Erythropoietin treatment in patients with acute myocardial infarction: A meta-analysis of randomized controlled trials

Erythropoietin treatment in patients with acutemyocardialinfarction: Ameta-analysis of randomized controlled trialsDengfengGao, PhD, a NingNing,MMed, b XiaolinNiu,MMed, a YinhuDang,MMed, a XinDong,MMed, a JinWei,MD, a

and Canzhan Zhu, MDa Shaanxi, PR China

Background In experimental models of acute myocardial infarction (AMI), erythropoietin (EPO) reduces infarct size andimproves left ventricular (LV) function. However, in the clinical setting, the effect of EPO in AMI was unclear. We conducted asystematic review and meta-analysis of randomized controlled trials (RCTs) of EPO to explore the safety and therapeutic effectsof EPO in patients with AMI.

Methods We identified reports of RCTs comparing EPO to placebo for AMI in adult humans in PubMed, CochraneCentral Register of Controlled Trials, and EMBASE. Outcomes included all-cause mortality, major cardiovascular events,cardiac function by LV ejection fraction and infarct size.

Results We included 13 articles of RCTs with data for 1,564 patients. Erythropoietin therapy did not improve LV ejectionfraction (weighted mean difference [WMD] 0.33, 95% CI −1.90 to 1.24, P = .68) and had no effect on infarct size, asmeasured by cardiac magnetic resonance imaging (WMD −0.12, −2.16 to 1.91, P = .90) or serum peak value of creatinekinase-MB (WMD −2.01, −25.70 to 21.68, P = .87). Erythropoietin treatment did not decrease the risk of total adversecardiac events (relative risk [RR] 1.02, 0.65-1.61, P = .92). Erythropoietin treatment also failed to decrease the risk of heartfailure (RR, 0.69, 0.27-1.72, P = .42) and all-cause mortality (RR 0.55, 0.22-1.33, P = .18). Moreover, EPO had no effect onthe risk of stent thrombosis (RR, 0.69, 0.29-1.64, P = .40).

Conclusion Erythropoietin in patients with AMI seems to have no clinical benefit for heart function or reducing infarctsize, cardiovascular events, and all-cause mortality. Erythropoietin may not be a choice for patients with AMI. (Am Heart J2012;164:715-727.e1.)

Despite advances in pharmacologic and early revascu-larization therapies over the past 2 decades, acutemyocardial infarction (AMI) remains one of the leadingcauses of death and disability around the world.1,2 Thissituation might be due, in part, to lethal myocardialreperfusion injury, whereby reperfusion of the ischemicmyocardium induces further cardiomyocyte death,thereby limiting the benefits of myocardial reperfusiontherapy.3 Beyond the acute phase, adverse ventricularremodeling, heart failure, and mortality are directlyrelated to infarct size and left ventricular (LV) dysfunc-tion, which may cause disability and death in thesepatients.4,5 Given the global burden of ischemic heart

From the aDepartment of Cardiology, The Second Affiliated Hospital, Xi'an JiaotongUniversity School of Medicine, Xi'an, Shaanxi, PR China, and bDepartment of Nucleamedicine, The Second Affiliated Hospital, Xi'an Jiaotong University School of MedicineXi'an, Shaanxi, PR China.Submitted December 12, 2011; accepted July 27, 2012.Reprint requests: Dengfeng Gao, PhD, Department of Cardiology, The Second AffiliatedHospital, Xi'an Jiaotong University School of Medicine, Xi'an, Shaanxi 710004, PR China.E-mail: [email protected]/$ - see front matter© 2012, Mosby, Inc. All rights reserved.http://dx.doi.org/10.1016/j.ahj.2012.07.031

r,

disease and heart failure, novel cardioprotective agentsare needed to reduce lethal myocardial reperfusioninjury, limit myocardial infarct (MI) size, preservecardiac function, and improve clinical outcomes inthese patients.6,7

Erythropoietin (EPO) is a 165-amino-acid glycoproteinhormone secreted by the kidney in response to hypoxiaand has a pivotal role in regulating plasma hemoglobin(Hb) concentrations.8 Considerable evidence suggeststhat EPO has pleiotropic effects beyond that of hemato-poiesis, including cardioprotection.9 In animal models ofAMI and ischemia-reperfusion, a number of cardiopro-tective effects of EPO including attenuating ischemic-reperfusion injury,10-12 reducing infarct size,12 andimproving LV function13 have been reported. Furtherinvestigations demonstrated that EPO-mediated cardio-protection was mainly through reducing apoptosis,increasing neovascularization, mobilizing endothelialprogenitor cells, and inducing angiogenesis throughphosphorylation and activation of signaling pathwayssuch as JAK1/2, STAT3, STAT5, and PI3K9,14 (Figure 1).In recent years, some small randomized controlled trials

(RCTs) have investigated EPO in patients with AMI.15-25

These trials aimed to determine whether EPO has

mailto:[email protected]

http://dx.doi.org/10.1016/j.ahj.2012.07.031

Figure 1

Possible mechanisms of the beneficial effects of EPO on AMI. Administration of EPO leading to mobilization of EPCs (1), inhibition of inflammation(2), inhibition of apoptosis (3), induction of angiogenesis (4), and decrease of MI size (5). EPCs, endothelial progenitor cells; PI3K,phosphoinositide 3 kinase; Akt, protein kinase; STAT, signal transducers and activators of transcription; CASP, caspase; PKC, protein kinase C;ERK1/2, phospho-p44/42 MAPK.

716 Gao et alAmerican Heart Journal

November 2012

cardioprotective effects in a clinical setting similar to thatin animal models. Data from these trials were controver-sial. Some studies observed a benefit with EPO for AMI,and others failed to exhibit any benefit. Moreover, thetrials raised safety concerns with the drug and suggestedthat potential benefits may be mitigated by prothrombo-tic effects. However, the small sample sizes producedunderpowered results.We therefore performed a meta-analysis to explore the

safety and therapeutic effects of EPO in patients with AMI.

MethodsSearch strategyWe adhered to the Preferred Reporting Items for Systematic

Reviews and Meta-Analyses (PRISRMA) statement26 andCochrane Collaboration guidelines.27 Two authors (D.G. and

N.N.) independently searched for reports of RCTs in Medline viaPubMed (published from 1966 to March 2012), EMBASE(published from 1980 to March 2012), and the CochraneCentral Register of Controlled Trials (using the OVID interfacefor publications up to the first quarter of 2012). To identifystudies of EPO involving patients with AMI, we used both themedical subject heading terms and searched the text withthe terms “myocardial infarction” plus “erythropoietin,”“darbepoetin,” “EPO,” or “Epoetin.” Sensitive filters identifiedclinical research or RCTs in the Medline database28 and theEMBASE database.29 The search was limited to human subjects,with no restriction on language. We included studies andabstracts that were presented at recent congresses (AmericanCollege of Cardiology [2005-2010], American Heart Association[2005-2010], European Society of Cardiology [2005-2010]). TheClinical Trials database (http://clinicaltrials.gov/) was searchedfor articles in other archived registries. We also searched thebibliographies of reports of RCTs and review articles for otherreports of RCTs.30

http://clinicaltrials.gov/

Figure 2

Gao et al 717American Heart JournalVolume 164, Number 5

Study selectionAll abstracts were scanned independently by 2 investigators

(D.G. and N.N.), who then retrieved the full text of potentialarticles. Disagreements were resolved by consensus and, ifnecessary, with a third author (X.N.). We considered reports ofRCTs comparing EPOs to placebo for AMI in adults andreporting at least one of the outcomes of interest, including(1) all-cause mortality, (2) major cardiovascular events (death,reinfarction, stoke, stent thrombosis, emergency re–percutane-ous coronary intervention), (3) heart failure, (4) cardiac function(ie, LV ejection fraction [LVEF]), (5) ventricular remodeling (ie,LV end-diastolic volume index [LVDSV], LV end-systolic volumeindex [LVESV]), and (6) MI size (enzymatic infarction size orpercentage of the area of delayed hyperenhancement that wasseen on cardiac magnetic resonance imaging [MRI]).

Quality assessment and data abstractionData were abstracted by use of a data collection form by the 2

authors (D.G. and N.N.) and checked for accuracy (by X.N.).Data on study characteristics and patient characteristics werecollected. The quality of the included RCTs was assessed by theJadad scale.31 We also evaluated any possible bias by using theCochrane Collaboration's tool for assessing risk of bias.32 If astudy did not clearly mention one of the key points, weconsidered that the point was not covered in the study, and theresults may have underestimated the reported characteristics.

Flow diagram of the report selection process.

Statistical analysisWe performed a separate meta-analysis for each comparison

and outcome. For dichotomous outcomes, results wereexpressed as relative risk (RR) with 95% CIs. For continuousoutcomes, pooled data were described with the weighted meandifference (WMD) and 95% CIs. For continuous variablesreported as median and interquartile range, the mean and SDwere estimated. The mean was estimated by mean = (a + 2m +b)/4, where m is the median and a and b are quartiles 25 and75, respectively. We estimated the SD using the formula SD =interquartile range/1.35. Heterogeneity of results across trialswas assessed by a standard χ2 test with significance P b .10 andthe I2 statistic with significance I2 N 50%. Significance of thepooled estimates was tested with the fixed-effects model, orwith the random-effects model with significant heterogeneity inestimates. Publication bias was evaluated by the funnel plot,Begg adjusted rank correlation test, and Egger regressionasymmetry test. We assessed the impact of methodologicalcharacteristics of studies on our results by sensitivity analysis oftrials of major adverse cardiac events (MACEs) and LVEF endpoints. Meta-regression analyses were conducted to estimate theextent to which other covariates might have influenced thetreatment effects. All P values were 2 tailed, and the statisticalsignificance was set at .05. Analyses were performed usingReview Manager v 5.14 (Revman; The Cochrane Collaboration,Oxford, UK) and STATA software version 11 (STATA Corpora-tion, College Station, TX).

No extramural funding was used to support this work. Theauthors are solely responsible for the design and conduct of thisstudy, all study analyses, the drafting and editing of themanuscript, and its final contents.

ResultsThe search revealed 679 reports; 647 were excluded

after reading the abstract and a further 19 were excludedafter reading the full report. Reports for 13 RCTs16-25,33,34

enrolling 1,564 patients fulfilled all eligibility criteria.Figure 2 shows the selection of articles for the study.

Study characteristicsThe study and patient characteristics are shown in Tables

I and II, respectively. Six of the reports described

image of

Table I. Study characteristics

StudyRandomized patients

(EPO/placebo)Intervention(dose/time)

Follow-upinterval Inclusion criteria End points

Prunier et al34 110 (57/53) 1000 U/kgafter reperfusion

3 mo Patients with STEMIfor PCI

CMR imaging, LVEF, indexed LVESV,LVDSV, infarction size, any AE

Ludman et al33 51 (26/25) 50 000 IU beforePCI, and 24 h after)

4 mo Patients with STEMIfor PCI

CMR imaging, LVEF, indexed LVESV,LVDSV, infarct size, and AE

REVEAL19 222 (123/99) Single dose/60,000 U/within 4 h of reperfusion

12 wk STEMI underwentsuccessful PCI within8 h of onset

CMR imaging; LVEF, indexed LVESV,LVDSV; vital signs, Hb level, reticulocyte count,markers of cardiac injury, clinical events

Suh et al22 57 (29/28) Single dose 50U/kg/before reperfusion

6 mo STEMI underwentsuccessful PCI within12-h onset

CMR imaging, infarct size, clinical events

Ferrario et al16 30 (15/15) 33,000 U/3 times (duringPCI, 24 and 48 h after)


CMR imaging, CD34+ cell mobilization,hematometric, coagulation and blood pressure,clinical event, gene expression in PBC

HEBE III25 529 (266/263) Single dose 60,000U/after PCI

6 wk STEMI underwentsuccessful PCI

LVEF (by radionuclide ventriculography);infarct size; clinical events

EPOC-AMI24 35 (17/18) 6000 U/d immediately,2 and 4 d after PCI


Neointimal volume (by IVUS); LV fractionand infarct size (by radionulideventriculography); clinical events;blood pressure; blood cell count

EPO/AMI-121 41 (22/19) Single dose 12,000 Uwithin 24 h after PCI


LV fraction and infarct size(by radionulide ventriculography); LVEF,indexed LVESV, LVDSV; coronary re-stenosis,change in BNP; clinical events

REVIVAL-320 138 (68/70) 33,000 U/immediately,1 and 2 d after PCI


CMR imaging; LVEF; indexed LVESVand LVDSV; clinical outcome

Tang et al23 44 (29/15) 200 U/kg daily for 3 d 1 mo STEMI underwentsuccessful PCI within16-h onset

Platelet function; soluble fas ligand;ene expression; blood cell count;blood pressure; clinical events

Binbrek et al15 236 (115/121) Single dose 30,000 U/before thrombolysis

30 d STEMI underwentthrombolysis

Infarct size index; echocardiography;blood cell count; blood pressure; clinical events

Liem et al17 51 (26/25) Single dose 40,000 U 1 y NSTEMI Infarct size; blood pressure; clinical eventsLipsic et al18 20 (10/10) 60,000 U/before PCI 30 d Patients with STEMI

for PCILV function (by radionulide ventriculography);CD34+ cell mobilization; blood cell count;blood pressure; clinical events

STEMI, ST-segment elevated MI; NSTEMI, non–ST-segment elevated MI; CMR, cardiac magnetic resonance; IVUS, intravascular ultrasound; PCI, percutaneous coronary intervention;AE, adverse event; PBC, peripheral blood cells; BNP, brain natriuretic peptide.


November 2012

multicenter RCTs.19-21,24,25,34 Reports for 12 trials de-scribed major adverse events15-23,25,33,34; 8, all-causemortality15,19-22,25,33,34; 10, LVEF15,18-22,24,25,33,34; and 11,infarct sizes.15-23,25,33,34 Erythropoietin treatment wasdescribed as intravenous in all included reports. A singledose of EPOwas described in 8 reports15,17-19,21,22,25,34 andmultidoses in 5,16,20,23,24,33 with regimens ranging from 50IU/kg to a cumulative dose of 100,000 IU. The treatmentperiods were from before revascularization to 3 days afterrevascularization. The time point for evaluation of theprimary end point was from 30 days to 1 year.

Methodological qualityTable III summarizes the methodological quality of the

reported studies. The studies were generally of moderatequality. Only 4 studies met all the validity criteria. Thequality of 4 was limited by their single-blind design. Fourstudies were open label, and 2 did not report generationof random sequence. Appendix Figure 1 shows the bias

assessments according to the risk of bias. More than 75%of the studies met only 3 of the items, and only 3 studiesmet all requirements. Interrater reliability for assessingquality items was good (κ = 0.83, P b .001). Funnel plotsof reports describing MACEs and LVEF showed symmet-rical data, which indicated a low risk of publication bias(Figure 3A and B). Furthermore, no publication bias wasalso shown by Egger test (LVEF: P = .70 and MACEs: P =.31) and Begg test (LVEF [P = .42] and MACE [P = .17]).

Left ventricular function, ventricular remodeling, andinfarct sizeErythropoietin was not associated with a change in

cardiac function, LVEF, at follow-up (WMD 0.33, 95% CI,−1.90 to 1.24, P = .68) (Figure 4A), and LVEF did notdiffer at 30 days (WMD 1.16, −3.81 to 6.12, P = .65) or 6months (WMD 0.84, −2.12 to 3.79, P = .58) (Table IV).Left ventricular ejection fraction was not improvedwith EPO administered before (WMD −0.85, −4.58 to

Table II. Patient characteristics

Prunier et al34 Ludman et al33 REVEAL19 Suh et al22Ferrarioet al16 HEBE III25

Patients,n (EPO/placebo)

110 (57/53) 51 (26/25) 222 (123/99) 57 (29/28) 30 (15/15) 519 (266/253)

Age (y), mean (SD)(EPO/placebo)

57.6 (11.7)/57.3 (12.8) 55.5(12.8)/61 (10)

56.8 (12.4)/58.8 (12.5)

59.3 (13.7)/59.3 (8.8)

53.2 (9.3)/56.5 (10.8)

60.8 (10.9)/61.0 (11.3)

Female,n (EPO/placebo)

10/10 3/4 23/21 6/8 0/0 65/51

ComorbidconditionsHypertension, n 19/24 10/13 58/53 15/12 6/3 84/90Hyperlipidemia, n 29/30 7/13 50/43 5/8 7/9 53/45Diabetes, n 9/7 3/2 14/25 8/6 1/1 25/22Baselinemedication use

Clopidogrel 57/53 26/25 118 (95.9)/94 (94.9)

28/24 NA 266/253

Aspirin 57/53 26/25 110 (89.4)/93 (93.9)

29/28 0/1 266/253

Heparin 57/53 26/25 108 (87.8)/86 (86.9)

29/28 NA 266/253

Statin NA NA NA NA 1/1 NAGlycoproteinIIb/IIIa inhibitor

NA 25/23 88 (71.5)/71 (71.7) NA 199/212

Follow-up medicationClopidogrel NA 12/12 NA NA NA 209/216Aspirin NA 16/14 NA NA NA 226/242Statin NA 15/13 NA NA NA 235/243ACEI/ARB NA 17/13 NA NA NA 182/203β-Blocker NA 14/12 NA NA NA 225/235Heart rate(beats/min),mean (SD)

77.49 (14.7)/76.3 (12.7) 77.4(16.1)/76.5 (13.6)

77.4 (16.1)/76.5 (13.6)

NA 30/25 74.9 (15.5)/74.2 (16.0)

Time from symptomonset torevascularization

233 (184-312)/210(169-268)

224(104)/257 (156)

207.5 (106.8)/208.3 (105.7)

336 (163)/341 (291)

229 ± 72/216 ± 77

180 (126-288)/174 (120-251)

MACE(EPO/placebo)

Death (2/1); stent thrombosis(2/1); MI (2/1); stroke (0/0);

revascularization (2/3)

Death (0/1); stroke(0/0); unplannedrevascularization

(1/0)

Death (1/0); stroke(1/0); MI (2/0); stentthrombosis (3/0)

Death (0/1);stroke (1/0);MI (1/0)

Stentthrombosis

(1/0)

Death (1/2); stentthrombosis (2/7); stroke(1/1); revascularization

(3/2)

EPOC-AMI24 EPO/AMI-121 REVIVAL-320

Tanget al23

Binbreket al15

Liemet al17 Lipsic et al18

Patients,n (EPO/placebo)

35 (17/18)

41 (22/19) 138 (68/70) 44 (29/15) 236 (115/121)

51 (26/25) 20 (10/10)

Age (y), mean (SD)(EPO/placebo)

57.3(11.9)/61.2(10.5)

62.5 (8.0)　/59.4 (13.9)

59.1 (13.0)/62.1(12.3)

58 (12)/53(9)

48 (0.82)/48 (0.70)

NA 61.7 (2.9)/55.7(2.9)

Female,n (EPO/placebo)

1/2 0/0 12/18 8/5 12/12 NA 0/0

Comorbid conditionsHypertension 9/7 9/10 43/44 NA 37/40 NA 4/5Hyperlipidemia 7/10 4/8 24/25 NA 43/57 NA 3/2Diabetes 3/6 2/2 11/10 NA 31/31 NA 1/1

Baseline medicationuseClopidogrel 100%/

100%NA 68/70 29/15 100%/100% NA NA

Aspirin 100%/100%

NA 68/70 29/15 100%/100% NA NA

(continued on next page)


Table II (continued)

EPOC-AMI24 EPO/AMI-121 REVIVAL-320

Tanget al23

Binbreket al15

Liemet al17 Lipsic et al18

Heparin NA 68/70 68%/62% 21/13 NA NAStatin 82.3%/

83.0%NA NA 96%/93% 100%/99 % NA NA

GlycoproteinIIb/IIIainhibitor

NA NA NA 57/62 NA NA NA

Follow-up medication NAClopidogrel 100%/

100%22/19 68/70 29/15 NA NA NA

Aspirin 100%/100%

22/19 68/70 29/15 NA NA NA

Statin 82.3%/83%

19/14 NA 96%/93% NA NA NA

ACEI/ARB 81.2%/66.7%

20/14 NA 72/73 NA NA NA

β-Blocker 70.6%/83%

9/12 NA 100/93 NA NA NA

Heart rate (beats/min),mean (SD)

NA NA 76 (15)/76 (14) 69 ± 8/70± 8

NA NA NA

Time from symptomonset torevascularization

NA 216 (21)/171 (14)

168 (109-315)/175 (108-361)

300 (570)/330 (600)

226 (12)/200 (10)

NA NA

MACE (EPO/placebo) 0/0 Death (0/1);stent thrombosis

(0/1)

Death (2/3);stent thrombosis (3/2);MI (3/1); stroke (1/0);revascularization (7/2)

Stentthrombosis

(0/1)

Death (1/2);re-MI (2/2);stroke (1/0)

Death (0/0);re-MI (2/2)

Revascularization(2/2)

NA, Unable to assess; ACEI, angiotensin-converting enzyme inhibitors; ARB, angiotensin receptor blocker; n, number of patients.

Table III. Quality features assessed by Jadad score of 12 reports of RCTs assessed in the meta-analysis

Study Year DesignReporting of

randomizationGeneration of random

sequenceCompleteness of

follow-upDescription ofwithdrawals

Qualityscore

Prunier et al34 2012 PROBE Y Web based Y Y 4Ludman et al33 2011 DB-P Y Computer generated Y Y 5REVEAL19 2011 DB-P Y Web based Y Y 5Suh et al22 2011 Open Y NA Y NA 3Ferrario et al16 2011 DB-P Y Sealed envelope Y NA 4HEBE III25 2010 PROBE Y Computer generated Y Y 4EPOC-AMI24 2010 PROBE Y Computer generated Y NA 3EPO/AMI-121 2010 SB-P Y Independent statistician Y NA 3REVIVAL-320 2010 DB-P Y Sealed envelope Y Y 5Tang et al23 2009 DB-P Y IVRS Y Y 5Binbrek et al15 2009 SB-P Y Computer generated Y Y 4Liem et al17 2007 SB-P Y NA NA NA 2Lipsic et al18 2006 SB-P Y Computer generated Y NA 3

DB-P, Double-blind parallel; PROBE, prospective randomized open-label blinded end points; SB-P, single-blind parallel;NA, unable to assess; IVRS, interactive voice response system.


November 2012

2.87, P = .65) or after revascularization (WMD 1.00, −0.34to 2.34, P = .14), with a high-dose (≥30,000 unites) orlow-dose or single- and multidose administration (seeTable IV). Furthermore, EPO had no effect with percuta-neous coronary intervention therapy (WMD 0.74, −1.05to 2.53, P = .42). According to the results of meta-regression analyses, no statistically significant associationwas found between the benefits of EPO treatment andyear of publication (P = .26), study size (P = .95), timefrom symptom onset to revascularization (P = .54),

patients' sex (P = .47), and percentage of diabetes (P =.29). Nevertheless, we found a statistical significantassociation between patients' age and slight improvementof LVEF (P = .01) (Figure 5).For ventricular remodeling, EPO treatment did not

affect indexed LVESV (WMD 0.38, −2.68 to 3.44, P = .81)or LVDSV (WMD −0.50, −4.28 to 3.29, P = .80) (Figure 4Band C).Erythropoietin therapy did not reduce infarct size as

estimated by both cardiac MRI (6 trials; WMD −0.12,

Figure 3

Funnel plot of publication bias for reports evaluating the effect of EPO on LVEF and MACEs.


−2.16 to 1.91, P = .90) and serum peak creatinine kinaselevel (5 trials; WMD −2.01, −25.70 to 21.68, P = .87)(Figure 6A and B).

Major adverse eventsOverall, EPO treatment did not increase MACEs (RR

1.02, 0.65-1.61, P = .92) (Figure 7). It did not reduce therisk of 30-day (RR 1.08, 0.63-1.86, P = .77) or 6-month (RR1.53, 0.68-3.45, P = .30) cardiovascular events (Table IV).Sensitive analysis showed that EPO administration before(RR 1.23, 0.54-2.83, P = .62) or after revascularization (RR1.07, 0.66-1.74, P = .78) both did not reduce cardiovas-cular events (Table IV), respectively. Erythropoietin hadno effect with percutaneous coronary interventiontherapy (RR 1.09, 0.70-1.71, P = .70). High-dose EPO didnot have a protective effect (RR 1.17, 0.74-1.85, P = .50).Moreover, prolonged application (3 times) increasedcardiovascular events but not significantly (RR 1.79,0.71-4.53, P = .22). In addition, EPO administration hadno effect when considering the methodological quality oftrials (excluding those with Jaded scores ≤3) (RR 1.09,0.70-1.69, P = .70). The logarithm of RR for MACE againstpublication year (P = .88), follow-up time (P = .36), study(P = .32), Time from symptom onset to revascularization(P = .91), patients' age (P = .45), percentage of female (P =.84), and percentage of diabetic patients (P = .73) did notreveal any effect modification.For the 8 reports15,19-22,25,33 containing data on all-

cause mortality, mortality was slightly lower with EPOthan with placebo (0.85% [6/705 patients] vs 1.77% [12/679 patients]); for decreased mortality with EPO treat-ment, there was no significant difference between the 2groups (RR 0.55, 0.22-1.33, P = .18) (Figure 8A). For the 7reports containing data on heart failure, EPO treatmentdid not reduce heart failure events (RR 0.69, 0.27-1.72,

P = .42) (Figure 8B). For 7 reports containing data onstent thrombosis, EPO did not increased the risk of stentthrombosis (RR 0.69, 0.29-1.64, P = .40) (Figure 8C).

DiscussionsWe performed a meta-analysis of RCTs evaluating the

effect of EPO compared with placebo on cardiac functionand ventricular remodeling, MI size, and major adverseevents in patients with AMI after reperfusion therapy.Erythropoietin treatment did not increase cardiovascularevents, so EPO seems safe for patients with AMI.However, our results do not show any beneficial effecton LV function, ventricular remodeling, or infarct size.Moreover, EPO did not reduce cardiovascular events andall-cause mortality during follow-up.The hormone EPO is secreted by the kidney in

response to hypoxia and plays a pivotal role in regulatingplasma Hb concentrations. However, EPO and itsreceptor have been found in the kidney, as well asneuronal, vascular, and myocardial tissue. Recent evi-dence suggests that EPO has a protective role in vasculardiseases with nonerythropoietic effects such as reducingthe number of apoptotic cells and stimulating endothelialprogenitor cells, which enhances new vessel formation.These properties are of potential benefit in the contextof AMI.9,14,35

Growing evidence supports that EPO may improvecardiac function and reduce infarct size, thereby improv-ing clinical outcomes in patients with AMI. In animalexperimental studies, EPO acutely reduced infarct size byinhibiting apoptotic cell death, thereby preserving ventric-ular function.35 However, previous clinical studies yieldedconflicting results: 3 pilot studies found improved ventric-ular function,18,21,24 but other studies failed to show anybeneficial effect.19,20,25 By pooling these results, we found

image of

Figure 4

Forest plot of meta-analysis of WMD for LVEF (A), indexed LVESV (B), and LV end-diastolic volume (C) (LVEDV) for patients with AMI receivingEPO or placebo. IV, inverse variance.


November 2012

no difference between EPO and placebo in ventricularfunction as assessed by LVEF. We also performed asensitive analysis of follow-up interval, therapeutic time,and dosage but found no beneficial effect of EPO on LVEF.In addition, our study showed that patientswith higher agewere more likely to get benefits from EPO therapy. Thesesuggest that the difference in the proportion of patients'age might be origins of the interstudy discrepancy. For 5 ofour included studies, infarct size was indirectly estimatedfrom serial measurements of enzymatic markers of cardiacinjury.16-18,22,25 The HEBE III trial and another small pilotstudy reported reduced enzymatic infarct size by 6.7% (P =.06) and 30%, respectively.16,25 However, other trials

found no protective effect. By pooling these results, wefound no difference between EPO and placebo in infarctsize. In 5 trials19-22,34 that assessed infarct size withcontrast-enhanced cardiac MRI (a powerful, validatedtool that allows accurate and reproducible measurementof infarct size), EPO did not reduce infarct size.Therefore, EPO may have little benefit on ventricularfunction and infarct size.Several reasons may explain why the benefit observed

in animal experiments was not replicated in the clinicalsetting. First, the effects of EPOmay differ among species.Although EPO reduced infarct size in MI models such asmice,36 rats,37,38 dogs,39 and rabbits,40 it failed to show

image of

Table IV. Sensitivity and subgroup analyses

RCT(n)

MACE LVEF

P value forheterogeneity RR (95% CI)

P value foroveralleffect

RCT(n)

P value forheterogeneity WMD 95% CI

P value foroveralleffect

30 d 8 P = .66, I2 = 0% 1.08 (0.63, 1.86) P = .77 2 P = .009, I2 = 75% 1.16 (−3.81 to 6.12) P = .656 mo 4 P = .36, I2 = 0% 1.53 (0.68-3.45) P = .30 4 P = .63, I2 = 0% 0.84 (−2.12 to 3.79) P = .58High dose of EPO 9 P = .52, I2 = 0% 1.17 (0.74-1.85) P = .50 6 P = .02, I2 = 64% 0.29 (−2.14 to 2.73) P = .81Low dose of EPO 3 P = .43, I2 = 0% 0.84. (0.32-2.24) P = .73 4 P = .42, I2 = 0% −0.90 (−2.14 to 0.35) P = .16Single dose 8 P = .89, I2 = 0% 0.75 (0.45-1.27) P = .29 7 P = .06, I2 = 52% 0.73 (−1.04 to 2.50) P = .42Triple dose 4 P = .72, I2 = 0% 1.79 (0.71-4.53) P = .22 3 P = .63, I2 = 0% 0.61 (−2.76 to 3.98) P = .72Before revascularization 4 P = .93, I2 = 0% 1.23 (0.54-2.83) P = .62 4 P = .008, I2 = 75% −0.85 (−4.58 to 2.87) P = .65After revascularization 8 P = .26, I2 = 23% 1.07 (0.66-1.74) P = .78 6 P = .81, I2 = 0% 1.00 (−0.34 to 2.34) P = .14PCI 11 P = .58, I2 = 0% 1.09 (0.70-1.71) P = .70 9 P = .05, I2 = 53% 0.74 (−2.53 to 1.05) P = .42STEMI 11 P = .58, I2 = 0% 1.09 (0.70-1.69) P = .70 10 P = .04, I2 = 49% 0.33 (−1.90 to 1.24) P = .68Jadad scale ≥ 3 11 P = .70, I2 = 0% 1.09 (0.70-1.69) P = .58 10 P = .04, I2 = 49% 0.33 (−1.90 to 1.24) P = .68

PCI, Percutaneous coronary intervention; STEMI, ST-segment elevated MI.

Figure 5

Plot of univariate meta-regression examining the effect of patients'age on the relationship between EPO use and LVEF (P = .01). Thearea of each circle is inversely proportional to the variance of theestimate of the log risk ratio.


any benefit in other mammals such as sheep41 and pigs.42

In addition, the timing of EPO administration in clinicaltrials may be different from that in animal experiments.Most of the animal studies administrated a single dose ofEPO before or immediately after the start of ischemia andat the time of reperfusion led to powerful benefit effects.However, in a clinical setting, delay of EPO administrationwas much longer. For example, the mean delay betweenreperfusion and the single dose of 60,000-U EPO infusionexceeded 170 minutes in the REVEAL trial and was within3 hours in the HEBE III trial. Furthermore, the time fromsymptom onset to reperfusion between animal andclinical studies varied significantly.

Recently, concerns have been expressed regarding thesafety of EPO.43 Studies have shown a significantlyincreased risk of adverse events with EPO in noncardiacpopulations.43,44 In the REVEAL and REVIVAL-3 trials,EPO significantly increased the risk of death, recurrentMI, stroke, or stent thrombosis in patients with AMI,especially in old patients, but other studies did not detectan increased risk of adverse events with EPO, and theHEBE III trial even reported a low risk of major adversecardiovascular outcomes with EPO (mainly by reducingthe incidence of heart failure),25 which indicated thatEPO may have some cardioprotective effect. We pooledthese results and found that EPO was not associated withdecreased number of cardiovascular events and heartfailure. Furthermore, meta-regression result showed thatMACE was not associated with increased mean age ofeach study, which indicated that older patients might notbe at higher risk. Therefore, temporary withholding ofEPO for patients who are receiving this medication forlabeled indications and who experience MI meritsfurther study.In patients with AMI, the effect of EPO on mortality is

still controversial. Eight reports15,19-22,25,33,34 of RCTswith small samples described all-cause mortality. Thepooled results indicated that EPO failed to decrease therisk of all-cause mortality. Given the small sample sizes ofthe studies, our study was not powered for this effect,and this statement should be carefully interpreted.Hence, the overall effect of EPO on mortality in AMIshould be tested in large-scale, randomized controlledclinical trials.Our meta-analysis contains some limitations. First, the

methodological quality of the studies included in thisanalysis was less than optimal, with only 5 studiesmeeting all validity criteria, 4 studies being limited bytheir single-blind design, and 3 studies being open label,

image of

Figure 6

Forest plot of WMD for infarct size as assessed by cardiac MRI (CMR) and peak creatinine kinase level (EM).

Figure 7

Forest plot of RR for MACE. IV, inverse variance.


November 2012

including the largest trial, HEBE III. We were not able toexclude the potential risk of bias in these trials. Second,the numbers of events in the reported studies were lowerthan those that have been reported in other AMI

populations45 (for death, 1.5% in our study vs 2.7% to8% in other study). Hence, there may be a selected biaswhen enrolling the patients. In addition, heterogeneityshould be kept in mind when interpreting the results of

image of

image of

Figure 8

Forest plot of RR for all-cause death, heart failure, and stent thrombosis. IV, inverse variance.


meta-analyses. However, we found little heterogeneity inthe studied outcomes. Third, although we did notobserve any benefits of EPO (including cardiac function,infarct size, and hard end points), owing to the smallsample sizes, we may have underestimated the clinicalbenefit, if one does exist. Fourth, we did not have accessto individual patient data. Subgroups of AMI patientsmight show a difference with EPO treatment, but wecould not perform these analyses given the limited

number of patients in the available reports, especiallyfor major adverse events. Finally, the dosage, thetherapeutic window, and the follow-up times variedwidely among reports.In conclusion, based on available evidence, short-

term treatment with EPO in patients with AMI afterprimary reperfusion therapy seems to have no clinicalbenefit for improving heart function as assessed byLVEF, or for reducing infarct size, cardiovascular

image of


November 2012

events, and all-cause mortality. Erythropoietin may notbe a choice of treatment for patients with AMI afterreperfusion therapy.

AcknowledgementsThis study was supported by a grant from National

Natural Science Foundation of China (No. 30900617 toDengfeng Gao), Research Fund for the Doctoral Programof Higher Education of China (No. 2008.6981036 toDengfeng Gao), and Major Basic Research DevelopmentProgram of China from the Ministry of Science andTechnology (No. 2011CB503904 to Xiaolin Niu).

References1. Roger VL, Go AS, Lloyd-Jones DM, et al. Heart disease and stroke

statistics–2011 update: a report from the American Heart Associa-tion. Circulation 2011;123(4):e18-209.

2. Fox KA, Steg PG, Eagle KA, et al. Decline in rates of death and heartfailure in acute coronary syndromes, 1999-2006. JAMA 2007;297(17):1892-900.

3. Boden WE, Eagle K, Granger CB. Reperfusion strategies in acute ST-segment elevation myocardial infarction: a comprehensive review ofcontemporary management options. J Am Coll Cardiol 2007;50(10):917-29.

4. Esposito G, Dellegrottaglie S, Chiariello M. The extent of irreversiblemyocardial damage and the potential for left ventricular repair afterprimary percutaneous coronary intervention. Am Heart J 2010;160(6 Suppl):S4-10.

5. Gajarsa JJ, Kloner RA. Left ventricular remodeling in the post-infarction heart: a review of cellular, molecular mechanisms, andtherapeutic modalities. Heart Fail Rev 2011;16(1):13-21.

6. Kashani A, Giugliano RP. Management of ST-elevation myocardialinfarction: an update on pharmacoinvasive recanalization. Am JCardiovasc Drugs 2008;8(3):187-97.

7. Saia F, Grigioni F, Marzocchi A, et al. Management of acute leftventricular dysfunction after primary percutaneous coronary inter-vention for ST elevation acute myocardial infarction. Am Heart J2010;160(6 Suppl):S16-21.

8. Henry DH, Bowers P, Romano MT, et al. Epoetin alfa. Clinicalevolution of a pleiotropic cytokine. Arch Intern Med 2004;164(3):262-76.

9. Ruifrok WP, de Boer RA, Westenbrink BD, et al. Erythropoietin incardiac disease: new features of an old drug. Eur J Pharmacol 2008;585(2–3):270-7.

10. Hirata A, Minamino T, Asanuma H, et al. Erythropoietin enhancesneovascularization of ischemic myocardium and improves leftventricular dysfunction after myocardial infarction in dogs. J Am CollCardiol 2006;48(1):176-84.

11. van der Meer P, Lipsic E, Henning RH, et al. Erythropoietin improvesleft ventricular function and coronary flow in an experimental modelof ischemia-reperfusion injury. Eur J Heart Fail 2004;6(7):853-9.

12. Moon C, Krawczyk M, Ahn D, et al. Erythropoietin reducesmyocardial infarction and left ventricular functional decline aftercoronary artery ligation in rats. Proc Natl Acad Sci U S A 2003;100(20):11612-7.

13. van der Meer P, Lipsic E, Henning RH, et al. Erythropoietin inducesneovascularization and improves cardiac function in rats with heart

failure after myocardial infarction. J Am Coll Cardiol 2005;46(1):125-33.

14. Chateauvieux S, Grigorakaki C, Morceau F, et al. Erythropoietin,erythropoiesis and beyond. Biochem Pharmacol 2011;82(10):1291-303.

15. Binbrek AS, Rao NS, Al Khaja N, et al. Erythropoietin to augmentmyocardial salvage induced by coronary thrombolysis in patientswith ST segment elevation acute myocardial infarction. Am J Cardiol2009;104(8):1035-40.

16. Ferrario M, Arbustini E, Massa M, et al. High-dose erythropoietin inpatients with acute myocardial infarction: a pilot, randomised,placebo-controlled study. Int J Cardiol 2011;147(1):124-31.

17. Liem A, van de Woestijne AP, Bruijns E, et al. Effect of EPOadministration on myocardial infarct size in patients with non-STEacute coronary syndromes; results from a pilot study. Int J Cardiol2009;131(2):285-7.

18. Lipsic E, van der Meer P, Voors AA, et al. A single bolus of a long-acting erythropoietin analogue darbepoetin alfa in patients withacute myocardial infarction: a randomized feasibility and safetystudy. Cardiovasc Drugs Ther 2006;20(2):135-41.

19. Najjar SS, Rao SV, Melloni C, et al. Intravenous erythropoietinin patients with ST-segment elevation myocardial infarction: REVEAL:a randomized controlled trial. JAMA 2011;305(18):1863-72.

20. Ott I, Schulz S, Mehilli J, et al. Erythropoietin in patients with acute ST-segment elevation myocardial infarction undergoing primary percu-taneous coronary intervention: a randomized, double-blind trial. CircCardiovasc Interv 2010;3(5):408-13.

21. Ozawa T, Toba K, Suzuki H, et al. Single-dose intravenousadministration of recombinant human erythropoietin is apromising treatment for patients with acute myocardial infarction—randomized controlled pilot trial of EPO/AMI-1 study. Circ J 2010;74(7):1415-23.

22. Suh JW, Chung WY, Kim YS, et al. The effect of intravenousadministration of erythropoietin on the infarct size in primarypercutaneous coronary intervention. Int J Cardiol 2011;149(2):216-20.

23. Tang YD, Hasan F, Giordano FJ, et al. Effects of recombinant humanerythropoietin on platelet activation in acute myocardial infarction:results of a double-blind, placebo-controlled, randomized trial. AmHeart J 2009;158(6):941-7.

24. Taniguchi N, Nakamura T, Sawada T, et al. Erythropoietin preventiontrial of coronary restenosis and cardiac remodeling after ST-elevatedacute myocardial infarction (EPOC-AMI): a pilot, randomized,placebo-controlled study. Circ J 2010;74(11):2365-71.

25. Voors AA, Belonje AM, Zijlstra F, et al. A single dose of erythropoietinin ST-elevation myocardial infarction. Eur Heart J 2010;31(21):2593-600.

26. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items forsystematic reviews and meta-analyses: the PRISMA statement. BMJ2009;339:b2535.

27. van Tulder M, Furlan A, Bombardier C, et al. Updated methodguidelines for systematic reviews in the Cochrane Collaboration BackReview Group. Spine (Phila Pa 1976) 2003;28(12):1290-9.

28. Haynes RB, McKibbon KA, Wilczynski NL, et al. Optimal searchstrategies for retrieving scientifically strong studies of treatment fromMedline: analytical survey. BMJ 2005;330(7501):1179.

29. Wong SS, Wilczynski NL, Haynes RB. Developing optimal searchstrategies for detecting clinically sound treatment studies in EMBASE.J Med Libr Assoc 2006;94(1):41-7.

30. Hopewell S, McDonald S, Clarke M, et al. Grey literature in meta-analyses of randomized trials of health care interventions. CochraneDatabase Syst Rev 2007(2):MR000010.


31. Jadad AR, Moore RA, Carroll D, et al. Assessing the quality of reportsof randomized clinical trials: is blinding necessary? Control Clin Trials1996;17(1):1-12.

32. Higgins JP, Altman DG, Gotzsche PC, et al. The CochraneCollaboration's tool for assessing risk of bias in randomised trials.BMJ 2011;343:d5928.

33. Ludman AJ, Yellon DM, Hasleton J, et al. Effect of erythropoietin as anadjunct to primary percutaneous coronary intervention: a rando-mised controlled clinical trial. Heart 2011;97(19):1560-5.

34. Prunier F, Biere L, Gilard M, et al. Single high-dose erythropoietinadministration immediately after reperfusion in patients with ST-segment elevation myocardial infarction: results of the erythropoietinin myocardial infarction trial. Am Heart J 2012;163(2) 200–7e1.

35. Mastromarino V, Volpe M, Musumeci MB, et al. Erythropoietin andthe heart: facts and perspectives. Clin Sci (Lond) 2011;120(2):51-63.

36. Brunner S,Winogradow J, Huber BC, et al. Erythropoietin administrationafter myocardial infarction in mice attenuates ischemic cardiomyopathyassociated with enhanced homing of bone marrow-derived progenitorcells via the CXCR-4/SDF-1 axis. FASEB J 2009;23(2):351-61.

37. Ben-Dor I, Hardy B, Fuchs S, et al. Repeated low-dose oferythropoietin is associated with improved left ventricular function inrat acute myocardial infarction model. Cardiovasc Drugs Ther 2007;21(5):339-46.

38. Hale SL, Sesti C, Kloner RA. Administration of erythropoietin fails toimprove long-term healing or cardiac function after myocardialinfarction in the rat. J Cardiovasc Pharmacol 2005;46(2):211-5.

39. Hirata A, Minamino T, Asanuma H, et al. Erythropoietin just beforereperfusion reduces both lethal arrhythmias and infarct size via thephosphatidylinositol-3 kinase–dependent pathway in canine hearts.Cardiovasc Drugs Ther 2005;19(1):33-40.

40. Kobayashi H, Miura T, Ishida H, et al. Limitation of infarct size byerythropoietin is associated with translocation of Akt to themitochondria after reperfusion. Clin Exp Pharmacol Physiol 2008;35(7):812-9.

41. Olea FD, Vera Janavel G, De Lorenzi A, et al. High-doseerythropoietin has no long-term protective effects in sheep withreperfused myocardial infarction. J Cardiovasc Pharmacol 2006;47(6):736-41.

42. Kristensen J, Maeng M, Rehling M, et al. Lack of acute cardiopro-tective effect from preischaemic erythropoietin administration in aporcine coronary occlusion model. Clin Physiol Funct Imaging 2005;25(5):305-10.

43. Pfeffer MA, Burdmann EA, Chen CY, et al. A trial of darbepoetin alfain type 2 diabetes and chronic kidney disease. N Engl J Med 2009;361(21):2019-32.

44. Unger EF, Thompson AM, Blank MJ, et al. Erythropoiesis-stimulatingagents—time for a reevaluation. N Engl J Med 2010;362(3):189-92.

45. Widimsky P, Wijns W, Fajadet J, et al. Reperfusion therapy forST elevation acute myocardial infarction in Europe: description ofthe current situation in 30 countries. Eur Heart J 2010;31(8):943-57.

Gao et al 727.e1American Heart JournalVolume 164, Number 5

Appendix

Appendix Figure 1

Methodological quality of included studies according to the Cochrane Collaboration's tool for assessing risk of bias.32

Unlabelled image

Documents

Erythropoietin treatment in patients with acute myocardial infarction: A meta-analysis of randomized controlled trials