Jacob Lønborg,1 Niels Vejlstrup,1 Henning Kelbæk,1 Lars Nepper-Christensen,1 Erik Jørgensen,1

Steffen Helqvist,1 Lene Holmvang,1 Kari Saunamäki,1 Hans Erik Bøtker,2 Won Yong Kim,2 Peter Clemmensen,1

Marek Treiman,3 and Thomas Engstrøm1

Impact of Acute Hyperglycemiaon Myocardial Infarct Size, Areaat Risk, and Salvage in PatientsWith STEMI and the AssociationWith Exenatide Treatment:Results From a RandomizedStudyDiabetes 2014;63:2474–2485 | DOI: 10.2337/db13-1849

Hyperglycemia upon hospital admission in patientswith ST-segment elevation myocardial infarction(STEMI) occurs frequently and is associated withadverse outcomes. It is, however, unsettled as towhether an elevated blood glucose level is the causeor consequence of increased myocardial damage. Inaddition, whether the cardioprotective effect of exena-tide, a glucose-lowering drug, is dependent on hyper-glycemia remains unknown. The objectives of thissubstudy were to evaluate the association betweenhyperglycemia and infarct size, myocardial salvage,and area at risk, and to assess the interaction betweenexenatide and hyperglycemia. A total of 210 STEMIpatients were randomized to receive intravenous exe-natide or placebo before percutaneous coronary in-tervention. Hyperglycemia was associated with largerarea at risk and infarct size compared with patientswith normoglycemia, but the salvage index and infarctsize adjusting for area at risk did not differ between thegroups. Treatment with exenatide resulted in in-creased salvage index both among patients withnormoglycemia and hyperglycemia. Thus, we con-clude that the association between hyperglycemiaupon hospital admission and infarct size in STEMI

patients is a consequence of a larger myocardial areaat risk but not of a reduction in myocardial salvage.Also, cardioprotection by exenatide treatment is in-dependent of glucose levels at hospital admission.Thus, hyperglycemia does not influence the effect ofthe reperfusion treatment but rather represents a sur-rogate marker for the severity of risk and injury to themyocardium.

In patients with ST-segment elevation myocardial in-farction (STEMI), timely treatment with primary percu-taneous coronary intervention (PCI) is recommended torestore coronary blood flow and increase myocardialsalvage, and thereby to minimize infarct size (1). How-ever, after the opening of an occluded vessel, reperfusioninjury may paradoxically cause additional irreversiblemyocardial damage that may account for as much as50% of the infarct size (2). The mechanisms behind reper-fusion injury have not been fully elucidated, but hyper-glycemia, which is observed in approximately half ofpatients with STEMI upon hospital admission (3), maybe unfavorable during reperfusion and has been linked

1Department of Cardiology, Rigshospitalet, Copenhagen University Hospital,Copenhagen, Denmark2Department of Cardiology, Aarhus University Hospital, Skejby, Denmark3Department of Biomedical Sciences and The Danish National FoundationResearch Centre for Heart Arrhythmia, Copenhagen University, Copenhagen,Denmark

Corresponding author: Jacob Lønborg, jacoblonborg@gmail.com.

Received 5 December 2013 and accepted 26 February 2014.

Clinical trial reg. no. NCT00835848, clinicaltrials.gov.

© 2014 by the American Diabetes Association. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

See accompanying article, p. 2209.

2474 Diabetes Volume 63, July 2014

PHARMACOLOGYAND

THERAPEUTIC

S

to the subsequent injury (4). Previous studies have dem-onstrated larger infarct size (5–8) and poorer prognosis inpatients with hyperglycemia upon hospital admissioncompared with patients without hyperglycemia, both inpatients with and without diabetes mellitus (3,5,8–20).Until now, the impact of hyperglycemia on myocardialsalvage has been evaluated in only a limited number ofpatients (21), and no data exist regarding the relationshipbetween hyperglycemia and area at risk. Thus, the causalrole of hyperglycemia in STEMI remains unknown, andwhether the association between hyperglycemia andmyocardial damage may be attributed to a larger myo-cardium at risk or is the cause of reperfusion injuryleading to decreased myocardial salvage needs to beelucidated. Cardiovascular magnetic resonance (CMR)provides an accurate method for in vivo assessment ofinfarct size (22,23), area at risk (24–27), and myocardialsalvage index (28,29).

In a previously published proof-of-concept study(30,31) on patients with STEMI and thrombolysis in myo-cardial infarction (TIMI) grade 0 or 1 flow before primaryPCI, we demonstrated a cardioprotective effect of exena-tide. Exenatide is a glucagon-like peptide analog that isknown to increase insulin secretion and cellular glucoseuptake, and subsequently reduce the level of blood glucose(32). The cardioprotective effect of exenatide may be re-lated to these glucose homeostatic effects (33), and it maybe hypothesized that the effect of exenatide depends onthe patient’s glucose level upon hospital admission andbefore undergoing PCI.

The aims of the current study were to evaluate (1) theassociation between hyperglycemia upon hospital admis-sion and area at risk and myocardial salvage index inSTEMI patients, and (2) the interaction between glycemicstate upon hospital admission and the cardioprotectiveeffect of exenatide.

RESEARCH DESIGN AND METHODS

Study PopulationThe patients in this post hoc study participated in a ran-domized clinical trial comparing intravenous administrationof exenatide with placebo in STEMI patients (30). Patientswith STEMI were randomized to receive either exenatideor placebo saline solution intravenously 15 min prior tointervention and continued 6 h post-PCI. Exenatidetreatment resulted in increased myocardial salvage in-dex, and exenatide also reduced the final infarct size inpatients with short system delays; for more details, seethe existing publications (30,31). STEMI was defined asST-segment elevation in two contiguous electrocardiogram(ECG) leads of .0.1 mV in V4–V6 or limb leads II, III, andaVF, or .0.2 mV in leads V1–V3. Patients were not con-sidered for study enrollment if they presented with cardio-genic shock or were unconscious. Similarly, patients withSTEMI caused by stent thrombosis, known renal insuffi-ciency, or previous coronary artery bypass graft surgerywere excluded from the study. Finally, patients with no

subsequent rise in levels of cardiac biomarkers were ex-cluded from the study. All patients eligible for primaryPCI were pretreated with aspirin (300 mg orally or 500mg i.v.), clopidogrel (600 mg orally), and heparin (10,000units i.v.). On arrival at the catheterization laboratory,a coronary angiography was performed to identify the cul-prit lesion, and primary PCI was performed according tocontemporary guidelines, as previously described (30). Allpatients were treated with clopidogrel, 75 mg daily for12 months, and aspirin, 75 mg daily indefinitely. Cardiacbiomarkers (troponin T) were measured before interven-tion and immediately after, 6 h after, 12–18 h after, andon the day after intervention. The proximal location ofthe culprit lesion was defined as the first segment of theright coronary artery, the left coronary descendingartery, and the left circumflex artery. All patients wereinformed orally and in writing, and all gave their writtenconsent before inclusion. The study was performedaccording to the Helsinki Declaration guidelines forgood clinical practice, and The Danish National Commit-tee on Biomedical Research Ethics approved the proto-col. The study was registered at www.clinicaltrial.gov(NCT00835848).

Blood Glucose and Definition of HyperglycemiaBlood glucose level was measured upon hospital admis-sion at the PCI center before the first angiogram wasperformed as part of a standard evaluation. For patientswithout known diabetes, a cutoff value of 8.3 mmol/L(149 mg/dL) was used for the definition of hyperglyce-mia, whereas for patients with known diabetes a cutoffvalue of 12.8 mmol/L (231 mg/dL) was used (17). Theseparticular cutoff values were chosen, since they previ-ously have been identified as the most optimal cutoffvalues for prediction of outcome in STEMI patients un-dergoing primary PCI for diabetic and nondiabeticpatients, respectively (17). Moreover, different cutoffvalues were used for diabetic and nondiabetic patients,since previous studies (5,17) have demonstrated that theinfarct size in nondiabetic patients increases with hyper-glycemia, but only with server hyperglycemia amongnondiabetic patients. However, a separate analysis wasalso performed using the same cutoff value independentof the diabetic state (8.3 mmol/L). Hypoglycemia wasdefined as a blood glucose level of ,3.3 mmol/L(60 mg/dL) (11).

AngiographyThe angiograms were analyzed for collateral flow to theinfarct-related artery according to the Rentrop classifica-tion grade (0–3), TIMI flow grade (0–3), and for area atrisk according to the APPROACH (Alberta Provincial Pro-ject for Outcome Assessment in Coronary Heart Disease)score (34).

CMRThe CMR protocol and image analyses has previouslybeen described in detail (30). In brief, a CMR scan was

diabetes.diabetesjournals.org Lønborg and Associates 2475

performed from day 1 to day 7 after the primary PCI (25–28), and a second CMR was repeated 3 months 6 3 weeksafter the primary PCI. Both scans were performed on a 1.5Tesla scanner (Avanto; Siemens, Erlangen, Germany) usinga 6-channel body array coil. The myocardial area at risk wasassessed on the first CMR scan as myocardial edema usinga T2-weighted short tau inversion recovery sequence (Fig.1A). On the second CMR examination, delayed enhance-ment images were obtained to determine the final infarctsize using an ECG triggered inversion-recovery sequence(Fig. 1B). These images were acquired 10 min after intra-venous injection of 0.1 mg/kg body weight gadolinium-diethylenetriaminepentaacetic acid (Gadovist; BayerSchering, Berlin, Germany). Left ventricular (LV) volumesand function were measured from both CMR examina-tions using an ECG-triggered, balanced steady-state, freeprecession cine sequence.

An observer blinded to all clinical data analyzed theimages, and for all analyses the endocardial and epicardial

borders were manually traced by the incorporation ofpapillary muscles as part of the LV cavity. The final infarctsize was measured using the free software Segment,version 1.8 (http://segment.heiberg.se) (Fig. 1C) (35).The infarct size, defined as the hyperenhanced myocardiumon the delayed enhancement images, was determined by anautomatic approach, as previously described (35). The in-farct size was expressed as a percentage of the total LVmass and as an absolute mass. Using a postprocessing tool(Argus; Siemens, Erlangen, Germany), the area at risk wasdefined as the hyperintense area on T2-weighted images. Amyocardial area was regarded as hyperintense wheneverthe signal intensity was .2 SDs of the signal intensity inthe normal myocardium (Fig. 1D). The salvage index wascalculated as follows: (area at risk 2 infarct size)/area atrisk (29). On cine short-axis CMR images, the LV volumewas calculated in all 25 phases by manually tracing theendocardial borders using a postprocessing tool (Argus;Siemens). According to the blood pool area, the LV diastolic

Figure 1—Examples of T2-weighted and delayed enhancement images. A delayed enhancement image was used for final infarct sizeanalysis (A), and a T2-weighted image was used for area-at-risk analysis (B). The endocardial and the epicardial borders were manuallytraced in all short-axis images, and the LV myocardial mass was calculated. Papillary muscles were considered parts of the LV cavity. C: Asemiautomatic approach was used to calculate the infarct size. The mean signal intensity was determined automatic in five sectors in eachslice. The region with the lowest mean signal intensity was considered “remote”myocardium. A slice-specific threshold was then set as themean of the remote sector plus 1.8 SDs. In order to take the partial volume effect into account, each pixel within the infarct region was thenweighted according to the signal intensity, where the minimal detectable pixel was set as 10%, with weight shown by pink and yellow,respectively. D: The area at risk was defined as the hyperintense area on T2-weighted images. The remote myocardium was de-termined visually, and the signal intensity within this region was calculated by tracing a region of at least 10 pixels in the middle of themyocardial wall. The signal intensity was then adjusted to the value of the normal + 2 SDs, and the areas of hyperintensity were tracedmanually.

2476 Hyperglycemia and Reperfusion Injury Diabetes Volume 63, July 2014

and systolic frames were then identified; and LV end-diastolic volume, LV end-systolic volume, and LV ejectionfraction were calculated.

Statistical AnalysisPatients with normoglycemia and hyperglycemia werecompared using the two-tailed x2 or Fisher exact test forcategorical variables and the Student t test or Mann-Whitney U test according to normality for continuousvariables. Normality was evaluated by histograms of thestandardized residuals. To compare the relationship be-tween the area at risk and the infarct size, a regressionanalysis was performed, and an ANCOVA was used to testfor equality of the regression lines for the patients withnormoglycemia and hyperglycemia. Multivariable regres-sion models were performed to adjust for potential con-founders using any baseline variable with P, 0.10 for thedifference between normoglycemia and hyperglycemia,and diabetes mellitus status was forced into the modelowing to the possible interaction with hyperglycemia.Among the nondiabetic patients, the correlation betweenblood glucose levels upon hospital admission as a contin-uous variable with area at risk, final infarct size, andmyocardial salvage index was also evaluated using linearregression analyses. Since reperfusion injury occurs in thefirst minutes after reperfusion and TIMI flow grade 0/1seems pivotal for cardioprotection (4), the patients withpre-PCI TIMI 0/1 flow were also analyzed separately.Exenatide treatment may potentially be a confounder;thus, a separate analysis was performed for patients treatedwith placebo. To evaluate the association between ex-enatide treatment and glycemic state upon hospital ad-mission, the patients with TIMI 0/1 flow were stratifiedaccording to normoglycemia and hyperglycemia, and theeffect of exenatide treatment upon myocardial salvageindex was evaluate for each of these groups. Interactionbetween glycemic state and exenatide treatment in termsof myocardial salvage index was evaluated by ANCOVA.The models were constructed as follows: glycemic state,exenatide/placebo, and glycemic state * exenatide/placebo.A two-sided P value ,0.05 was considered statistically sig-nificant. All statistical analyses were performed with SPSSsoftware, version 20 (SPSS, Chicago, IL).

RESULTS

Study PopulationThe patient flowchart is shown in Fig. 2. Blood glucoselevel upon hospital admission was not available in 41patients, and a further 44 patients were not consideredfor study inclusion owing to having aborted STEMI.A total of 92 patients (30%) considered for inclusion inthis study were lost to CMR. The patients lost to CMRwere older, had shorter system delays, and had better pre-PCI TIMI flow compared with the included patients (Table1). We included 210 STEMI patients, of whom 25 patientswere lost to the first CMR, leaving 185 patients with

measurements of both the area at risk and final infarctsize. The median blood glucose levels (interquartile range[IQR]) in patients without and with diabetes were 7.9mmol/L (6.7–9.0 mmol/L) and 11.2 mmol/L (9.2–20.3mmol/L), respectively. No patients had hypoglycemiaupon hospital admission. One hundred twenty-fivepatients (60%) were in a normoglycemic state upon hos-pital admission, and 85 patients (40%) were in a hypergly-cemic state. Baseline characteristics are presented in Table2. Patients with hyperglycemia had a higher frequency ofpre-PCI TIMI flow 0/1, higher levels of glucose, higherpeak troponin T level, higher maximum ST-segment ele-vation before PCI, higher heart rate at hospital discharge,and shorter time from symptom onset until PCI; theywere also more frequently treated with thrombectomyand less frequently randomized to receive exenatide (Ta-ble 2). Moreover, these patients tended to have a higherfrequency of proximal occlusion (Table 2). Importantly,patients with known diabetes were equally distributedbetween the two groups (Table 2). The first and secondCMRs were performed a median time of 2 days (IQR 1–2days) and 90 days (IQR 83–95 days) after the STEMI,respectively, with no difference between the groups.A total of 78% of the population underwent the firstCMR within 48 h after the STEMI.

Hyperglycemia and Area at Risk and MyocardialSalvage IndexSTEMI patients with hyperglycemia had a larger area atrisk measured by CMR than patients with normoglycemia,with medians of 32% LV (IQR 26–39% LV) and 29% LV(IQR 23–36% LV), respectively (Fig. 3A), as measured bythe angiographic APPROACH score (Table 2). As expected,the final median infarct size measured by CMR was alsolarger in the hyperglycemic patients compared with thenormoglycemic patients (11% LV [IQR 6–17% LV] vs. 8%LV [IQR 5–13% LV]) (Fig. 3B). However, the median myo-cardial salvage index determined by CMR did not differbetween the groups (0.71 [IQR 0.64–0.82] and 0.73 [IQR0.61–0.82], respectively) (Fig. 3C). Using the angiographicAPPROACH score to calculate the median myocardial sal-vage index did not change the result (0.61 [IQR 0.47–0.74] vs. 0.62 [IQR 0.45–0.80], respectively; P = 0.58).The final infarct was not different adjusting for area atrisk, since the regression lines for the hyperglycemic andnormoglycemic patients were superimposed (Fig. 3D). Ina multivariable analysis, hyperglycemia was not indepen-dently associated with final infarct size (Table 3).

Using a cutoff value of 8.3 mmol/L to define hyper-glycemia, patients with hyperglycemia still had a largermedian area at risk measured by CMR (29% LV [IQR 22–37% LV] vs. 32% LV [IQR 25–39% LV]; P = 0.049) and theangiographic APPROACH score (27% LV [IQR 19–29%LV] vs. 28% LV [IQR 21–33% LV]; P = 0.036). The asso-ciation between hyperglycemia and final infarct sizeseems to be weaker (9% LV [5–13% LV] vs. 10% LV[IQR 6–16% LV]; P = 0.06). The myocardial salvage index

diabetes.diabetesjournals.org Lønborg and Associates 2477

did not differ between the groups (0.72 [IQR 0.61–0.82]vs. 0.72 [IQR 0.64–0.83]; P = 0.59). Accordingly, the in-farct size was not different adjusting for area at risk (P =0.57). Among the nondiabetic patients the level of bloodglucose upon hospital admission as a continuous variablecorrelated with final infarct size (r = 0.17; P = 0.018), areaat risk by CMR (r = 0.28; P , 0.001), and APPROACHscore (r = 0.15; P = 0.042), but not myocardial salvageindex (r = 0.07; P = 0.36).

A total of 141 patients had pre-PCI TIMI grade 0/1flow, of whom 76 (54%) had normal glucose levels uponhospital admission and 65 (46%) had elevated levels.Evaluating only patients with TIMI 0/1 grade flow beforeundergoing PCI, hyperglycemia was associated witha larger area at risk compared with normoglycemia (33%LV [IQR 27–44% LV] vs. 30% LV [IQR 25–38]; P = 0.042),just as final infarct size tended to be larger (13% LV [IQR

8–18% LV] vs. 10% LV [IQR 7–14% LV]; P = 0.07).However, either the median myocardial salvage index deter-mined by CMR (0.70 [IQR 0.62–0.77] vs. 0.68 [IQR 0.57–0.76]; P = 0.44) or the final infarct size adjusting for the areaat risk (P = 0.98) were significantly different between thetwo groups.

When the patients randomized to receive placebo alone(n = 89) were analyzed, the overall results did not change.Patients with hyperglycemia had a larger myocardiumarea at risk than normoglycemic with median values of32% LV (IQR 26–39% LV) and 27% LV (IQR 22–32% LV)(P = 0.008). The final infarct size was numerically largerwith a median of 12% LV (IQR 6–17% LV) compared with9% LV (IQR 5–13% LV) (P = 0.15). The median myocar-dial salvage index as determined by CMR was 0.70 (IQR0.60–0.79) for patients with hyperglycemia and 0.69 (IQR0.58–0.84) for patients with normoglycemia (P = 0.98).

Figure 2—Flowchart of patient disposition. CABG, coronary artery bypass graft surgery.

2478 Hyperglycemia and Reperfusion Injury Diabetes Volume 63, July 2014

Exenatide and HyperglycemiaAmong patients with TIMI 0/1 grade flow before un-dergoing PCI, exenatide treatment was associated witha lower blood glucose level upon hospital admission,immediately after the PCI, and 6 h after the PCI (Fig. 4).In patients with normoglycemia, treatment with exena-tide resulted in a mean myocardial salvage index deter-mined by CMR of 0.68 6 0.17 compared with 0.62 60.12 for patients treated with placebo (P = 0.08). Amongthe patients with hyperglycemia, the myocardial salvageindex determined by CMR was 0.73 6 0.11 for patientstreated with exenatide, and 0.64 6 0.15 for patients

treated with placebo (P = 0.017). However, there was nostatistically significant interaction between the glycemicstate upon hospital admission and treatment allocation(exenatide vs. placebo) with regard to myocardial salvageindex (P = 0.71).

DISCUSSION

In the current study, we found that hyperglycemia uponhospital admission in STEMI patients treated withprimary PCI is related to a larger myocardium area atrisk and final infarct size, without affecting the potentialfor myocardial salvage. Also, despite the lower blood

Table 1—Baseline clinical, angiographic, and procedural characteristics for patients included and lost to CMR

Characteristics Included (n = 210) Lost to CMR (n = 92) P value

Age, years 61 6 11 64 6 12 0.045

Male 80 75 0.45

Known diabetes mellitus 8 10 0.83

Hypertension 34 43 0.19

Hypercholesterolemia 49 44 0.53

Previous PCI 7 10 0.35

Preinfarct angina 17 23 0.26

Preinfarct medical treatmentb-Blockers 8 8 1.00ACE inhibitors 11 10 0.84Statin 15 18 0.60Aspirin 14 17 0.59

Hospital admission glucose, mmol/L 8.0 (6.8–9.3) 7.9 (6.7–9.6) 0.87

Hyperglycemia upon hospital admission 41 44 0.61

Creatinine, mmol/L 71 (61–86) 71 (58–89) 0.97

Symptom-to-balloon, min 176 (120–270) 195 (137–265) 0.76

System delay, min 132 (102–161) 115 (92–145) 0.009

Maximum ST-segment elevation pre-PCI, mm 3.0 (1.6–5.2) 2.9 (1.5–4.7) 0.69

LAD infarct location 44 42 0.80

TIMI flow pre-PCI0/1 68 662 17 9 0.0433 15 25

Proximal culprit location 31 24 0.27

Bifurcation 7 5 0.45

Collateral flow (Rentrop grade 2/3) 14 13 0.86

Multiple vessel disease 19 18 0.87

Visual thrombus 79 74 0.36

TIMI grade 3 after procedure 94 97 0.42

Randomized to exenatide 57 45 0.06

Thrombectomy 56 62 0.31

StentNo stenting 11 16

0.39Bare metal stent 22 29Drug-eluting stent 67 55

Treatment with GP IIb/IIIa inhibitor 85 87 0.72

Peak troponin T, mg/L 4.1 (1.9–7.3) 3.8 (1.7–8.1) 0.74

Data are presented as the mean6 SD, median (IQR), or %, unless otherwise indicated. GP, glycoprotein; LAD, left anterior descending artery.

diabetes.diabetesjournals.org Lønborg and Associates 2479

glucose level in patients treated with exenatide, there wasno interaction between glycemic state and exenatidetreatment, indicating that the cardioprotective effect ofexenatide treatment is independent of glucose level uponhospital admission.

This is the first study to assess the association ofglycemic state upon hospital admission with the myocar-dium area at risk and myocardial salvage in a large cohortof patients with STEMI. Our data suggest that theexcessive myocardial damage (5–8) and adverse prognosis

Table 2—Baseline clinical, angiographic, and procedural characteristics according to glycemic state upon hospital admission

CharacteristicsNormoglycemic(n = 125 [60%])*

Hyperglycemic(n = 85 [40%])† P value

Age, years 61 6 11 61 6 10 0.84

Male 83 75 0.16

Known diabetes mellitus 8 8 0.97

Hypertension 32 39 0.27

Hypercholesterolemia 49 48 0.89

Previous PCI 6 8 0.46

Preinfarct angina 18 17 0.81

Preinfarct medical treatmentb-Blockers 8 7 0.79ACE inhibitors 9 14 0.26Statin 14 17 0.57Aspirin 12 16 0.50

Hospital admission glucose, mmol/L 6.7 (6.1–7.4) 9.1 (8.4–10.8) ,0.001

Glucose at end of PCI, mmol/L 6.1 (5.4–6.8) 8.2 (7.3–9.9) ,0.001

Time to CMR 1, days 2 (1–2) 2 (1–2) 0.28

Time to CMR 2, days 89 (80–95) 91 (85–95) 0.53

Symptom-to-balloon, min 190 (145–275) 165 (120–260) 0.040

System delay, min 131 (102–165) 132 (101–158) 0.25

Maximum ST-segment elevation pre-PCI, mm 2.6 (1.3–4.8) 3.5 (2.0–5.5) 0.003

LAD infarct location 39 48 0.18

TIMI flow pre-PCI0/1 61 762 21 12 0.063 18 12

Proximal culprit location 27 39 0.07

Bifurcation 6 10 0.42

APPROACH score, % LV (n = 208) 27 (18–29) 28 (22–34) 0.014

Collateral flow (Rentrop grade 2/3) 15 13 0.71

Multiple vessel disease 17 20 0.55

Visual thrombus 78 81 0.64

TIMI grade 3 after procedure 94 95 0.68

Randomized to exenatide 65 47 0.011

Thrombectomy 49 66 0.017

StentNo stenting 12 10

0.50Bare metal stent 24 31Drug-eluting stent 64 59

Treatment with GP IIb/IIIa inhibitor 84 86 0.66

Heart rate at discharge, bpm 73 6 12 77 6 13 0.044

Peak troponin T, mg/L 3.5 (1.6–6.2) 4.4 (2.0–7.9) 0.023

Acute LVEF 54 (47–59) 52 (45–59) 0.34

Follow-up LVEF 58 (52–62) 56 (51–65) 0.64

Data are presented as mean 6 SD, median (IQR), or %, unless otherwise indicated. GP, glycoprotein; LAD, left anterior descendingartery; LVEF, LV ejection fraction. *Blood glucose level ,8.3 or 12.8 mmol/L. †Blood glucose level .8.2 or 12.7 mmol/L.

2480 Hyperglycemia and Reperfusion Injury Diabetes Volume 63, July 2014

(3,5,8–19) reported in patients with hyperglycemia arethe results of a larger myocardial area at risk, and notof a smaller myocardial salvage. Similarly, since therewas no association between hyperglycemia and infarctsize when adjusting for the area at risk, the current studydemonstrates that the association between hyperglycemiaand infarct size is dependent on the size of the area atrisk, and hyperglycemic patients can therefore be consid-ered to be at higher risk per se. The patients with hyper-glycemia more frequently have anterior infarct locationand pre-PCI TIMI flow grade 0/1 and a shorter timefrom symptom onset until balloon (5,8,17). However, ithas previously been demonstrated that the blood glucoselevel upon hospital admission fails to predict prognosiswhen adjusted for other predictors including duration of

symptoms, TIMI flow, and infarct location (8); likewise,hyperglycemia in the current study failed to predict in-farct size in the multivariable analysis. The relationshipbetween hyperglycemia and duration of symptoms mightindicate that the early presenters with more pronouncedsymptoms and faster reaction are at higher risk (36). Im-portantly, there was no difference in system delay, which isa stronger predictor than time from the onset of symptomsto balloon angioplasty (37,38). Thus, it seems more plau-sible that hyperglycemia upon hospital admission in STEMIpatients is an indicator of increased myocardium area atrisk and a marker for the severity of the myocardial dam-age, rather than being a deteriorating factor in itself.

Cell death owing to lethal reperfusion injury cannot bedistinguished from ischemia-induced cell death in vivo,

Figure 3—Impact of hyperglycemia on area at risk, infarct size, and salvage index in patients with STEMI. Box plots (25th percentile,median, and 75th percentile) and whisker (5th and 95th percentile) showing the influence of hospital admission glucose level on CMR-measured myocardial area at risk (A), final infarct size (B), and myocardial salvage index (C). D: Infarct size was also plotted against themyocardial area at risk. There was no difference between the regression lines for patients with normoglycemia and hyperglycemia uponhospital admission (P = 0.82). In both groups, the infarct size correlates with the area at risk (r = 0.80 and r = 0.75, P < 0.001). DM, diabetesmellitus.

diabetes.diabetesjournals.org Lønborg and Associates 2481

but since it occurs during the reperfusion phase a decreasein myocardial salvage or a larger infarct size adjusted forarea at risk would be expected. In this study, hypergly-cemia was not related to either of these in the entirecohort of patients with STEMI or in patients with pre-PCITIMI grade 0/1 flow, which indicates that hyperglycemiadoes not lead to further irreversible myocardial damageduring and after reperfusion with PCI. Thus, our findings

suggest that hyperglycemia upon hospital admission isnot related to lethal reperfusion injury and does notreduce the potential of salvage. This suggests thathyperglycemia is not directly involved in the death ofthe myocytes, but is the consequence of metabolicderangements caused by an increased myocardium atrisk leading to more extensive myocardial damage. In-terestingly, after an acute myocardial infarction the levelof serum insulin is not associated with biomarkersreflecting the infarct size and paradoxically is weaklypositively correlated with glucose levels (39). Hyperglyce-mia upon hospital admission is therefore not explained bya decrease in serum insulin level, but may be the result ofincreased levels of stress hormones leading to greaterglucose availability, which may in fact benefit the hiber-nating myocytes and prevent further apoptosis (40).This may also help to explain the overall disappointingresults of previous studies of glucose-lowering therapyin patients with acute coronary syndrome (41). In ran-domized settings, de Mulder et al. (41) evaluates theeffect of insulin treatment on infarct size in patientspresenting with STEMI and hyperglycemia, and con-cluded that glucose-lowering therapy with insulin didnot reduce infarct size, but paradoxically tended to in-crease the infarct size. In contradiction, in the POST-CON II study exenatide therapy resulted in lowerglucose levels, an increase in myocardial salvage anddecrease in the infarct size adjusted for the area atrisk (30). But the cardioprotective effect of exenatideis still not determined and probably involves a directcardiac action (see below), and the glucose-lowering ef-fect of exenatide is not as dramatic and abrupt as theeffect of insulin and does not result in hypoglycemia.Interestingly, Selker et al. (42) reported a reduction ininfarct size in patients with suspected acute coronarysyndrome and in STEMI patients after prehospital treat-ment with glucose-insulin-potassium. However, the pro-tective effect of glucose-insulin-potassium treatment isconsidered to be the cause of increased glucose uptakeand improved metabolic support to the ischemic myo-cardium and not a reduction in the blood glucose level(40,42).

The cardioprotective effects of glucagon-like peptide 1(GLP-1) and its analogs may be mediated through bothcardiac and noncardiac sites of action. GLP-1 receptorshave been described in cardiomyocytes, endocardium,vascular endothelium, and smooth muscle cells of rodenthearts (43), and several survival signaling pathways havebeen proposed to be activated by these receptors (33).However, the expression of GLP-1 receptors in humanhearts remains controversial (44). Cardiac as well as non-cardiac insulinotropic and insulinomimetic effects, includ-ing an increased glucose uptake, might also contribute tothe cardioprotective action of GLP-1 analogs. Indeed, weobserved a reduced level of glucose before and after thePCI after treatment with intravenous exenatide. However,the cardioprotective effect of exenatide is independent

Figure 4—Effect of exenatide treatment on blood glucose levels inpatients with STEMI and TIMI grade 0/1 flow. The blood glucoselevel after treatment with exenatide (blue columns) is lower com-pared with treatment with placebo (green columns), measured uponhospital admission (before PCI), immediately after the PCI, and 6 hafter the PCI.

Table 3—Multivariable predictors for final infarct size

r value P value

Hyperglycemia 20.08 0.31

Maximum ST-segment elevation pre-PCI 0.36 ,0.001

TIMI flow pre-PCI 20.18 0.028

Proximal culprit location 0.13 0.09

Symptom-to-balloon 0.14 0.06

Thrombectomy 0.06 0.46

Randomized to exenatide 20.04 0.62

Heart rate at discharge, bpm 0.33 ,0.001

2482 Hyperglycemia and Reperfusion Injury Diabetes Volume 63, July 2014

of hyperglycemia upon hospital admission in STEMIpatients, but whether the effect is dependent on increasedglucose uptake remains unknown.

In contrast to findings in the current study, a previousstudy by Teraguchi et al. (21) reports a decreased myo-cardial salvage index in patients with hyperglycemia. Thefollowing important differences between that study andours should be taken into account: 1) Teraguchi et al. (21)did not report data on the area at risk or infarct size, andinformation regarding pre-PCI TIMI flow is missing,which is an important predictor for myocardial salvageindex (45); 2) the time from symptom onset until reper-fusion is twice as long in the study by Teraguchi et al. (21)compared with the present study, with means of 375 and180 min, respectively; and 3) Teraguchi et al. (21) useda cutoff value of 10 mmol/L (180 mg/dL) to identify thepatients with hyperglycemia among both diabetic andnondiabetic patients, which does not seem to be optimalin terms of predicting outcome. Eitel et al. (5) demon-strated that a glucose level .7.8 mmol/L is related tolarger infarct size in nondiabetic patients, whereas inpatients with diabetes a glucose level exceeding 11.1mmol/L was related to larger infarct size. Similar, Teraguchiet al. (21) found that a glucose level exceeding 10 mmol/Lwas related to infarct size in nondiabetic patients but notin patients with diabetes. Planer et al. (17) also reportdifferent cutoff values for diabetic and nondiabetic pa-tients. In 3,405 STEMI patients, they identified 8.3 and12.8 mmol/L as the most optimal cutoff values for pre-dicting 30-day mortality in nondiabetic and diabeticpatients, respectively (17). Thus, these particular cutoffvalues were used in the current study, and the findingsconfirm these values as related to outcome. However, it isstill uncertain whether or not to use the same or differentcutoff values according to patients with diabetes, but pre-vious observations indicate that the importance of glucoselevels in STEMI patients depends on the diabetic state.The current study includes an additional analysis withone definition of hyperglycemia (8.3 mmol/L) withoutchanging the overall results, but the association betweenhyperglycemia and final infarct size may be weaker. Toofew patients with diabetes were included in this study, andit lacks sufficient power to draw any firm conclusions re-garding the use of the same or different cutoff values todefine hyperglycemia. Using one cutoff value will placemost patients with diabetes in the hyperglycemic group,resulting in an inherent bias. Interestingly, Eitel et al. (5)found that hyperglycemia was related to microvascularobstruction by CMR, but the patients were not stratifiedaccording to pre-PCI TIMI flow, a pivotal factor for thePCI-induced reperfusion injury (4). Also, there are no dataon the area at risk or multivariate analysis; thus, whetherhyperglycemia is a cause or a consequence of microvascularobstruction remains unknown. Unfortunately, no data re-garding microvascular obstruction were available, sincedelayed enhancement images were not acquired duringthe initial scan, which is a limitation to this study.

Study LimitationsOwing to the post hoc nature of our study findings,regarding the association between exenatide and bloodglucose levels may only be considered hypothesis gener-ating. Patients in this study were randomized to receiveeither placebo or exenatide, which could have affected theassociation between hyperglycemia and myocardial dam-age, especially since the blood samples were collected afterrandomization. It is therefore important to underscorethat evaluating only the patients treated with placebo didnot change the results. Thus, a possible confounding roleof exenatide appears unlikely, but using all patients in thissubstudy increases the statistical power. Area at risk wasevaluated by assessing myocardial edema after reperfu-sion; thus, hyperglycemia may potentially also change thelevel of myocardial edema. Using CMR has some obviousadvantages, as mentioned in the INTRODUCTION, but alsocarries inherent limitations owing to contraindications.In the current study, a total of 30% of the patientsintended for analysis were lost to CMR, which inducesa certain risk of selection bias. Furthermore, the excludedpatients and the patients lost to CMR may present someof the most critically ill patients (e.g., patients with renalfailure, unconsciousness, or cardiogenic shock), whichintroduces a risk of selection bias. However, the patientslost to CMR were older, but in contrast had better pre-PCITIMI flow and shorter system delay; thus, these patientswere not sicker, which further limits the risk for selectionbias. A relatively wide time window was allowed for thefirst CMR, but the same time window has been used inprevious studies that have validated T2-weighted CMR todelineate area at risk (25–28). The majority of thepatients underwent the first CMR scan within 48 h(78%), and there was no difference between the groups.Importantly, other estimates of the area at risk, such asby the angiography APPROACH score and ST-segmentelevation prior to PCI, were also higher among patientswith hyperglycemia, which supports the CMR findings. Inthe original study, patients were included when they hadTIMI flow grade 0/1 before intervention and no otherstenosis .70% than the culprit lesion (30). While prepro-cedural TIMI flow is well-established as a pivotal determi-nant of reperfusion injury, the role of multivessel diseaseis more controversial (4). Consequently, in order to in-crease the statistical power of the present analysispatients with multivessel disease were not excluded. Ow-ing to many statistical comparisons, some significant dif-ferences may be observed by chance leading to a risk oftype II errors, and the results must be interpreted cau-tiously and need to be confirmed.

ConclusionsThe association between hyperglycemia upon hospitaladmission and larger infarct size in STEMI patientstreated with PCI is explained by a larger myocardial areaat risk but not by a reduction in myocardial salvage. Also,cardioprotection by exenatide treatment is independent

diabetes.diabetesjournals.org Lønborg and Associates 2483

of hospital admission glucose levels. Thus, we concludethat hyperglycemia does not influence the effect of thereperfusion treatment but rather serves as a marker forthe severity of myocardium at risk and injury.

Funding. This research was supported by the Danish National ResearchFoundation for Heart Arrhythmia, the Novo Nordisk Foundation, Danielsen’sFoundation, Rigshospitale’s Research Foundation, and the Danish HeartFoundation.Duality of Interest. H.E.B. is a shareholder in CellAegis Inc. P.C. hasreceived grants from Eli Lilly. T.E. has received grants from and is a memberof the advisory board of Eli Lilly. No other potential conflicts of interest relevantto this article were reported.Authors Contributions. J.L. contributed to the conceptual design, thefollow-up of patients, and data quality control; performed the cardiovascularmagnetic resonance scan and analysis and angiography; researched the data;performed the general data analysis; and wrote the manuscript, including sta-tistics, tables, and figures. N.V. and W.Y.K. contributed to conceptual design andparticipated in the cardiovascular magnetic resonance scan protocol and analy-sis. H.K., E.J., S.H., L.H., K.S., H.E.B., P.C., and M.T. contributed to conceptualdesign and data analysis and recruitment and enrollment of patients. L.N.-C.contributed to conceptual design, data analysis, data research, and follow-up ofpatients. T.E. was principal investigator and contributed to conceptual design anddata analysis, angiographic data analysis, data research, and recruitment andenrollment of patients. All authors have contributed significantly to the making ofthe manuscript, to trial design, and to the reviewing and editing of the manuscript.J.L. and T.E. are the guarantors of this work and, as such, had full access to all thedata in the study and take responsibility for the integrity of the data and theaccuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 25thTranscatheter Cardiovascular Therapeutics Annual Scientific Symposium, SanFrancisco, CA, 27 October–1 November 2013.

References1. Hamm CW, Bassand JP, Agewall S, et al.; ESC Committee for PracticeGuidelines. ESC Guidelines for the management of acute coronary syndromes inpatients presenting without persistent ST-segment elevation: the Task Force forthe management of acute coronary syndromes (ACS) in patients presentingwithout persistent ST-segment elevation of the European Society of Cardiology(ESC). Eur Heart J 2011;32:2999–30542. Yellon DM, Hausenloy DJ. Myocardial reperfusion injury. N Engl J Med2007;357:1121–11353. Kosiborod M, McGuire DK. Glucose-lowering targets for patients with car-diovascular disease: focus on inpatient management of patients with acutecoronary syndromes. Circulation 2010;122:2736–27444. Hausenloy DJ, Erik Bøtker H, Condorelli G, et al. Translating cardioprotectionfor patient benefit: position paper from the Working Group of Cellular Biology ofthe Heart of the European Society of Cardiology. Cardiovasc Res 2013;98:7–275. Eitel I, Hintze S, de Waha S, et al. Prognostic impact of hyperglycemia innondiabetic and diabetic patients with ST-elevation myocardial infarction: in-sights from contrast-enhanced magnetic resonance imaging. Circ CardiovascImaging 2012;5:708–7186. Jensen CJ, Eberle HC, Nassenstein K, et al. Impact of hyperglycemia atadmission in patients with acute ST-segment elevation myocardial infarction asassessed by contrast-enhanced MRI. Clin Res Cardiol 2011;100:649–6597. Mather AN, Crean A, Abidin N, et al. Relationship of dysglycemia to acutemyocardial infarct size and cardiovascular outcome as determined by cardio-vascular magnetic resonance. J Cardiovasc Magn Reson 2010;12:618. Timmer JR, Hoekstra M, Nijsten MW, et al. Prognostic value of admission gly-cosylated hemoglobin and glucose in nondiabetic patients with ST-segment-elevation

myocardial infarction treated with percutaneous coronary intervention. Circulation2011;124:704–7119. Capes SE, Hunt D, Malmberg K, Gerstein HC. Stress hyperglycaemia andincreased risk of death after myocardial infarction in patients with and withoutdiabetes: a systematic overview. Lancet 2000;355:773–77810. Foo K, Cooper J, Deaner A, et al. A single serum glucose measurementpredicts adverse outcomes across the whole range of acute coronary syndromes.Heart 2003;89:512–51611. Goyal A, Mehta SR, Díaz R, et al. Differential clinical outcomes associatedwith hypoglycemia and hyperglycemia in acute myocardial infarction. Circulation2009;120:2429–243712. Hoebers LP, Damman P, Claessen BE, et al. Predictive value of plasmaglucose level on admission for short and long term mortality in patients with ST-elevation myocardial infarction treated with primary percutaneous coronary in-tervention. Am J Cardiol 2012;109:53–5913. Ishihara M, Kagawa E, Inoue I, et al. Impact of admission hyperglycemia anddiabetes mellitus on short- and long-term mortality after acute myocardial in-farction in the coronary intervention era. Am J Cardiol 2007;99:1674–167914. Ishihara M, Kojima S, Sakamoto T, et al.; Japanese Acute Coronary Syn-drome Study Investigators. Acute hyperglycemia is associated with adverseoutcome after acute myocardial infarction in the coronary intervention era. AmHeart J 2005;150:814–82015. Kadri Z, Danchin N, Vaur L, et al.; USIC 2000 Investigators. Major impact ofadmission glycaemia on 30 day and one year mortality in non-diabetic patientsadmitted for myocardial infarction: results from the nationwide French USIC 2000study. Heart 2006;92:910–91516. Kosiborod M, Rathore SS, Inzucchi SE, et al. Admission glucose andmortality in elderly patients hospitalized with acute myocardial infarction: im-plications for patients with and without recognized diabetes. Circulation 2005;111:3078–308617. Planer D, Witzenbichler B, Guagliumi G, et al. Impact of hyperglycemia inpatients with ST-segment elevation myocardial infarction undergoing percuta-neous coronary intervention: the HORIZONS-AMI trial. Int J Cardiol 2013;167:2572–257918. Stranders I, Diamant M, van Gelder RE, et al. Admission blood glucose levelas risk indicator of death after myocardial infarction in patients with and withoutdiabetes mellitus. Arch Intern Med 2004;164:982–98819. Wahab NN, Cowden EA, Pearce NJ, Gardner MJ, Merry H, Cox JL; ICONSInvestigators. Is blood glucose an independent predictor of mortality in acutemyocardial infarction in the thrombolytic era? J Am Coll Cardiol 2002;40:1748–175420. Ekmekci A, Cicek G, Uluganyan M, et al. Admission hyperglycemia predictsinhospital mortality and major adverse cardiac events after primary percutaneouscoronary intervention in patients without diabetes mellitus. Angiology 2014;65:154–15921. Teraguchi I, Imanishi T, Ozaki Y, et al. Impact of stress hyperglycemia onmyocardial salvage following successfully recanalized primary acute myocardialinfarction. Circ J 2012;76:2690–269622. Mahrholdt H, Wagner A, Holly TA, et al. Reproducibility of chronic infarct sizemeasurement by contrast-enhanced magnetic resonance imaging. Circulation2002;106:2322–232723. Kim RJ, Fieno DS, Parrish TB, et al. Relationship of MRI delayed contrastenhancement to irreversible injury, infarct age, and contractile function. Circu-lation 1999;100:1992–200224. Aletras AH, Tilak GS, Natanzon A, et al. Retrospective determination of thearea at risk for reperfused acute myocardial infarction with T2-weighted cardiacmagnetic resonance imaging: histopathological and displacement encoding withstimulated echoes (DENSE) functional validations. Circulation 2006;113:1865–187025. Carlsson M, Ubachs JF, Hedström E, Heiberg E, Jovinge S, Arheden H.Myocardium at risk after acute infarction in humans on cardiac magnetic reso-nance: quantitative assessment during follow-up and validation with single-

2484 Hyperglycemia and Reperfusion Injury Diabetes Volume 63, July 2014

photon emission computed tomography. JACC Cardiovasc Imaging 2009;2:569–57626. Hadamitzky M, Langhans B, Hausleiter J, et al. The assessment of area atrisk and myocardial salvage after coronary revascularization in acute myocardialinfarction: comparison between CMR and SPECT. JACC Cardiovasc Imaging2013;6:358–36927. Wright J, Adriaenssens T, Dymarkowski S, Desmet W, Bogaert J. Quanti-fication of myocardial area at risk with T2-weighted CMR: comparison withcontrast-enhanced CMR and coronary angiography. JACC Cardiovasc Imaging2009;2:825–83128. Friedrich MG, Abdel-Aty H, Taylor A, Schulz-Menger J, Messroghli D, DietzR. The salvaged area at risk in reperfused acute myocardial infarction as visu-alized by cardiovascular magnetic resonance. J Am Coll Cardiol 2008;51:1581–158729. Eitel I, Desch S, Fuernau G, et al. Prognostic significance and determinantsof myocardial salvage assessed by cardiovascular magnetic resonance in acutereperfused myocardial infarction. J Am Coll Cardiol 2010;55:2470–247930. Lønborg J, Vejlstrup N, Kelbæk H, et al. Exenatide reduces reperfusion injuryin patients with ST-segment elevation myocardial infarction. Eur Heart J 2012;33:1491–149931. Lønborg J, Kelbæk H, Vejlstrup N, et al. Exenatide reduces final infarct sizein patients with ST-segment-elevation myocardial infarction and short-duration ofischemia. Circ Cardiovasc Interv 2012;5:288–29532. Holst JJ. The physiology of glucagon-like peptide 1. Physiol Rev 2007;87:1409–143933. Treiman M, Elvekjaer M, Engstrøm T, Jensen JS. Glucagon-like peptide1—a cardiologic dimension. Trends Cardiovasc Med 2010;20:8–1234. Ortiz-Pérez JT, Meyers SN, Lee DC, et al. Angiographic estimates ofmyocardium at risk during acute myocardial infarction: validation study usingcardiac magnetic resonance imaging. Eur Heart J 2007;28:1750–175835. Heiberg E, Ugander M, Engblom H, et al. Automated quantification ofmyocardial infarction from MR images by accounting for partial volume effects:animal, phantom, and human study. Radiology 2008;246:581–588

36. Nallamothu BK, Bates ER, Herrin J, Wang Y, Bradley EH, Krumholz HM;NRMI Investigators. Times to treatment in transfer patients undergoing primarypercutaneous coronary intervention in the United States: National Registry ofMyocardial Infarction (NRMI)-3/4 analysis. Circulation 2005;111:761–76737. Lønborg J, Schoos MM, Kelbæk H, et al. Impact of system delay on infarctsize, myocardial salvage index, and left ventricular function in patients with ST-segment elevation myocardial infarction. Am Heart J 2012;164:538–54638. Terkelsen CJ, Sørensen JT, Maeng M, et al. System delay and mortalityamong patients with STEMI treated with primary percutaneous coronary in-tervention. JAMA 2010;304:763–77139. Stubbs PJ, Laycock J, Alaghband-Zadeh J, Carter G, Noble MI. Circulatingstress hormone and insulin concentrations in acute coronary syndromes: iden-tification of insulin resistance on admission. Clin Sci (Lond) 1999;96:589–59540. Oliver MF, Opie LH. Effects of glucose and fatty acids on myocardial is-chaemia and arrhythmias. Lancet 1994;343:155–15841. de Mulder M, Umans VA, Cornel JH, et al. Intensive glucose regulation inhyperglycemic acute coronary syndrome: results of the randomized BIOMarkerstudy to identify the acute risk of a coronary syndrome-2 (BIOMArCS-2) glucosetrial. JAMA Intern Med 2013;173:1896–190442. Selker HP, Beshansky JR, Sheehan PR, et al. Out-of-hospital administrationof intravenous glucose-insulin-potassium in patients with suspected acute cor-onary syndromes: the IMMEDIATE randomized controlled trial. JAMA 2012;307:1925–193343. Ban K, Noyan-Ashraf MH, Hoefer J, Bolz SS, Drucker DJ, Husain M.Cardioprotective and vasodilatory actions of glucagon-like peptide 1 receptorare mediated through both glucagon-like peptide 1 receptor-dependent and-independent pathways. Circulation 2008;117:2340–235044. Sivertsen J, Rosenmeier J, Holst JJ, Vilsbøll T. The effect of glucagon-likepeptide 1 on cardiovascular risk. Nat Rev Cardiol 2012;9:209–22245. Lønborg J, Kelbæk H, Vejlstrup N, et al. Influence of pre-infarction angina,collateral flow, and pre-procedural TIMI flow on myocardial salvage index bycardiac magnetic resonance in patients with ST-segment elevation myocardialinfarction. Eur Heart J Cardiovasc Imaging 2012;13:433–443

diabetes.diabetesjournals.org Lønborg and Associates 2485