lable at ScienceDirect

Atherosclerosis 237 (2014) 410e417

Contents lists avai

Atherosclerosis

journal homepage: www.elsevier .com/locate/atherosclerosis

ORAl iMmunosuppressive therapy to prevent in-Stent rEstenosiS(RAMSES) cooperation: A patient-level meta-analysis of randomizedtrials

Salvatore Cassese a, *, 1, Giuseppe De Luca b, 1, Flavio Ribichini c, Carmelo Cernigliaro d,Mara Sansa e, Francesco Versaci f, g, Igino Proietti h, Goran Stankovic i, Sinisa Stojkovic i,Carlos Fernandez-Pereira j, Fabrizio Tomai k, Corrado Vassanelli c, David Antoniucci l,Patrick W. Serruys m, Adnan Kastrati a, Alfredo E. Rodriguez j

a Deutsches Herzzentrum, Technische Universit€at München, Lazarettstr. 36, 80636 Munich, Germanyb Department of Clinical and Experimental Medicine, Eastern Piedmont University, Novara, Italyc Department of Medicine, University of Verona, Verona, Italyd Interventional Cardiology, San Gaudenzio Clinic, Novara, Italye Catheterization Laboratory, Division of Cardiology, Maggiore Hospital, Novara, Italyf Department of Cardiology, Cardarelli Hospital, Campobasso, Italyg Department of Cardiology, Veneziale Hospital, Isernia, Italyh Department of Cardiology, Vannini Hospital, Rome, Italyi Department of Diagnostics and Catheterization Laboratory, Division of Cardiology, Clinical Center of Serbia, Medical School of Belgrade, Serbiaj Cardiovascular Research Center, Sanatorio Otamendi, Buenos Aires, Argentinak Department of Cardiovascular Sciences, European Hospital, Rome, Italyl Division of Cardiology, Careggi Hospital, Florence, Italym Thorax Center, Erasmus University Medical Center, Rotterdam, The Netherlands

a r t i c l e i n f o

Article history:Received 6 April 2014Received in revised form31 August 2014Accepted 5 September 2014Available online 30 September 2014

Keywords:Oral immunosuppressive therapyCoronary stentsRestenosisMeta-analysis

* Corresponding author.E-mail addresses: cassese@dhm.mhn.de, salvatore

1 Contributed equally to the manuscript and are jo

http://dx.doi.org/10.1016/j.atherosclerosis.2014.09.0210021-9150/© 2014 Elsevier Ireland Ltd. All rights rese

a b s t r a c t

Objective: The role of oral immunosuppressive therapy (OIT) to prevent restenosis after percutaneouscoronary intervention (PCI) and stenting is still controversial. This study evaluates the impact of oraladministration of prednisone or sirolimus to prevent restenosis. Methods: We conducted a meta-analysis of trials in which PCI-patients were randomized to bare metal stents (BMS) plus OIT(BMS þ OIT group) versus BMS or drug-eluting stents alone (BMS/DES group). Primary endpoints weretarget lesion revascularization and death/myocardial infarction (MI). Secondary endpoints were death,MI, stent thrombosis and in-stent late lumen loss. Hazard ratio and weighted geometric mean difference[95% confidence intervals] served as summary statistics. Results: Individual data of seven trials (1246patients [BMS þ OIT, n ¼ 608 versus BMS/DES, n ¼ 638] with 1456 coronary lesions) were merged. At amedian follow-up of 360 days, BMS þ OIT versus BMS/DES significantly reduced the risk of revascu-larization (0.49 [0.24e0.98], P ¼ 0.04). In particular, BMS þ OIT reduced the risk of revascularization (0.38[0.21e0.67], P < 0.001) and late lumen loss (�0.39 mm [�0.67, �0.11], P < 0.001) as compared with BMSalone. BMS þ OIT versus BMS/DES showed a similar risk of death/MI (0.67 [0.29e1.53], P ¼ 0.34), death(0.82 [0.25e2.69], P ¼ 0.71), MI (0.58 [0.24e1.39], P ¼ 0.22) and stent thrombosis (0.43 [0.10e1.87],P ¼ 0.26). Conclusion: In patients undergoing PCI the use of BMS and oral immunosuppressive therapyreduces the risk of revascularization as compared with BMS alone but not as compared with DES alone,while these therapies display a similar risk of death/MI. The advantage of adding oral immunosup-pressive therapy to BMS is due to a lower risk of restenosis as compared with BMS alone.

© 2014 Elsevier Ireland Ltd. All rights reserved.

cassese@libero.it (S. Cassese).int first authors.

rved.

1. Introduction

Percutaneous coronary intervention (PCI) represents the mostfrequent revascularization option in patients suffering from

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417 411

coronary artery disease (CAD) [1] and has become synonymouswith stenting.

The stent restenosis is the major drawback of contemporary PCI[2]. Being the inflammatory reaction at the stented site one of theprincipal components of restenosis [3], the local use of immuno-suppressive therapy is reasonable in patients undergoing PCI. Theuse of stents eluting immunosuppressive drugs (drug-elutingstents, DES) aimed at minimize the high frequency of restenosisobserved after PCI with bare metal stents (BMS) [4,5]. Indeed, DESreduced the relative risk of reintervention from 50% to 70% ascompared to BMS [6] though at the expense of a higher risk ofthrombosis time after PCI, particularly with first-generation plat-forms [7].

Notwithstanding the merits of DES, BMS have not been aban-doned [2]. Concerns about a higher risk of bleeding or non-compliance with the mandatory dual antiplatelet therapy suggestsome relative contraindications to DES in PCI-patients with specificclinical and socioeconomic conditions [8,9]. Despite socioeconomicconsiderations are more relevant in certain geographic areas, BMSare still used in one fourth of patients receiving a PCI in the US [10].Thus, a considerable proportion of contemporary PCI-patients areat risk of restenosis since not receiving DES.

In concomitance with the widespread adoption of DES, anobvious question was whether the immunosuppressive drugsacting locally at stented site would have similar efficacy and safetywhen administered systemically. For this reason, the immunosup-pressive efficacy of steroids [11] or sirolimus (the inhibitor ofmammalian target of rapamycin) [12] has earlier been studied inPCI-patients with conflicting results. Thus, the role of oral immu-nosuppressive therapy (OIT) in patients treated with coronarystents still remains a matter of controversy.

The aim of the oRAl iMmunosuppressive therapy to prevent in-Stent rEstenosiS (RAMSES) cooperation was to perform anindividual-patient pooled-analysis of randomized trials investi-gating the outcomes of OIT with prednisone or sirolimus in PCI-patients treated with stents.

2. Methods

2.1. Search strategy and selection criteria

We searchedMedline, EMBASE, the Cochrane Central Register ofControlled Trials (CENTRAL), scientific sessions abstracts and rele-vant websites (www.cardiosource.com, www.clinicaltrialresults.org, www.escardio.org, www.tctmd.com, www.theheart.org),without language or publication status restrictions. Reference listsof the eligible studies and previous reviews [11,12] were checked toidentify further evaluable articles. The last search was run on 1March 2013. Search terms included: “percutaneous coronaryintervention”, “stent(s)”, “oral immunosuppressive therapy”,“sirolimus”, “rapamycin”, “cortisone”, “prednisone”, “trial” and“randomized trial”. Inclusion criteria were: (1) randomized design;(2) intention-to-treat analysis; and (3) �6-month follow-up aftertreatment allocation. Exclusion criteria were: (1) oral therapy otherthan immunosuppressive as adjunct to standard of care, (2)duplicated data and (3) ongoing trials. We discarded those trials,which did not study OIT after coronary stenting.

2.2. Data collection and assessment of risk of bias

Two independent investigators (SC and GDL) assessed reportsfor eligibility at title and/or at abstract level. A third investigator(AK) was contacted to resolve conflicts (consensus meeting).Studies that met inclusion criteria were selected for further anal-ysis. Freedom from bias was evaluated in accordance with the

Cochrane Collaboration method [13] based on the following items:adequacy of random sequence generation and allocation conceal-ment, blinding (at participants or outcome assessors level),description of incomplete outcome data reporting, depiction ofselective outcome, adequate description of sample size calculationand detailed funding sources disclosure. Formal quality scoreadjudication was not used, since previous investigations failed todemonstrate its usefulness [14].

All principal investigators agreed to share data of all patientsenrolled. Individual data were transferred without patient identi-fiers to the ISAResearch Center (Deutsches Herzzentrum, TechnischeUniversit€at München - Munich, Germany) and merged. The finaldataset was checked for completeness and consistency andcompared with the results from any publications. Principal in-vestigators were contacted in the case of inconsistency. Data weremanaged according to the intention-to-treat principle. All studiesincluded in the present analysis were approved from InstitutionalReview Boards.

2.3. Outcome variables

Primary endpoints were target lesion revascularization (TLR)and the composite of death/myocardial infarction (MI). Secondaryendpoints were death, MI, stent thrombosis (ST) and in-stent latelumen loss at angiographic surveillance. All endpoints were eval-uated according to per protocol definitions and events count washierarchical, considering always the most severe event when morethan one occurred.

2.4. Statistical analysis

A complete case analysis approach was used to manage missingoutcomes data assuming this latter was missing completely atrandom. Only subjects who had complete informations regardingoutcomes of interest entered the final analyses. Statistical analysiswas performed using Stata 11.2 (STATA Corp, College Station, Texas,USA). The к statistic was used to assess agreement between re-viewers for study selection. Categorical variables were presented asproportions, whilst continuous variables as median (interquartilerange) and mean ± SD. Hazard ratio (HR) [95% confidence intervals- 95% CI] served as summary statistics and was first estimated foreach trial and then combined using standard meta-analyticmethods. Survival analyses were performed for each trial usingthe ManteleCox method, which is not based on the proportionalhazards assumption. Treatment effect could not be assessed intrials in which no event was reported within groups. For trials inwhich only 1 of the treatment groups had no events of interest, thetreatment effect estimate and its standard error were approxi-mated from 2 � 2 contingency tables, after adding 0.5/cell [15]. Therandom effects model (DerSimonian and Laird) was used to poolHRs derived from individual trials and were calculated for BMS plusOIT (BMS þ OIT group) versus BMS or DES alone (BMS/DES group).The random effects model warrants a more conservative approachsince it accounts for the variation between studies and within eachindividual study, thus, typically yielding wider confidence intervals.The risk estimates were calculated overall, as well as stratified forimmunosuppressive therapies (for primary and secondary end-points) and stent type (for primary endpoints). In the case of sta-tistical significance, absolute risk reduction (ARR) and the numberneeded to treat (NNT) with [95% CI] were provided. Meta-analysisof angiographic data was performed according to Higgins et al.for meta-analyses of continuous variables with skewed distribution[16]. The weighted geometric mean difference [95% CI] served assummary statistic. The Breslow-Day c2 test (p < 0.1) and the I2

statistic tested the presence of heterogeneity across the studies

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417412

[13]. The potential for small-study effects was assessed both visu-ally and mathematically [17]. KaplaneMeier curves were displayedaccording to the treatments received.

A sensitivity analysis investigated whether some variables d

gender, age (under/above the median value), diabetes mellitus,acute coronary syndrome (ACS) at admission, angiographic sur-veillance, placebo use, sirolimus dose (under/above the medianvalue) d might have influenced the treatment-effect for the pri-mary endpoints. Two adjusted indirect comparisons according toBucher [18] and Song [19] investigated differences in the mainoutcomes between the immunosuppressive therapies (prednisoneversus sirolimus). An influence analysis, in which meta-analysisestimates are computed omitting one study at time, was per-formed for the primary endpoints. The study accomplished thePreferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [20].

3. Results

3.1. Eligible studies

We screened the title and/or the abstract of 210 potentiallyeligible publications (Fig. 1 d Supplementary Data). Of these, 170citations were excluded since they were not relevant to this studyor duplicated. Thus, 40 studies were assessed for eligibility and 33studies were eliminated as inclusion criteria were not met. Amongthese latter studies, four randomized trials were excluded sincepatients were treated with OIT after plain angioplasty [21e23] orwith short-term intravenous immunosuppressive therapy afterstenting [24]. Finally, 7 trials (all full-length manuscripts [25e31])enrolling a total of 1246 patients (608 randomized to BMS þ OITand 638 randomized to BMS/DES) and 1456 lesions (711 random-ized to BMS þ OIT and 745 randomized to BMS/DES) were includedin the meta-analysis. The inter-observer agreement for study se-lection was good (к 0.90).

In the present meta-analysis, the OIT consisted of prednisone orsirolimus in all cases. In two trials [25,26], 1 mg/kg/day of predni-sone was administered after PCI for 10e15 days with subsequenttapering until complete drug withdrawal after 40e45 days. In fivetrials [27e31], the sirolimus dose ranged between 2 and 3 mg/dayfor a period of time lasting 7e30 days after index procedure. Aloading dose of sirolimus (range 4e24 mg) was administered in 4trials [27e29,31]. The cumulative dose of sirolimus was 60 mg(52e62; 57.2 ± 6.8). One trial with a three-arm design [27]randomly allocated patients to high versus usual loading dose ofsirolimus versus control therapy: in this case, data of patientsbelonging to sirolimus arms were grouped to derive pooled esti-mates. Overall, the median OIT duration in the experimental armswas 14 days (7e40; 21.5 ± 14.3).

In the BMS/DES group, patients underwent PCI with BMS or DESalone, with [25,27e30] or without additional placebo [26,31]. Onetrial exclusively enrolled patients presenting restenosis afterstenting: in this trial the protocol of repeated intervention con-sisted of balloon angioplasty with provisional BMS [27].

The Table 1 d Supplementary Data presents an overview of thestudies included in the meta-analysis. Patients with significant CADand indication to PCI for de novo or restenotic lesions were ran-domized to BMS þ OIT versus BMS/DES.

Main common exclusion criteria among trials depended oncomorbidities precluding the safe administration of prednisone (i.e.diabetes mellitus, uncontrolled arterial hypertension, and recentMI d <2e6 weeks) or sirolimus (i.e. haematological disorders,infective diseases, and uncontrolled dyslipidaemia). Further com-mon key exclusions criteria among trials were bleeding diathesis orreduced life expectancy. Per protocol endpoints definitions are

listed in Table 2 d Supplementary Data. Each trial reported thatindependent committees adjudicated clinical and angiographicendpoints. The risk of bias among studies is reported in Table 3 d

Supplementary Data.Antithrombotic and anticoagulant therapies reflected the

changing practices during the period of observation: all patientsreceived a loading dose of thienopyridines, as well as aspirin at thetime of index intervention. At discharge, aspirinwas recommendedindefinitely, whilst thienopyridines were prescribed for a period oftime depending on clinical presentation or type of stent implanted(range 1e12 months). Anticoagulation during coronary in-terventions was mainly accomplished through administration ofunfractionated heparin. All interventions were performed inaccordance with standard of care. All study subjects received thetreatments assigned on top of other cardioactive therapies.

Clinical features were available for all patients (Table 1). Overall,patients had a median age of 64 years (57e71; 63.6 ± 9.8), werepredominantly males, up to one third diabetics and with a highfrequency of ACS at admission. Left ventricular function wasavailable for 1151 patients (90.7%) and showed a value of 57%(50e60; 55.1 ± 9.8). Angiographic features were available for 1446lesions (99.3%): median reference vessel diameter was 2.81 mm(2.50e3.18; 2.87 ± 0.55) and median lesion length was 13 mm(10e18; 14.28 ± 6.37).

3.2. Clinical endpoints

Of those randomized, 1238 patients (99.3%) were available foroutcomes assessment. Median follow-up was 360 days (260e1440;759 ± 548). KaplaneMeier curves for primary and secondaryendpoints are displayed according to the inclusion in the BMSþOIT[prednisone or sirolimus] or in the BMS/DES group (Fig. 2, PanelAeD d Supplementary Data). In addition, KaplaneMeier curvesfor primary and secondary endpoints are displayed according to thetype of stent used as comparator (Fig. 3, Panel AeD d

Supplementary Data and Fig. 4, Panel AeDd Supplementary Data).TLR occurred in 183 patients (14.7%; Fig. 1). BMS þ OIT versus

BMS/DES significantly decreased the risk of TLR, though the pres-ence of high heterogeneity (12.4% versus 17.0%; 0.49 [0.24e0.98],P ¼ 0.04; I2 ¼ 67%, P for heterogeneity - Phet ¼ 0.01; ARR ¼ 4.6%[0.6%, 8.5%]; NNT ¼ 22 [12e157]). When the risk estimates of TLRwere analysed according to the type of stent used as comparator,BMS þ OIT versus BMS lowered the risk of TLR without significantheterogeneity (Fig. 5 d Supplementary Data), while BMS þ OITversus DES showed a similar risk of TLR (8.8% versus 6.6%; 1.43[0.50e4.07], P ¼ 0.50; I2 ¼ 50%, Phet ¼ 0.16). A total of 316 patients(25.5%) with restenosis after stenting were randomized toBMS þ OIT versus BMS/DES: the treatment of in-stent restenosisversus de novo lesions did not modify the risk estimates for TLR (Pfor interaction ¼ 0.36).

The composite of death/MI occurred in 44 patients (3.5%, Fig. 2).BMSþ OIT versus BMS/DES showed a similar risk of death/MI (3.3%versus 3.7%; 0.67 [0.29e1.53], P ¼ 0.34; I2 ¼ 20%, Phet ¼ 0.27).BMS þ OIT versus BMS showed a similar risk of death/MI (Fig. 6 d

Supplementary Data). Similarly, BMS þ OIT versus DES showed asimilar risk of death/MI (2.2% versus 5.7%; 0.36 [0.12e1.07],P ¼ 0.07; I2 ¼ 0%, Phet ¼ 0.44).

Death occurred in 20 patients (1.6%, Fig. 3). BMS þ OIT versusBMS/DES showed a similar risk of death (1.6% versus 1.5%; 0.82[0.25e2.69], P ¼ 0.71; I2 ¼ 17%, Phet ¼ 0.30).

MI occurred in 24 patients (1.9%, Fig. 4). BMS þ OIT versus BMS/DES showed a similar risk of MI (1.5% versus 2.3%; 0.58 [0.24e1.39],P ¼ 0.22; I2 ¼ 0%, Phet ¼ 0.86).

ST occurred in 8 patients (1.2%, data available for 658 [53.1%]patients, three trials [25,26,31]). BMS þ OIT versus BMS/DES

Table 1Characteristics of patients included in the meta-analysis stratified by trial.

Trial Age, years Males, (%) Diabetes, (%) Acute coronarysyndrome, (%)

Left ventricularfunction, (%)

LAD, (%) Multivesselintervention, (%)

MLD pre, mm Surveillanceangiography, (%)

CEREA DES [26] 65 [58; 71] 85 N/A 59 60 [52; 60] 42 23 N/A N/ACernigliaro et al. [29] 62 [56; 71] 84 18 67 50 [44; 55] 50 14 1.03 [0.81; 1.41] 96IMPRESS [25] 65 [57; 71] 85 N/A 11 57 [52; 60] 53 N/A 0.90 [0.74; 1.05] 98ORAR II [28] 64 [59; 71] 90 16 65 N/R 46 22 1.00 [0.59; 1.32] 84ORAR III [31] 63 [55; 71] 82 27 59 56 [49; 60] 47 35 N/A N/AOSIRIS [27] 65 [59; 73] 78 29 N/A 58 [49; 54] 49 N/A 0.85 [0.61; 1.12] 87Stojkovic et al. [30] 59 [51; 54] 70 29 52 55 [50; 60] 41 7 0.86 [0.60; 1.05] 90

Median values and proportions are reported. LAD: left anterior descending artery; MLD: minimum lumen diameter; N/A: not applicable; N/R: not reported. Trial acronyms:CEREA DES: Cortisone plus BMS or DES versus BMS alone to eliminate restenosis; IMPRESS: Immunosuppressive Therapy for the Prevention of Restenosis After CoronaryArtery Stent Implantation; ORAR: Oral Rapamycin in Argentina; OSIRIS: Oral Sirolimus to Inhibit Recurrent In-stent Stenosis.

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417 413

showed a similar risk of ST (0.7% versus 1.5%; 0.43 [0.10e1.87],P ¼ 0.26; I2 ¼ 0%, Phet ¼ 0.77). Due to the paucity of data, we didneither build a Forest plot nor present KaplaneMeier curves. EarlyST (�30 days after index procedure) occurred in 3 patients (0.3%versus 0.5%; 0.83 [0.10e6.85], P ¼ 0.86; I2 ¼ 0%, Phet ¼ 0.85). Late ST(>30e360 days after index procedure) occurred in 5 patients (0.3%versus 1.0%; 0.66 [0.01e3.69], P ¼ 0.65). Among those patientsrandomized to BMS/DES and suffering ST, 1 patient had receivedBMS and 5 patients had received DES.

The proportion of patients complaining side effects associatedwith OIT was 13.4% and 1.1% required permanent OIT discontinu-ation. Prednisone and sirolimus did not differ in the proportion ofside effects (11.4% versus 14.1%, respectively, P ¼ 0.46; Fig. 7 e

Supplementary Data).

3.3. Angiographic endpoint

Among those studies [25,27e30] with angiographic surveillanceat 6e9 months after index procedure, a total of 602 patients (89.7%,

Fig. 1. Risk estimates of target lesion revascularization associated with BMS þ OIT versus Bconfidence intervals, CI] of TLR associated with BMS þ OIT versus BMS/DES. The squares andthe size of each square is proportional to the statistical weight of a trial in the meta-analysisindicating the point estimate and the left and the right ends the [95% CI]. OIT: oral immunosurevascularization.

666 lesions) randomized to BMS þ OIT versus BMS were availablefor assessment of late lumen loss. Patients enrolled in theremaining trials did not undergo routine angiographic follow-up.Overall, BMS þ OIT versus BMS significantly reduced late lumenloss (Fig. 5).

3.4. Small-study effects, influence and sensitivity analyses, adjustedindirect comparison

Funnel plot distribution of primary endpoints was derived fromthe standard error of the natural logarithm HR plotted against theHR of primary endpoints. Of note, the absence of bias due to small-study effects was confirmed both visually and mathematically(Fig. 8, Panel AeB d Supplementary Data). Additionally, influenceanalyses demonstrated that no single study significantly altered thesummary HR for primary endpoints (Fig. 9, Panel AeB d

Supplementary Data). The pre-specified covariates did not modifythe risk estimates for primary endpoints (Fig. 10, Panel AeB d

Supplementary Data). At adjusted indirect comparison PCI-

MS/DES. Hazard ratio (HR, random effects model d DerSimonian and Laird) with [95%the horizontal lines indicate the HR and the [95% CI] respectively for each trial included;; the diamonds indicate the effect estimate derived from meta-analysis, with the centreppressive therapy; BMS: bare metal stents; DES: drug-eluting stents; TLR: target lesion

Fig. 2. Risk estimates of death/myocardial infarction associated with BMS þ OIT versus BMS/DES. Hazard ratio (HR, random effects modeld DerSimonian and Laird) with [95% CI] ofdeath/MI associated with BMS þ OIT versus BMS/DES. MI: myocardial infarction. Other abbreviations are as in Fig. 1.

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417414

patients randomized to either prednisone or sirolimus showed asimilar risk of TLR (1.05 [0.12e8.72], P ¼ 0.96; I2 ¼ 87%) and death/MI (0.37 [0.05e3.07], P ¼ 0.36; I2 ¼ 28%).

4. Discussion

This meta-analysis investigates the impact of oral immuno-suppressive therapy to prevent in-stent restenosis. At 1-year

Fig. 3. Risk estimates of death for BMS þ OIT versus BMS/DES. Hazard ratio (HR, random effeversus BMS/DES. Other abbreviations are as in Fig. 1.

follow-up the main findings are: i) BMS þ OIT versus BMS/DESreduced the risk of TLR by virtue of higher antirestenotic efficacy ascompared to BMS ii) without affecting the risk of death, MI and ST.

DES significantly lowered restenosis in comparison with BMS[32]. The improved clinical outcome seen with DES facilitated thewidespread use to increasingly complex patient and lesion subsetsand significantly reduced the proportion of BMS implanted [33].Notwithstanding the undisputed merits associated with DES,

cts model d DerSimonian and Laird) with [95% CI] of death associated with BMS þ OIT

Fig. 4. Risk estimates of myocardial infarction for BMS þ OIT versus BMS/DES. Hazard ratio (HR, random effects model d DerSimonian and Laird) with [95% CI] of MI associatedwith BMS þ OIT versus BMS/DES. MI: myocardial infarction. Other abbreviations are as in the Fig. 1.

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417 415

nowadays one fourth of PCI-patients in the US receive only BMS[10]. In addition, some safety considerations from cardiovascularguideline-writing authorities suggest that specific clinical condi-tions (increased bleeding risk, planned extra-cardiac surgery, co-morbid conditions and polypharmacy) or socioeconomic aspects(reimbursement policy, limited access to medications) representrelative contraindications to DES even in the era of safer second-

Fig. 5. In-stent late lumen loss for BMS þ OIT versus BMS. Weighted geometric mean differensquares and the horizontal lines indicate the geometric mean difference with [95% CI]: for efor final analysis and presentation; the diamonds indicate the effect estimate derived fromends the [95% CI]. Other abbreviations are as in the Fig. 1.

generation platforms [8,9] As a result, a significant proportion ofPCI-patients, as those living in developing countries, may beexposed at higher risk of restenosis since not receiving DES [4].

The key feature of DES is the local inhibition of smooth musclecell proliferation and migration after stenting by progressivelyreleasing into the vessel wall drugs with an immunosuppressiveaction [34]. Whether the oral administration of such drugs after PCI

ce with [95% CI] of in-stent late lumen loss associated with BMS þ OIT versus BMS. Theach trial results were transferred on log-transformed scales and then back transformedmeta-analysis, with the centre indicating the point estimate and the left and the right

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417416

would replicate the effect achieved in situ with DES has previouslyfocused attention [12]. The anti-inflammatory effect of steroids wasinitially studied in animal models [35], and in patients receivingPCI, with uncertain results: the use of plain angioplasty [21,22],different potency, low dosages and very short duration of thetreatments [24] were likely themain reasons for this lack of efficacy[11]. Similarly, earlier studies tested the antiproliferative efficacy ofsirolimus after plain angioplasty, but results were inconclusive[23,36] Thereafter, small randomized trials showed some benefitsof oral sirolimus after stenting [27,28], though the number of pa-tients limited definitive conclusions.

This meta-analysis merges for the first time the datasets of 7randomized trials investigating BMS þ OIT versus BMS/DES toprevent restenosis. As main finding, BMS þ OIT versus BMS/DESdecreases TLR, although the risk estimates are only borderlinesignificant. OIT significantly reduces the degree of late lumen loss,thus providing a mechanistic explanation for the observed clinicaladvantages. However, these findings should be interpreted withcaution. Firstly, BMSþOIT reduces the risk of TLR as comparedwithBMS alone, but not as compared with DES alone. Secondly, angio-graphic follow-up data was available only for patients receivingBMS.

BMS þ OIT versus BMS/DES displayed a similar risk of death, MIand ST. These results merit careful consideration. As first, DESinhibit neointimal hyperplasia by achieving high drug concentra-tionwithin a limited vessel surface. Although the potential for localvascular toxicity [37], there is no measurable systemic effect. At theopposite, the systemic administration of immunosuppressive drugsmay reduce reactions at stented site, butmay hypothetically elevatethe proportion of adverse reactions or unsought consequences(such as new malignancies). In accordance with previous data[38,39] this meta-analysis shows that a therapy with BMS þ OIT incomparison with BMS/DES does not increase mortality at 1-yearfollow-up. This follow-up may be insufficient to capture latesequelae after OIT. However, in those studies with a longer follow-up, [29,38,40,41] the occurrence of de novo malignancies, newlydiagnosed diabetes mellitus or auto-immune diseases do notsignificantly differ from the general population. As second, thepresent study is not powered to discriminate differences in rareroutcomes such as MI and ST, thus precluding an exhaustiveinvestigation of the safety profile of BMS þ OIT versus DES. Inaddition, the present analysis pooled only platforms belonging tofirst-generation DES, though recent evidence supports a superiorsafety of second-generation DES over both first-generation DES aswell as BMS [42]. Therefore, it remains unstudied whether astrategy of BMS þ OIT may still have a role in the contemporaryinterventional portfolio in light of safer new DES with metallicplatforms.

The scarce compliance remains an indication for BMS implan-tation among patients undergoing PCI. In accordance with theduration of OIT reported in this meta-analysis, the compliance withOIT is supposed a minor issue. In addition, the side effects observedwith OIT had a low frequency, were well tolerated, and led in a veryfew number of cases to definitive study drug discontinuation.Prednisone and sirolimus showed similar efficacy and safety pro-files. These considerations notwithstanding, sirolimus might bepreferred over prednisone since it could be administered to awiderspectrum of high-risk patients irrespective of the diabetes status orthe history of a recent MI.

The risk estimates for TLR associated with BMS þ OIT versusBMS/DES appeared to be independent from gender, median age,diabetes mellitus, ACS as clinical presentation, the performance ofangiographic surveillance, the use of placebo in the control arms orthe dose of sirolimus in the experimental arm. However, the pop-ulation enrolled resulted from randomized trials, for which

somewhat restrictive inclusion/exclusion criteria have beenapplied, a fact that further limits the generalizability of the presentfindings. Moreover, these analyses are post-hoc and have to beintended as hypothesis generating and sensitive to the play ofchance.

4.1. Study limitations

This meta-analysis has some limitations. The overall number ofevents observed was modest since the majority of the trialsincluded originally focused on the angiographic efficacy ofBMS þ OIT versus BMS/DES in patients undergoing PCI. Eventhough in this setting this represents the largest cohort of patientspooled so far, the relative small number of patients and the selectednature per se of the population preclude fully exploring differencesin relatively infrequent adverse events, such MI and ST. The in-formations regarding MI type (Q-wave or non-Q-wave) orregarding DAPT duration were not routinely collected in the orig-inal trials. The risk estimates for clinical endpoints were derived bygrouping different immunosuppressive therapies as well as stentplatforms. The type of stent (BMS or DES) was responsible for theheterogeneity observed in the risk estimates for TLR. However,unknown sources of heterogeneity cannot be definitively excluded.The lack of per protocol definitions precludes considering the sideeffects associated with OIT as endpoint. The median follow-up of360 days available for this analysis is not sufficient to evaluate thesystemic side effects of OIT, such as malignancies: a longer clinicalfollow-up would shed more light on late events associated with thetreatments of interest. Data of angiographic surveillance wasavailable only for patients treated with BMS, precluding definiteconclusions on the antirestenotic efficacy of BMS þ OIT versus DES.Finally, the cost-effectiveness of a strategy of OIT after stenting ascompared to first-generation DES cannot be assessed with thepresent meta-analysis, since studies were performed withindifferent health care systems.

In conclusion, in patients undergoing PCI the addition of oralimmunosuppressive therapy to BMS reduces the risk of TLR ascompared with BMS alone but not as compared with DES alone,while there is no difference in the composite of death/MI betweenthese therapies. The advantage of adding oral immunosuppressivetherapy to BMS is due to a lower risk of vessel renarrowing ascompared with BMS alone.

Conflict of interest disclosures/funding

AK has submitted patents in relation to a number of DES tech-nologies and received consulting or lecture fees from Abbott, Bio-sensors and Biotronik. The other authors declare no potentialconflict of interest. No extramural funding was used for the currentstudy.

Authors contributions

SC and GDL had full access to all the data used in the study andtake the responsibility for the integrity of the data and the accuracyof the analysis. Study concept and design: GDL, AK and AER. Acqui-sition of the data: SC, GDL, FR, CC, FV, SS, AK, and AER. Analysis andinterpretation of the data: SC, GDL and AK. Draughting of themanuscript: SC and GDL. Critical revision of the manuscript forimportant intellectual content: MS, IP, GS, CF-P, FT, CV, DA and PWS.Statistical analysis: SC. Administrative, technical or material support:not applicable. Study supervision: GDL, AK and AER. All authorsapproved the manuscript for final submission.

S. Cassese et al. / Atherosclerosis 237 (2014) 410e417 417

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atherosclerosis.2014.09.021.

References

[1] Nabel EG, Braunwald E. A tale of coronary artery disease and myocardialinfarction. N. Engl. J. Med. 2012;366:54e63.

[2] Alfonso F, Byrne RA, Rivero F, et al. Current treatment of in-stent restenosis.J. Am. Coll. Cardiol. 2014 Jun 24;63(24):2659e73.

[3] Farb A, Sangiorgi G, Carter AJ, et al. Pathology of acute and chronic coronarystenting in humans. Circulation 1999;99:44e52.

[4] Kastrati A, Mehilli J, Pache J, et al. Analysis of 14 trials comparing sirolimus-eluting stents with bare-metal stents. N. Engl. J. Med. 2007;356:1030e9.

[5] Dangas GD, Claessen BE, Caixeta A, et al. In-stent restenosis in the drug-eluting stent era. J. Am. Coll. Cardiol. 2010;56:1897e907.

[6] Amin AP, Spertus JA, Cohen DJ, et al. Use of drug-eluting stents as a function ofpredicted benefit: clinical and economic implications of current practice. Arch.Intern. Med. 2012;172:1145e52.

[7] Tada T, Byrne RA, Simunovic I, et al. Risk of stent thrombosis among bare-metal stents, first-generation drug-eluting stents, and second-generationdrug-eluting stents: results from a registry of 18,334 patients. JACC Car-diovasc. Interv. 2013;6:1267e74.

[8] Task Force on Myocardial Revascularization of the European Society of Car-diology, the European Association for Cardio-Thoracic Surgery, European As-sociation for Percutaneous Cardiovascular Interventions, et al. Guidelines onmyocardial revascularization. Eur. Heart J. 2010;31:2501e55.

[9] Levine GN, Bates ER, Blankenship JC, et al. 2011 ACCF/AHA/SCAI guideline forpercutaneous coronary intervention: a report of the American College ofCardiology Foundation/American Heart Association Task Force on PracticeGuidelines and the Society for Cardiovascular Angiography and Interventions.Circulation 2011;124:e574e651.

[10] Dehmer GJ, Weaver D, Roe MT, et al. A contemporary view of diagnosticcardiac catheterization and percutaneous coronary intervention in the UnitedStates: a report from the CathPCI registry of the National Cardiovascular DataRegistry, 2010 through June 2011. J. Am. Coll. Cardiol. 2012;60:2017e31.

[11] Sardar P, Chatterjee S, Mukherjee D, et al. Steroids for the prevention ofrestenosis in bare-metal stents-a systematic review and meta-analysis.J. Invasive Cardiol. 2012;24:98e103.

[12] Rodriguez AE. Emerging drugs for coronary restenosis: the role of systemicoral agents the in stent era. Expert Opin. Emerg. Drugs 2009;14:561e76.

[13] Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. Br. Med. J. 2003;327:557e60.

[14] Juni P, Witschi A, Bloch R, et al. The hazards of scoring the quality of clinicaltrials for meta-analysis. J. Am. Med. Assoc. 1999;282:1054e60.

[15] Schomig A, Mehilli J, de Waha A, et al. A meta-analysis of 17 randomized trialsof a percutaneous coronary intervention-based strategy in patients withstable coronary artery disease. J. Am. Coll. Cardiol. 2008;52:894e904.

[16] Higgins JP, White IR, Anzures-Cabrera J. Meta-analysis of skewed data:combining results reported on log-transformed or raw scales. Stat. Med.2008;27:6072e92.

[17] Sterne JA, Egger M, Smith GD. Systematic reviews in health care: investigatingand dealing with publication and other biases in meta-analysis. Br. Med. J.2001;323:101e5.

[18] Bucher HC, Guyatt GH, Griffith LE, et al. The results of direct and indirecttreatment comparisons in meta-analysis of randomized controlled trials.J. Clin. Epidemiol. 1997;50:683e91.

[19] Song F, Altman DG, Glenny AM, et al. Validity of indirect comparison forestimating efficacy of competing interventions: empirical evidence frompublished meta-analyses. Br. Med. J. 2003;326:472.

[20] Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematicreviews and meta-analyses: the PRISMA statement. Ann. Intern. Med.2009;151:264e9. W264.

[21] Stone GW, Rutherford BD, McConahay DR, et al. A randomized trial of corti-costeroids for the prevention of restenosis in 102 patients undergoing repeatcoronary angioplasty. Cathet. Cardiovasc. Diagn. 1989;18:227e31.

[22] Pepine CJ, Hirshfeld JW, Macdonald RG, et al. A controlled trial of cortico-steroids to prevent restenosis after coronary angioplasty. M-HEART Group.Circulation 1990;81:1753e61.

[23] Brara PS, Moussavian M, Grise MA, et al. Pilot trial of oral rapamycin forrecalcitrant restenosis. Circulation 2003;107:1722e4.

[24] Lee CW, Chae JK, Lim HY, et al. Prospective randomized trial of corticosteroidsfor the prevention of restenosis after intracoronary stent implantation. Am.Heart J. 1999;138:60e3.

[25] Versaci F, Gaspardone A, Tomai F, et al. Immunosuppressive therapy for theprevention of restenosis after coronary artery stent implantation (IMPRESSstudy). J. Am. Coll. Cardiol. 2002;40:1935e42.

[26] Ribichini F, Tomai F, De Luca G, et al. Immunosuppressive therapy with oralprednisone to prevent restenosis after PCI. A multicenter randomized trial.Am. J. Med. 2011;124:434e43.

[27] Hausleiter J, Kastrati A, Mehilli J, et al. Randomized, double-blind, placebo-controlled trial of oral sirolimus for restenosis prevention in patients with in-stent restenosis: the Oral Sirolimus to Inhibit Recurrent In-stent Stenosis(OSIRIS) trial. Circulation 2004;110:790e5.

[28] Rodriguez AE, Granada JF, Rodriguez-Alemparte M, et al. Oral rapamycin aftercoronary bare-metal stent implantation to prevent restenosis: the Prospec-tive, Randomized Oral Rapamycin in Argentina (ORAR II) study. J. Am. Coll.Cardiol. 2006;47:1522e9.

[29] Cernigliaro C, Sansa M, Vitrella G, et al. Preventing restenosis after implan-tation of bare stents with oral rapamycin: a randomized angiographic andintravascular ultrasound study with a 5-year clinical follow-up. Cardiology2010;115:77e86.

[30] Stojkovic S, Ostojic M, Nedeljkovic M, et al. Systemic rapamycin withoutloading dose for restenosis prevention after coronary bare metal stent im-plantation. Catheter. Cardiovasc. Interv. 2010;75:317e25.

[31] Rodriguez AE, Rodriguez-Granillo AM, Antoniucci D, et al. Randomized com-parison of cost-saving and effectiveness of oral rapamycin plus bare-metalstents with drug-eluting stents: three-year outcome from the randomizedoral rapamycin in Argentina (ORAR) III trial. Catheter. Cardiovasc. Interv.2012;80:385e94.

[32] Bangalore S, Kumar S, Fusaro M, et al. Short- and long-term outcomes withdrug-eluting and bare-metal coronary stents: a mixed-treatment comparisonanalysis of 117 762 patient-years of follow-up from randomized trials. Cir-culation 2012;125:2873e91.

[33] Beohar N, Davidson CJ, Kip KE, et al. Outcomes and complications associatedwith off-label and untested use of drug-eluting stents. J. Am. Med. Assoc.2007;297:1992e2000.

[34] Stefanini GG, Holmes Jr DR. Drug-eluting coronary-artery stents. N. Engl. J.Med. 2013;368:254e65.

[35] Ribichini F, Joner M, Ferrero V, et al. Effects of oral prednisone after stenting ina rabbit model of established atherosclerosis. J. Am. Coll. Cardiol. 2007;50:176e85.

[36] Waksman R, Ajani AE, Pichard AD, et al. Oral rapamycin to inhibit restenosisafter stenting of de novo coronary lesions: the Oral Rapamune to InhibitRestenosis (ORBIT) study. J. Am. Coll. Cardiol. 2004;44:1386e92.

[37] Joner M, Finn AV, Farb A, et al. Pathology of drug-eluting stents in humans:delayed healing and late thrombotic risk. J. Am. Coll. Cardiol. 2006;48:193e202.

[38] Kufner S, Hausleiter J, Ndrepepa G, et al. Long-term risk of adverse outcomesand newmalignancies in patients treated with oral sirolimus for prevention ofrestenosis. JACC Cardiovasc. Interv. 2009;2:1142e8.

[39] Rodriguez AE, Maree A, Tarragona S, et al. Percutaneous coronary interventionwith oral sirolimus and bare metal stents has comparable safety and efficacyto treatment with drug eluting stents, but with significant cost saving: long-term follow-up results from the randomised, controlled ORAR III (OralRapamycin in ARgentina) study. EuroIntervention 2009;5:255e64.

[40] Ribichini F, Tomai F, Pesarini G, et al. Long-term clinical follow-up of themulticentre, randomized study to test immunosuppressive therapy with oralprednisone for the prevention of restenosis after percutaneous coronary in-terventions: cortisone plus BMS or DES veRsus BMS alone to EliminAteRestenosis (CEREA-DES). Eur. Heart J. 2013;34:1740e8.

[41] Rodriguez AE, Palacios I, Rodriguez-Granillo AM, et al. Comparison of cost-effectiveness of oral rapamycin plus bare-metal stents versus first genera-tion of drug-eluting stents (from the Randomized Oral Rapamycin inArgentina [ORAR] 3 trial). Am. J. Cardiol. 2014;113:815e21.

[42] Sarno G, Lagerqvist B, Frobert O, et al. Lower risk of stent thrombosis andrestenosis with unrestricted use of 'new-generation' drug-eluting stents: areport from the nationwide Swedish Coronary Angiography and AngioplastyRegistry (SCAAR). Eur. Heart J. 2012;33:606e13.