vs. BA | PEB vs. PES | SES vs. BA | SES vs. EES |
| No. of patients | 197 | 50 | 450 | 402 | 110 | 215 | 96 | 66 |
| No. of lesions | 202 | 50 | 483 | 500 | 110 | 221 | 96 | 66 |
| Type of DES restenosis | SES or PES | SES | SES | SES, EES, BES, ZES | SES ∗ , EES, PES | N/A | N/A | N/A |
| Mean age (years) | 68 | 69 | 67 | 67 | 67 | 62 | 63 | 63 |
| Men (%) | 71 | 86 | 76 | 71 | 71 | 87 | 73.9 | 51.5 |
| Diabetes (%) | 39 | 62 | 36 | 41.5 | 35.4 | 36.7 | 34.4 | 30.3 |
| Focal vs. diffuse restenosis (%) | N/A | 58 vs. 42 | 76 vs. 24 | 66.8 vs. 33.2 | 65.4 vs. 34.6 | 63.4 vs. 36.6 | 100 vs. 0 | 0 vs. 100 |
| Lesion length (mm) | 14.6 | 13 | 12.6 | N/A | 11.7 | 12.8 | 8 | 22.3 |
| RVD (mm) | 2.6 | 2.8 | 2.76 | 2.75 | 2.29 | 2.69 | 3.05 | 3.31 |
| Dual antiplatelet therapy prescription (months) | 6 or 1 † | 3 | 6 | 6 | 6 | 12 | 6 | 6 |
| Angiographic follow-up (months) | 9–12 | 6 | 6.5 | 6.5 | 6 | 9 | 9.3 | 9.5 |
| Diameter stenosis at follow-up (%) | 21 ± 19.2 vs. 29.8 ± 18.5 | 34.2 ± 15.2 vs. 58 ± 22.7 | 26.6 ± 23.6 vs. 25.4 ± 21.5 | 38 ± 21.5 vs. 37.4 ± 21.8 vs. 54.1 ± 25 | 29.6 ± 24.3 vs. 51.1 ± 31 | 29 ± 21.3 vs. 30.8 ± 25.3 | 13.5 ± 14.8 vs. 20.9 ± 25.2 | 23.8 ± 20.2 vs. 20 ± 13.3 |
| In-stent late lumen loss (mm) | 0.37 ± 0.57 vs. 0.41 ± 0.63 | 0.17 ± 0.45 vs. 0.72 ± 0.56 | 0.40 ± 0.65 vs. 0.38 ± 0.59 | 0.37 ± 0.59 vs. 0.37 ± 0.61 vs. 0.70 ± 0.69 | 0.43 ± 0.61 vs. 1.03 ± 0.77 | 0.46 ± 0.51 vs. 0.55 ± 0.61 | 0.012 ± 0.14 vs. 0.37 ± 0.53 | 0.15 ± 0.26 vs. 0.12 ± 0.25 |
| Angiographic restenosis (%) | 11.1 vs. 14 | 8.7 vs. 62.5 | 19.6 vs. 20.6 | 27 vs. 24 vs. 57 | 17.2 vs. 58.1 | 17.5 vs. 21.4 | 3.1 vs. 17.2 | 5 vs. 14.3 |
| Target-lesion revascularization (%) | 5.9 vs. 13.1 | 4.3 vs. 41.7 | 16.6 vs. 14.6 | 22.1 vs. 13.5 vs. 43.5 | 15.3 vs. 36.8 | 14.7 vs. 10.4 | 0 vs. 6.3 | 3.1 vs. 5.8 |
| Year of publication | 2012 | 2011 | 2010 | 2012 | 2012 | 2014 | 2012 | 2012 |
∗ Yukon stent (Translumina, Hechingen, Germany) included.
| Treatment | Diameter Stenosis at Angiographic Follow-Up (%) Mean Difference (95% CI) | Late Lumen Loss (mm) Mean Difference (95% CI) | Restenosis Odds Ratio (95% CI) | Target-Lesion Revascularization Odds Ratio (95% CI) |
|---|---|---|---|---|
| PEB versus: | ||||
| PES | −2.40 (−9.82, 4.16) | −0.07 (−0.34, 0.16) | 0.76 (0.21, 2.28) | 1.48 (0.53, 4.02) |
| SES | −6.59 (−16.61, 1.24) | −0.16 (−0.51, 0.12) | 0.64 (0.12, 2.56) | 1.12 (0.30, 4.24) |
| EES | −2.67 (−20.22, 12.87) | −0.12 (−0.66, 0.36) | 0.16 (0.00, 3.82) | 0.42 (0.01, 14.86) |
| BA | −17.74 (−25.17, −11.31) | −0.43 (−0.68, −0.22) | 0.18 (0.05, 0.45) | 0.29 (0.09, 0.57) |
| EES versus: | ||||
| DEB | 2.67 (−12.87, 20.22) | 0.12 (−0.36, 0.66) | 6.36 (0.26, 445.85) | 2.38 (0.07, 112.12) |
| PES | 0.25 (−15.19, 17.02) | 0.05 (−0.41, 0.56) | 4.70 (0.19, 299.92) | 3.44 (0.10, 154.95) |
| SES | −3.8 (−18.08, 10.14) | −0.03 (−0.43, 0.36) | 3.80 (0.22, 181) | 2.60 (0.10, 102.53) |
| BA | −14.93 (−33.47, 1.16) | −0.30 (−0.77, 0.18) | 1.10 (0.05, 64.82) | 0.65 (0.02, 26.86) |
| PES versus: | ||||
| DEB | 2.40 (−4.16, 9.82) | 0.07 (−0.16, 0.34) | 1.31 (0.44, 4.66) | 0.68 (0.25, 1.89) |
| SES | −4.14 (−13.12, 2.8) | −0.09 (−0.40, 0.18) | 0.84 (0.20, 3.27) | 0.76 (0.24, 2.52) |
| EES | −0.25 (−17.02, 15.19) | −0.05 (−0.56, 0.41) | 0.21 (0.00, 5.29) | 0.29 (0.01, 9.64) |
| BA | −15.3 (−22.96, −8.35) | −0.35 (−0.62, −0.10) | 0.24 (0.06, 0.75) | 0.20 (0.05, 0.45) |
| SES versus: | ||||
| DEB | 6.59 (−1.24, 16.61) | 0.16 (−0.12, 0.51) | 1.57 (0.39, 8.14) | 0.89 (0.24, 3.38) |
| PES | 4.14 (−2.8, 13.12) | 0.09 (−0.18, 0.40) | 1.20 (0.31, 5.01) | 1.32 (0.40, 4.23) |
| EES | 3.8 (−10.14, 18.08) | 0.03 (−0.36, 0.43) | 0.26 (0.01, 4.61) | 0.39 (0.01, 9.76) |
| BA | −11.08 (−17.89, −3.4) | −0.27 (−0.51, 0.00) | 0.29 (0.07, 0.99) | 0.25 (0.06, 0.68) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
