www.elsevier.com/locate/ejctsEuropean Journal of Cardio-thoracic Surgery 32 (2007) 481—487

Myocardial injury after off-pump coronary arterybypass grafting operation

Domenico Paparella *, Giangiuseppe Cappabianca, Piergiorgio Malvindi,Andreas Paramythiotis, Antonella Galeone, Nicola Veneziani,

Corrado Fondacone, Luigi de Luca Tupputi Schinosa

Division of Cardiac Surgery, Department of Emergency and Organ Transplant, University of Bari, Italy

Received 19 March 2007; received in revised form 29 May 2007; accepted 14 June 2007

Abstract

Objective: Perioperativemyocardial ischemia is less pronounced in off-pump coronary artery bypass (OPCAB) compared to on-pump coronaryartery bypass; however, the threshold over which the postoperative release of cardiac troponin I (cTnI) release and creatine kinase-MB (CK-MB)after OPCAB should be considered clinically relevant is unknown. The study was designated to evaluate if perioperative myocardial damage,measured by means of postoperative release of cTnI and CK-MB, has an influence on short- and mid-term outcome after OPCAB operations.Methods: Two hundred and sixty-one unselected patients undergoing OPCAB had cTnI and CK-MB measured preoperatively and nine timespostoperatively. Postoperative peak values were evaluated and the 80th percentiles were used to segregate the population into two groups foreach marker. The following cut-offs were used: 7.1 ng/dl for cTnI peak and 36.3 ng/dl for CK-MB peak. Results: Patients with cTnI>7.1 ng/ml(n = 51) and CK-MB >36.3 ng/ml (n = 48) had a longer mechanical ventilation and ICU length of stay. Nevertheless, hospital mortality did notdiffer between groups. Survival after 3 years was 92.8 � 2.3% and 81.8 � 6.2 for patients with postoperative cTnI peak �7.1 ng/ml and>7.1 ng/ml, respectively ( p = 0.003). It was 93 � 2.2% and 80 � 6.8% for patients with CK-MB �36.3 ng/ml and >36.3 ng/ml, respectively( p = 0.005). Adjusted hazard ratios for mid-termmortality were HR 2.7 (CI 1—7.6), p = 0.05 for cTnI>7.1 ng/dl and HR 3.1 (CI 1—9.1), p = 0.04for CK-MB >36.3 ng/ml. Conclusion: Perioperative myocardial damage should not be considered an innocuous event following OPCABoperations since the survival rate over 3 years is significantly worse in patients with the highest postoperative peak release of cTnI andCK-MB.# 2007 European Association for Cardio-Thoracic Surgery. Published by Elsevier B.V. All rights reserved.

Keywords: Coronary artery bypass grafting; CABG; Myocardial injury; Survival analysis; Off-pump

1. Background

Perioperative myocardial ischemia is a serious complica-tion, increasing mortality andmorbidity after cardiac surgery[1]. While postoperative electrocardiogram (ECG) variationshave a limited clinical value [2], several studies demon-strated the power of cardiac troponin I (cTnI) [3—6] andcreatine kinase-MB (CK-MB) [7—9] to predict hospitalmortality, major complications and graft failure afterstandard on-pump coronary artery bypass grafting (CABG).It is still debated whether perioperative myocardial ischemiamay influence survival in the mid-term: postoperative cTnIrelease provided conflicting results on this regard [5,10],while there is consensus about the prognostic role of

* Corresponding author. Address: Division of Cardiac Surgery, Dipartimentod’Emergenza e Trapianti d’Organo (D.E.T.O.), University of Bari, Piazza GiulioCesare 11, 70100 Bari, Italy. Tel.: +39 0805595075.

E-mail address: dpaparella@cardiochir.uniba.it (D. Paparella).

1010-7940/$ — see front matter # 2007 European Association for Cardio-Thoracicdoi:10.1016/j.ejcts.2007.06.015

postoperative CK-MB release on mid-term survival afteron-pump CABG [7—9].

Clinical experiences showed that the release of cTnIfollowing off-pump coronary artery bypass (OPCAB) issignificantly lower than after on-pump CABG [11]. In centerswith experience of beating heart operations, early results arein favor of this technique [12], with graft patency equivalentto the one obtained with cardiac arrest [13]. However, tworecent randomized controlled trials [14,15] revealed a worsepatency rate in OPCAB patients in comparison with patientsoperated on cardiopulmonary bypass. Moreover, data from alarge registry demonstrate that OPCAB patients experience ahigher rate of perioperative myocardial infarction and have asignificantly worse 3-year risk-adjusted survival [16].

No study evaluated before the threshold over which thepostoperative release of cTnI and CK-MB after OPCAB shouldbe considered clinically relevant. The aim of the presentstudy is to evaluate if perioperative myocardial damage,measured by means of postoperative release of cTnI and CK-MB, influences short- and mid-term outcome after OPCAB.

Surgery. Published by Elsevier B.V. All rights reserved.

D. Paparella et al. / European Journal of Cardio-thoracic Surgery 32 (2007) 481—487482

2. Material and methods

From January 2002 to December 2004, 261 consecutivepatients underwent OPCAB at our institution. In the sameperiod 642 patients underwent on-pump CABG. This numberincludes 16 patients converted to cardiopulmonary bypasssurgery during an OPCAB procedure; reasons for conversionwere: hemodynamic instability (nine patients), intramyo-cardial coronary arteries (four patients) and ECG changes(three patients). No strict decision criteria exist in ourinstitution to indicate OPCAB procedures; this is mainlydependent on surgeons’ (D.P., C.F., L.d.L.T.) choice. Gen-erally, high-risk patients, with several comorbidities, areoperated without cardiopulmonary bypass. All OPCABpatients (261) were enrolled in the study; they had serialmeasurements of cTnI and CK-MB at the following times:preoperatively at 6, 12, 24 and 36 h after the end of theoperations and every day from the second to the sixthpostoperative day. Trained personnel prospectively recordedpreoperative patients’ characteristics, intraoperative vari-ables and clinical outcomes in our institutional database.A local ethical committee approved the study protocol.Individual consent was waived.

The design of the study was a retrospective, single-centerobservation, and the primary end-point was to evaluate thepower of postoperative peaks of cTnI and CK-MB to predict 1-and 3-year survival after OPCAB. The secondary end-pointwas to evaluate the ability of these two markers to predicthospital morbidity and mortality.

With this aim, we tested four cut-offs for each marker:0.57 ng/ml (20th percentile), 1.25 ng/ml (40th percentile),2.45 ng/ml (60th percentile), 7.1 ng/ml (80th percentile) forcTnI; 5.3 ng/ml (20th percentile), 9.1 ng/ml (40th percen-tile), 14.9 ng/ml (60th percentile), 36.3 ng/ml (80th per-centile) for CK-MB. Each of these cut-offs was used in order todivide the whole cohort in two groups and also to calculatecrude odds ratios for hospital mortality and crude hazardratios for follow-up mortality (Table 1).

Hazard ratios for follow-up mortality progressivelyincreased across the percentiles. In particular, they becamesignificant up to the 80th percentile for both cTnI and CK-MB.

Table 1Crude odds ratios for hospital mortality, crude hazard ratios for mid-termmortality according to cTnI and CK-MB

Crude odds ratios for hospital mortality

cTnI CK-MB

OR (CI) p OR (CI) p

20th 0.7 (0.2—3.7) 0.72 2.0 (0.7—7) 0.2740th 1.5 (0.5—4.8) 0.48 0.6 (0.2—2.5) 0.6160th 1.3 (0.4—4.7) 0.63 0.9 (0.3—2.9) 0.8780th 1.0 (0.3—5.9) 0.79 1.1 (0.2—5.3) 0.87

Percentile Crude hazard ratios for mid-term mortality

cTnI CK-MB

HR (CI) p HR (CI) p

20th 1.7 (0.5—6) 0.39 1.3 (0.4—4.6) 0.6340th 1.4 (0.5—3.8) 0.50 1.5 (0.7—3.9) 0.4260th 2.1 (0.9—5.3) 0.10 2.2 (1—5.4) 0.0980th 3.6 (1.4—9) 0.006 3.5 (1.4—8.7) 0.008

A similar trend was not found analyzing odd ratios forhospital mortality. Therefore, the 80th percentile was usedfor both markers (7.1 ng/dl for cTnI peak and 36.3 ng/ml forCK-MB peak) to segregate our study population into twogroups. The study has a power of 0.8 to find a difference notless than 13% concerning mid-term mortality and a power of0.8 to find a difference not less than12% concerning hospitalmortality.

2.1. Operative technique

Normothermia was maintained during the operations bywarming the operative theatre, using a heating mattress andinfusing warm intravenous fluids. The heart was approachedthrough standard median sternotomy. Temporary pacemakerwires were placed in the right atrial appendage. A single 3/0Prolene suture, which had been passed previously through agauze swab, was placed in the posterior pericardium at themidline between the inferior vena cava and the left inferiorpulmonary vein. This retraction stitch enabled verticaldisplacement of the cardiac apex. Intravenous heparin(300 IU/kg) was administered prior to beginning the distalanastomoses with a target activated clotting time (ACT) of400 s. The distal anastomosis of left internal mammary arteryon left anterior descending was performed first. Proximalanastomoses of venous or free arterial grafts were madebefore performing distal anastomoses.

A tissue stabilizer (Octopus 3 Medtronic, Inc., Minneapo-lis, MN) was used in all patients, and intracoronary shunts(Clearview Medtronic, Inc., Minneapolis, MN) were used tosuture most of the distal anastomoses (77%), except for small(<1 mm) and very atheromatous coronary vessels. Distalanastomoses were suturedwith the aid of a CO2-water blower(Clearview Medtronic, Inc., Minneapolis, MN).

2.2. Definitions

Systemic hypertension was defined as blood pressureexceeding 140/90 mmHg, having a history of high bloodpressure or needing anti-hypertensive medications. Thosehaving a history of diabetes, regardless of the duration ofdisease or the need of anti-diabetic agents, were considereddiabetic. Hypercholesterolemia was defined as a fastingcholesterol level>200 mg/dl. Smoking history was defined asany current or past form of tobacco use. Chronic obstructivepulmonary disease (COPD) was defined as forced expiratoryvolume during the first second (FEV1) <75% or the need ofpharmacological therapy for the treatment of chronicpulmonary compromise. On-going refractory angina thatrequired the use of intravenous nitrate therapy for controlwas regarded as unstable angina. Left ventricular ejectionfraction was obtained in all patients by planimetry of leftventriculogram. EuroScore was calculated by additivemethod.

Death within the same hospital admission, regardless ofthe cause, was defined as in-hospital mortality. Low outputsyndromewas defined as the need for postoperative inotropicsupport or an intra-aortic balloon pump for more than 30 minto maintain systolic blood pressure greater than 90 mmHg,mean blood pressure greater than 60 mmHg or the cardiacindex greater than 2.2 l/min/m2, despite sufficient volume

D. Paparella et al. / European Journal of Cardio-thoracic Surgery 32 (2007) 481—487 483

Fig. 1. Box plots of cTnI and CK-MB across times before and after OPCAB.

substitution. Extubation criteria were hemodynamic stabi-lity, absence of surgical bleeding, fully re-warming, con-sciousness and optimal blood gases with FIO2 �0.3.Cerebrovascular disease (CVD) was regarded as any transientischemia accident, reversible ischemic neurological deficit orstroke.

2.3. Biochemistry

Blood samples for cTnI and CK-MB mass were collectedpreoperatively, at 6, 12, 24 and 36 h after the end of theoperations and every day from the second to the sixthpostoperative day. Cardiac troponin I and CK-MB mass weremeasured according to the manufacturer’s recommendationby standard immunoassay techniques (Dade Behring, Inc.,Newark, DE).

2.4. Electrocardiogram

Daily 12-lead ECGs were performed in all patients beforeand after the operations and, after the evaluation ofcardiologist blinded to the clinical outcome of the patient,presence and the localization of new q-waves were routinelyrecorded in the institutional database.

2.5. Follow-up

After discharge home, follow-up was carried out byperiodic evaluation of patients at our institution or contact-ing patients’ general practitioner by telephone call. Max-imum, minimum and median lengths of follow-up were 50, 12and 30 (interquartile range 22—41) months, whereascompleteness was 94.8% (238/251).

2.6. Statistical analysis

All continuous variables were evaluated by Shaphiro-Wilktest for normality: if non-normal, they were presented asmedian and interquartile range (IQR: 25th—75th percentile)and were compared by Mann—Whitney test. Instead, ifnormally distributed, they were presented as mean � stan-standard deviation and compared by t-test. Categoricalvariables were presented as an absolute number andpercentage and were compared through Fisher’s exact test.Spearman rank correlation coefficient evaluated correlationbetween the peaks of cTnI and CK-MB. ROC curves were usedto calculate areas under the curve (AUC) of postoperativecTnI and CK-MB peaks in function of in-hospital mortality. Asno patient was censored during the first 12 months of follow-up it was also possible to draw receiver operatingcharacteristic (ROC) curves and calculate AUC for 1-yearmortality. Logistic regression was used to calculate crudeodds ratios for hospital mortality (Table 1). Cox hazard modelwas initially used to calculate crude hazard ratios for follow-up mortality according to different cut-offs of cTnI and CK-MB (Table 1). A stepwise model (Table 4) for follow-upmortality containing all preoperative, intraoperative andpostoperative variables was created. Variables whichreached p � 0.20 at the univariable step were introducedin a multivariable model in order to adjust crude hazardratios of cTnI > 7.1 ng/ml and CK-MB > 36.3 ng/ml for

follow-up mortality. Kaplan and Meier curves and log ranktest were used to estimate 1- and 3-year survival prob-abilities and patients at risk. For all the tests, p-values wereconsidered significant when p � 0.05. Statistical analyseswere performed using the Stat-View Statistical SoftwarePackage 5.0 (SAS Institute, Inc. Cary, NC), NCSS 2001 (NumberChruncher Statistical System, Kaysville, Utah).

3. Results

Fig. 1 shows median values of cTnI and CK-MB acrossdifferent times before and after OPCAB. Trends of the twoenzymes were quite similar but CK-MB peak (24 h) wasreached earlier than cTnI (36 h). Moreover, there was asignificant correlation between cTnI and CK-MB peaks(Spearman’s rank test coefficient 0.71, CI 0.64—0.76,p < 0.0001).

During the operations, 704 distal anastomoses wereperformed (mean 2.7 � 1.1), and of these 546 (77%) wereperformed using shunts (mean 2.1 � 1.2) and 270 (38%) werearterial grafts (mean 1 � 0.3).

Overall in-hospital mortality for the study population was3.8% (10/261), whereas the mean EuroScore was 5 (IQR 3—7).Therefore, the ratio between observed and expectedmortality was 0.76. Causes of in-hospital death were: lowcardiac output syndrome (six), arrhythmia (three) andmediastinitis (one). In our population no patient experiencedpostoperative cerebrovascular accident after OPCAB. One-year survival was 95.2 � 4.8% (228 patients at risk), whereas3-year survival was 89.5 � 10.5% (80 patients at risk).Nineteen patients died during follow-up and the causes ofdeath were: acute myocardial infarction (three), heartfailure (six), sudden death (six), renal failure (two), stroke(one) and cancer (one).

3.1. Cardiac TnI

Preoperative and intraoperative characteristics of patientswith cTnI postoperative peak >7.1 ng/ml after OPCAB werequite similar to the ones of patients with cTnI �7.1 ng/ml(Table 2). Minor differences were found regarding the

D. Paparella et al. / European Journal of Cardio-thoracic Surgery 32 (2007) 481—487484

Table 2Preoperative and intraoperative data according to cTnI and CK-MB

cTnI CK-MB

�7.1 ng/ml >7.1 ng/ml p �36.3 ng/ml >36.3 ng/ml p

Patients 210 51 213 48Age > 70 years 74 (35.2%) 19 (37.3%) 0.87 74 (34.7%) 19 (39.6%) 0.62Male 164 (78%) 36 (71%) 0.27 165 (77%) 35 (73%) 0.57BMI > 30 kg/m2 32 (15.2%) 9 (17.6%) 0.67 33 (15.5%) 8 (16.7%) 0.82Hypertension 149 (71%) 38 (75%) 0.72 152 (71%) 35 (73%) 0.99Diabetes 84 (40%) 24 (47%) 0.42 86 (40%) 22 (46%) 0.51Insulin therapy 37 (18%) 15 (29%) 0.08 37 (17%) 15 (31%) 0.04Hypercholesterolemia 126 (60%) 29 (57%) 0.75 126 (59%) 29 (60%) 0.99Statin therapy 99 (47%) 18 (35%) 0.15 98 (46%) 19 (40%) 0.52Smoke 124 (59%) 26 (51%) 0.34 126 (59%) 24 (50%) 0.26COPD 71 (34%) 15 (29%) 0.62 68 (32%) 18 (38%) 0.5AMI < 21 days 60 (29%) 22 (43%) 0.06 65 (31%) 17 (35%) 0.5AMI > 21 days 103 (49%) 19 (37%) 0.15 102 (48%) 20 (42%) 0.52Preop. TnI > 75th percentile (0.3 ng/ml) 42 (20%) 17 (33%) 0.06 47 (22.1%) 12 (25%) 0.7Preop. CK-MB > 75th percentile (1.2 ng/ml) 47 (22.4%) 14 (27.4%) 0.46 48 (22.5%) 13 (27.1%) 0.57Unstable angina 37 (18%) 12 (24%) 0.32 39 (18%) 10 (21%) 0.68EF � 40% 68 (32.4%) 17 (33.3%) 0.99 67 (31.5%) 18 (37.5%) 0.49Main stem > 50% 21 (10%) 8 (16%) 0.32 22 (10%) 7 (15%) 0.44C.A. stenosis > 60% 16 (8%) 8 (16%) 0.1 16 (8%) 8 (17%) 0.06Previous CVA 27 (13%) 3 (6%) 0.22 26 (12%) 4 (8%) 0.61PVD 29 (14%) 9 (18%) 0.5 26 (12%) 10 (21%) 0.17CRF 13 (6%) 6 (12%) 0.22 13 (6%) 6 (13%) 0.13EuroScore 4 (3—6) 5 (4—7) 0.21 4 (3—6) 5 (4—7) 0.1Grafts/patient 3 (2—4) 3 (2—3.5) 0.89 3 (2—4) 3 (2—4) 0.25Grafts � 3 110 (52.4%) 28 (54.9%) 0.75 109 (51.2%) 29 (60.4%) 0.26Arterial grafts/patients 1 (1—1) 1 (1—1) 0.43 1 (1—1) 1 (1—1) 0.61Arterial graft > 1 14 (6.7%) 0 (0%) 0.08 14 (6.6%) 0 (0%) 0.08Coronaries shunted/patient 2 (1—3) 2 (2—3) 0.37 2 (1—3) 2 (2—3) 0.32Coronaries shunted > 2 78 (37%) 23 (45%) 0.34 79 (37%) 22 (46%) 0.32

Abbreviations: AMI, acute myocardial infarction; BMI, body mass index; C.A., carotid arteries; COPD, chronic obstructive pulmonary disease; CRF, chronic renalfailure; CVA, cerebrovascular accident; EF, ejection fraction; PVD, peripheral vascular disease.

incidence of diabetic patients undergoing insulin therapybefore the operation (p = 0.08), incidence of patients withacute myocardial infarction less than 21 days (p = 0.06),incidence of patients arriving to the operation with apreoperative cTnI over the 75th percentile (p = 0.06) andincidence of patients who received more than one arterialgraft during the operation (p = 0.08). However, neither thenumbers of graft nor intracoronaric shunts usedwere differentbetween these two groups (Table 3).

Table 3Postoperative results according to postoperative cTnI and CK-MB peaks

cTnI

�7.1 ng/ml >7.1 ng/m

Patients 210 51Hospital mortality 8 (3.8%) 2 (3.9%)Mechanical vent (h) 11.5 (9.5—17.6) 13 (11—20Mechanical vent >75th percentile (18 h) 49 (23%) 16 (31%)ICU stay (h) 41 (22—44) 42 (24—46ICU >75th percentile (44 h) 40 (19%) 18 (35%)Overall stay (days) 7 (6—7.7) 7 (6.5—8)Overall stay >75th percentile (8 days) 30 (14%) 10 (20%)LCOS 16 (8%) 8 (16%)New q-wave 25 (12%) 14 (27%)AF 49 (23%) 14 (27%)ARF 18 (9%) 8 (16%)Blood loss 720 (550—920) 800 (615—Blood loss 75th percentile (945 ml) 52 (25%) 13 (25%)Blood products usage 90 (43%) 29 (57%)

Abbreviations: ICU, intensive care unit; LCOS, low cardiac output syndrome; AF, atr

In the comparison of the two groups, hospital mortalityresults were found to be very similar. In fact, AUCs forhospital mortality, according to cTnI peak value, was closeto 0.5 (0.53 � 0.1). Significant differences were observedin regards of the duration of mechanical ventilation,ICU stay and incidence of new q-waves after the operation.One- and 3-year survivals of patients with postoperativecTnI peak >7.1 ng/ml were 87.2 � 5.4 (39 at risk) and81.8 � 6.2 (10 at risk), respectively, versus 97.1 � 1.3%

CK-MB

l p �36.3 ng/ml >36.3 ng/ml p

213 480.97 8 (3.8%) 2 (4.2%) 0.89

) 0.01 12 (9.5—17.7) 14 (12—21) 0.020.28 50 (23%) 15 (31%) 0.27

) 0.02 41 (22—44) 42 (37—47) 0.010.03 40 (19%) 18 (38%) 0.010.21 7 (6—7) 7 (6—8) 0.180.39 30 (14%) 10 (21%) 0.270.1 17 (8%) 7 (15%) 0.160.01 26 (12%) 13 (27%) 0.010.59 51 (24%) 12 (25%) 0.850.19 18 (8%) 8 (17%) 0.1

955) 0.16 730 (560—970) 700 (547—836) 0.30.99 58 (27%) 7 (15%) 0.090.09 96 (45%) 23 (48%) 0.75

ial fibrillation; ARF, acute renal failure.

D. Paparella et al. / European Journal of Cardio-thoracic Surgery 32 (2007) 481—487 485

(189 at risk) and 92.8 � 2.3% (70 at risk) for patientswith cTnI peak �7.1 ng/ml (log rank test p = 0.003).Introducing the multivariable hazard model for follow-upmortality (Table 4) the crude hazard ratio previouslycalculated for cTnI postoperative peak >7.1 ng/ml(HR = 3.6, CI 1.4—9, p = 0.006) was adjusted to HR 2.7,CI 1—7.6, p = 0.05. Moreover, AUC obtained drawingROC curve for 1-year mortality according to cTnI post-operative peak was only 0.68 � 0.1, suggesting that onlyvery high release of cTnI (>7.1 ng/dl) affects survival afterOPCAB.

Preoperative clinical state was found to be similarbetween patients with postoperative CK-MB peak greaterthan or �36.3 ng/dl (Table 2); only insulin-dependentdiabetes was significantly more frequent in the >36.3 ng/dlCK-MB postoperative peak. The number of grafts, thenumber of arterial grafts or the number of intracoronaric

Table 4Cox hazard model for follow-up mortality

Variables Univariable analysis forfollow-up mortality

HR p

PreoperativeAge >70 2.5 (1—6.5) 0.03Female 0.4 (0.2—2) 0.21BMI > 30 0.8 (0.3—3.2) 0.64Hypertension 1.6 (0.5—4.9) 0.37Diabetes 1.2 (0.5—2.9) 0.67Insulin therapy 1.3 (0.5—3.7) 0.55Hypercholesterolemia 0.7 (0.3—1.7) 0.37Statin therapy 0.5 (0.2—1.4) 0.13Smoke 2 (0.6—4) 0.15COPD 2 (0.7—4.1) 0.07AMI < 21 days 1.2 (0.4—2.7) 0.67AMI > 21 days 0.7 (0.3—1.7) 0.48Preop. TnI > 75th percentile 1 (0.5—4.3) 0.89Preop. CK-MB > 75th percentile 1.3 (0.7—4.8) 0.55Unstable angina 1.7 (0.4—3.8) 0.27EF � 40% 2.7 (1.3—8.3) 0.02Main stem >50% 2.1 (0.9—7.9) 0.17C.A. stenosis >60% 1.7 (0.7—6.7) 0.3CVA 0.7 (0.2—3.7) 0.61PVD 2.9 (1.1—8) 0.02CRF 1.7 (0.6—7.3) 0.37EuroScore > 5 5 (1.9—15.2) 0.003

IntraoperativeGrafts � 3 0.6 (0.3—1.9) 0.32Arterial graft >1 0.6 (0.1—5.5) 0.61Coronaries shunted >2 0.8 (0.3—2) 0.61

PostoperativeMechan. ventilation >75th percentile 2.3 (1—6.4) 0.06ICU >75th percentile 2.4 (1.1—6.6) 0.05Overall stay >75th percentile 5.5 (2.4—14.9) 0.0002TnI peak >7.1 ng/dl 3.6 (1.4—9) 0.006CK-MB peak >36.3 ng/dl 3.5 (1.4—8.7) 0.008LCOS 2 (0.5—9.7) 0.34New q-wave 0.9 (0.3—3.3) 0.87AF 0.9 (0.3—2.9) 0.83ARF 2.5 (0.9—7.9) 0.11Blood loss 75th percentile 1.7 (0.7—4.7) 0.26Blood products usage 2.7 (1.2—9) 0.04

Abbreviations: AF, atrial fibrillation; AMI, acute myocardial infarction; ARF, acuteobstructive pulmonary disease; CRF, chronic renal failure; CVA, cerebrovascular accisyndrome; PVD, peripheral vascular disease.

shunts used was not associated with higher release ofCK-MB.

Hospital mortality was similar between the groups. Infact, AUC for hospital mortality was 0.48 � 0.1. Patients withpostoperative CK-MB peak >36.3 ng/dl experienced longermechanical ventilation, ICU stay and had a higher incidenceof new q-waves after the operation. One- and 3-year survivalsfor patients with CK-MB postoperative peak >36.3 ng/mlwere 89.2 � 5.1% (38 at risk) and 80 � 6.8% (14 at risk),respectively, versus 96.5 � 1.4% (190 at risk) and 93 � 2.2%(66 at risk) for patients with CK-MB �36.3 ng/ml (log ranktest p = 0.005). Introducing in the multivariable hazardmodel (Table 4) the crude hazard ratio of CK-MB peak>36.3 ng/ml on mid-term mortality (HR 3.5, CI = 1.4—8.7,p = 0.008), it was adjusted to HR 3.1, CI 1—9.1, p = 0.04. AUCof 1-year mortality according to CK-MB peak value was0.66 � 0.1.

Multivariable Analyses for follow-up mortality

Including TnI peak >7.1 ng/dl Including CK-MB peak>36.3 ng/dl

HR p HR p

0.8 (0.3—2.6) 0.78 0.7 (0.2—2.1) 0.48

0.5 (0.2—1.4) 0.17 0.4 (0.1—1.2) 0.14.5 (1.3—15.2) 0.02 4.2 (1.3—14.2) 0.020.6 (0.2—1.9) 0.35 0.6 (0.2—1.8) 0.37

2.6 (0.9—7.3) 0.08 2.6 (0.9—7.6) 0.071.5 (0.4—5) 0.52 1.4 (0.4—4.8) 0.59

1 (0.3—3.5) 0.99 1.1 (0.3—3.6) 0.94

3.2 (0.7—14.7) 0.14 2.9 (0.7—12) 0.15

1.5 (0.6—4.1) 0.42 1.6 (0.6—4.6) 0.340.9 (0.3—3) 0.9 1 (0.3—3.1) 0.983.9 (1.1—13.3) 0.03 4.1 (1.2—13.6) 0.022.7 (1—7.6) 0.05 — —— — 3.1 (1—9.1) 0.04

0.7 (0.2—3.1) 0.68 0.6 (0.1—2.7) 0.47

2.7 (0.9—8.2) 0.09 3.2 (1—10.3) 0.06

renal failure; BMI, body mass index; C.A., carotid arteries; COPD, chronicdent; EF, ejection fraction; ICU, intensive care unit; LCOS, low cardiac output

D. Paparella et al. / European Journal of Cardio-thoracic Surgery 32 (2007) 481—487486

3.2. Onset of new q-wave

The number of patients who experienced the onset of anew q-wave was 39 (14.9%). Hospital mortality for thesepatients was not different compared to the rest of thepopulation (5.1% vs 3.6%, p = 0.64), while the incidence ofpostoperative low cardiac output was significantly higher(25% vs 6.3%, p = 0.0001). One- and 3-year survivals forpatient with onset of new q-waves were 97.1 � 2.8 (39 atrisk) and 86.3 � 10.5 (10 at risk), respectively, whereaspatients without new q-waves after the operation had thefollowing survivals: 94.8 � 1.7% at 1 year (189 patients atrisk) and 89.4 � 2.6 at 3 years (70 at risk) (log rank testp = 0.87).

4. Discussion

The present study confirms that, during OPCAB, a smalldegree of myocardial damage occurs, as demonstrated by therelease of cTnI and CK-MB. The postoperative peak results ofthese two markers were strictly correlated. In our popula-tion, observed hospital mortality appears high (3.8%) whencompared to previously published series [17], but it is in linewith the expected mortality (5%) calculated by EuroScore.This result reflects the choice of our institution to addresshigh-risk patients (older age, presence of left ventriculardysfunction, chronic renal failure, history of cerebrovascularaccidents), when possible, to OPCAB.

In our study, the degree of postoperative cTnI and CK-MBdoes not predict in-hospital mortality and incidence of lowcardiac output, but we found that ICU length of stay andduration of mechanical ventilation were significantly longerin patients with cTnI and CK-MB over the 80th percentile.Moreover, postoperative electrocardiograms revealed amorefrequent onset of new q-waves in patients belonging to the80th percentile.

Our study confirms that ECG changes observed in theimmediate postoperative period do not possess a clinicalrelevance and the ability to predict postoperative outcome[2]; unfortunately we are not able to document furtherevolution of the ECG in the following months.

We first demonstrate that very high peaks of cardiacenzyme release after OPCAB is associated with a worsemid-term survival. In fact, considering the 80th percentile asthe cut-off for cTnI and CK-MB, both crude and adjustedhazard ratios are significant predictors of death during thefollow-up.

Similarly, in patients undergoing percutaneous coronaryintervention (PCI) an elevation in cardiac enzymes has beenshown to predict outcome. Kini et al. [18] found on 2873patients who had undergone successful PCI that post-procedural CK-MB greater than five times the normal levelis an independent risk factor (OR 6.7, p = 0.002) for 1-yearmortality. Ricciardi et al. [19] found on 286 patients who hadundergone PCI that a threefold elevation of cTnI aftersuccessful PCI independently predicts major adverse cardiacevents (MACE), especially the need for early repeatedrevascularization, 1 year after the procedure. To explain therelationship between post-procedural cardiac markers ele-vation and adverse follow-up after PCI, Nageh et al. suggest

[20] that myocardial damage due to procedural complica-tions and the related tissue remodeling and scar formationmay form a substrate for an adverse outcome secondary toreduced left ventricular function and arrhythmic events.

We can hypothesize that, in a similar way, postoperativerelease of cardiac enzymes may influence mid-term outcomebecause of increased myocardial damage. Using magneticresonance imaging (MRI), Steuer et al. [21] showed thatelevated cTnI, CK-MB and troponin T after on-pump CABGcorrespond to the amount of perioperatively infarctedmyocardium. The correlation between the release of cardiacenzymes and the loss of vital myocardium could probablyexplain the excess of almost 20% on 4-year survival that weobserved after OPCAB in patients belonging to the 80thpercentile of cTnI and CK-MB peak release.

Several factors can influence the release of cardiacenzymes during a surgical myocardial revascularization: inon-pump CABG, the length of cardioplegic arrest, thedelivery and the composition of cardioplegic solutions;during OPCAB, the number of grafts done, the time requiredto complete each anastomosis and the use of intracoronaricshunts [22]. Recently, Rastan et al. [23] demonstrated thatmyocardial revascularization carried out with cardiopulmon-ary bypass without cardioplegic arrest induces a greaterrelease of cardiac enzymes compared to OPCAB. However,independent of the techniques used, we think that graftfailure and incomplete revascularization are the mostprobable causes of postoperative release of cardiac enzymes:microembolism, graft thrombosis, spasm, kinking or over-stretching of the conduits, stenosis or occlusion of theanastomotic sites can be the cause of acute graft failureeither after OPCAB or on-pump CABG. A recent work byThielmann et al. [6] on 55 patients with perioperativemyocardial infarction over 2078 patients who had undergoneon-pump CABG evidenced, after early angiographic control,that 63% had graft failure. Patients with graft-relatedperioperative myocardial infraction had similar hospitalmortality, incidence of major complications, onset of newq-wave, but a higher peak of catnip compared with patientswith no graft-related perioperative myocardial infraction.Rasmussen et al. [24] demonstrated graft failure in 43 out of59 patients (73%) who received acute postoperative re-angiography because of a highly suspected perioperative MI.Unfortunately a similar evaluation is not yet available forOPCAB patients.

According to several surgeons’ opinion other mechanismscan provoke cardiac enzymes release, such as surgicaldissection required to expose intramyocardial coronaryarteries, heart manipulation and epicardium suction byheart stabilizer employed during OPCAB. However, there isno scientific evidence to support these hypotheses.

In our experience, the predictive role of cTnI after on-pump CABG appears quite different from OPCAB. A recentanalysis by our institution [5] on 230 patients who hadundergone on-pump CABG evidenced that a postoperativepeak release of cTnI > 13 ng/ml increases 10-fold the risk ofin-hospital mortality, is associated with higher incidence ofshort-term complications, but does not predict mid-termsurvival. The adoption of 13 ng/dl as cut-off of cTnI wassupported by previous publications and confirmed by ROCcurves. After OPCAB, the degree of myocardial damage, also

D. Paparella et al. / European Journal of Cardio-thoracic Surgery 32 (2007) 481—487 487

considering very high cTnI and CK-MB peak release, is notenough to impair mortality. However, even a limited release(compared to on-pump CABG) of cardiac enzyme after theOPCAB may suggest an imperfect revascularization (graftfailure or incomplete revascularization) and this may explainthe worse survival during follow-up.

Major limitations of our study are the retrospectivedesign, the consequent lack of angiographic evaluation ofgraft patency and the absence of serial echocardiographic orMRI evaluation of left ventricular function to possiblycorrelate myocardial function to survival.

In conclusion, a very high release of cTnI and CK-MBsignificantly affects mid-term survival, but it appears lesseffective in predicting short-term outcome after OPCAB.Along with previous work, the onset of new q-waves seemsless useful predicting short- and mid-term outcome afterOPCAB. The major cause of very high cardiac enzyme releaseafter OPCAB could be graft failure and incomplete myocar-dial revascularization. In order to confirm this hypothesis andimprove mid-term results, future studies with angiographiccontrols should be performed in patients with high cTnI andCK-MB postoperative peaks, despite an acceptable short-term outcome.

References

[1] Multicenter Study of Perioperative Ischemia (MCSPI) Research Group.Effects of acadesine on the incidence of myocardial infarction andadverse cardiac outcomes after coronary artery bypass graft surgery.Anesthesiology 1995;83:658—73.

[2] Crescenzi G, Bove T, Pappalardo F, Scandroglio AM, Landoni G, Aletti G,Zangrillo A, Alfieri O. Clinical significance of a new Q wave after cardiacsurgery. Eur J Cardiothorac Surg 2004;25:1001—5.

[3] Eigel P, van Ingen G, Wagenpfeil S. Predictive value of perioperativecardiac troponin I for adverse outcome in coronary artery bypass surgery.Eur J Cardiothorac Surg 2001;20:544—9.

[4] Lasocki S, Provenchere S, Benessiano J, Vicaut E, Lecharny JB, DesmontsJM, Dehoux M, Philip I. Cardiac troponin I is an independent predictor ofin-hospital death after adult cardiac surgery. Anesthesiology 2002;97:405—11.

[5] Paparella D, Cappabianca G, Visicchio G, Galeone A, Marzovillo A, GalloN, Memmola C, Schinosa de L. Cardiac troponin I release after coronaryartery bypass grafting operation: effects on operative and midtermsurvival. Ann Thorac Surg 2005;80:1758—64.

[6] Thielmann M, Massoudy P, Margaaf G, Knipp S, Schmermund A, PiotrowskiJ, Erbel R, Jackob H. Role of troponin I, myoglobin and creatine kinase fordetection of early graft failure following coronary artery bypass grafting.Eur J Cardiothorac Surg 2004;26:102—9.

[7] Costa MA, Carere RG, Lichtenstein SV, Foley DP, de Valk V, LindenboomW,Roose PC, van Geldorp TR, Macaya C, Castanon JL, Fernandez-Avilez F,Gonzales JH, Heyer G, Unger F, Serruys PW. Incidence, predictors, andsignificance of abnormal cardiac enzyme rise in patients treated withbypass surgery in the arterial revascularization therapies study (ARTS).Circulation 2001;104:2689—93.

[8] Brener SJ, Lytle BW, Schneider JP, Ellis SG, Topol EJ. Association betweenCK-MB elevation after percutaneous or surgical revascularization andthree-year mortality. J Am Coll Cardiol 2002;40:1961—7.

[9] Steuer J, Horte LG, Lindahl B, Sthale E. Impact of perioperative myo-cardial injury on early and long-term outcome after coronary arterybypass grafting. Eur Heart J 2002;23:1219—27.

[10] Fellahi JL, Gue X, Richomme X, Monier E, Guillou L, Riou B. Short- andlong-term prognostic value of postoperative cardiac troponin I concen-tration in patients undergoing coronary artery bypass grafting. Anesthe-siology 2003;99:270—4.

[11] Alwan K, Falcoz PE, Alwan J, Mouawad W, Oujaimi G, Chocron S, EtieventJP. Beating versus arrested heart coronary revascularization: evaluationby cardiac troponin I release. Ann Thorac Surg 2004;77:2051—5.

[12] Mack MJ, Pfister A, Bachand D, Emery R, Magee MJ, Connolly M, Sub-ramanian V. Comparison of coronary bypass surgery with and withoutcardiopulmonary bypass in patients with multivessel disease. J ThoracCardiovasc Surg 2004;127:167—73.

[13] Puskas JD, Williams WH, Mahoney EM, Huber PR, Block PC, Duke PG,Staples JR, Glas KE, Marshall JJ, Leimbach ME, McCall SA, Petersen RJ,Bailey DE, Weintraub WS, Guyton RA. Off-pump vs conventional coronaryartery bypass grafting: early and 1-year graft patency, cost, and quality-of-life outcomes: a randomized trial. JAMA 2004;291:1841—9.

[14] Khan NE, De Souza A, Mister R, Flather M, Clague J, Davies S, Collins P,Wang D, Sigwart U, Pepper J. A randomized comparison of off-pump andon-pump multivessel coronary-artery bypass surgery. N Engl J Med2004;350:21—8.

[15] Widimsky P, Straka Z, Stros P, Jirasek K, Dvorak J, Votava J, Lisa L,Budesinsky T, Kolesar M, Vanek T, Brucek P. One-year coronary bypassgraft patency: a randomized comparison between off-pump and on-pumpsurgery angiographic results of the PRAGUE-4 trial. Circulation 2004;110:3418—23.

[16] Racz MJ, Hannan EL, Isom OW, Subramanian VA, Jones RH, Gold JP, RyanTJ, Hartman A, Culliford AT, Bennett E, Lancey RA, Rose EA. A comparisonof short- and long-term outcomes after off-pump and on-pump coronaryartery bypass graft surgery with sternotomy. J Am Coll Cardiol 2004;43:557—64.

[17] Cleveland Jr JC, Shroyer AL, Chen AY, Peterson E, Grover FL. Off-pumpcoronary artery bypass grafting decreases risk-adjusted mortality andmorbidity. Ann Thorac Surg 2001;72:1282—8.

[18] Kini AS, Lee P, Marmur JD, Agarwal A, Duffy ME, Kim MC, Sharma SK.Correlation of postpercutaneous coronary intervention creatinine kinase-MB and troponin I elevation in predicting mid-term mortality. Am JCardiol 2004;93:18—23.

[19] Ricciardi MJ, Davidson CJ, Gubernikoff G, Beohar N, Eckman LJ, ParkerMA, Bonow RO. Troponin I elevation and cardiac events after percuta-neous coronary interventions. Am Heart J 2003;145:522—8.

[20] Nageh T, Sherwood RA, Harris BM, Byrne JA, Thomas MR. Cardiac troponinT and I and creatine kinase-MB as markers of myocardial injury andpredictors of outcome following precutaneous coronary intervention.Int J Cardiol 2003;92:285—93.

[21] Steuer J, Bjerner T, Duvernoy O, Jideus L, Johansson L, Ahlstrom H, StahleE, Lindahl B. Visualization and quantification of peri-operative myocar-dial infarction after coronary artery bypass surgery with contrast-enhanced magnetic resonance imaging. Eur Heart J 2004;25:1293—9.

[22] Gurbuz A, Emrecan B, Yilik L, Ozsoyler I, Kestelli M, Ozbek C, Karahan N.Intracoronary shunt reduces postoperative troponin leaks: a prospectiverandomized study. Eur J Cardiothorac Surg 2006;29:186—9.

[23] Rastan AJ, Bittner HB, Gummert JF, Walther T, Schewick CV, Girdauskas E,Mohr FW. On-pump beating heart versus off-pump coronary artery bypasssurgery-evidence of pump-induced myocardial injury. Eur J CardiothoracSurg 2005;27:1057—64.

[24] Rasmussen C, Thiis JJ, Clemmensen P, Efsen F, Arendrup HC, Saunamaki K,Madsen JK, Pettersson G. Significance and management of early graftfailure after coronary artery bypass grafting. Feasibility and results ofacute angiography and re-revascularization. Eur J Cardiothorac Surg1997;12:847—52.