metformina iam

ORIGINAL ARTICLE

Chronic Metformin Treatment is Associated with ReducedMyocardial Infarct Size in Diabetic Patients with ST-segmentElevation Myocardial Infarction

Chris P. H. Lexis & Wouter G. Wieringa & Bart Hiemstra & Vincent M. van Deursen &

Erik Lipsic & Pim van der Harst & Dirk J. van Veldhuisen & Iwan C. C. van der Horst

# Springer Science+Business Media New York 2013

AbstractPurpose Increased myocardial infarct (MI) size is associatedwith higher risk of developing left ventricular dysfunction,heart failure and mortality. Experimental studies have sug-gested that metformin treatment reduces MI size after inducedischaemia but human data is lacking. We aimed to investigatethe effect of metformin on MI size in patients presenting withan acute MI.Methods All consecutive patients (n =3,288) presenting toour hospital with ST-segment elevation myocardial infarction(STEMI) undergoing primary PCI between January 2004 andDecember 2010 were included in this retrospective analysis.Patients with diabetes were divided according to metforminversus non-metformin based pharmacotherapy. MI size wasestimated using peak values of serum creatine kinase (CK),myocardial band of CK (CK-MB), and troponin-T.Results We identified 677 (20.6 %) patients with diabetes, ofwhom 189 (27.9 %) were treated with metformin. Chronicmetformin treatment was associated with lower peak levels ofCK (1,101 vs. 1,422 U/L, P=0.005), CK-MB (152 vs. 182 U/L, P=0.018) and troponin-T (2.5 vs. 4.0 ng/L, P=0.021)compared to non-metformin using diabetics. After adjustmentfor age, sex, TIMI flow post PCI, and previous MI, the use ofmetformin treatment remained an independent predictor of

smaller MI size. Patient with diabetes treated with metforminhad even smaller MI size than patients without diabetes.Conclusions Chronic metformin treatment is associated withreduced MI size compared to non-metformin based strategiesin diabetic patients presenting with STEMI. Metformin mighthave additional beneficial effects beyond glucose loweringefficacy.

Keywords Myocardial infarction .Metabolism . Diabetesmellitus . Metformin .Myocardial infarct size

Introduction

Acute myocardial infarction (AMI) often results inmyocardialdamage and impaired outcome. Despite improvements inreperfusion therapy and additional pharmacotherapy in thelast decades, mortality rates remain substantial with a 1 yearmortality rate of approximately 10 % as observed in a recentretrospective analysis in over 70,000 patients of the SwedishCoronary Angiography and Angioplasty Registry (SCAAR)[1]. In this population myocardial damage during AMI wasidentified as an important predictor of mortality. Myocardialdamage often results in decreased myocardial function andheart failure, and the amount of myocardium damaged isstrongly related to outcome [2–5]. Therefore, myocardial in-farct (MI) size is a primary determinant of prognosis in pa-tients presenting with AMI [2].

The United Kingdom Prospective Diabetes Study(UKPDS) 34 trial demonstrated that metformin in overweightpatients with type 2 diabetes resulted in a significant riskreduction of 32 % for any diabetes related endpoint, of 42 %for diabetes related death, and of 36 % risk reduction for allcause mortality compared to other antihyperglycaemic strate-gies [6]. In the Diabetes Mellitus Insulin-Glucose Infusion in

C. P. H. Lexis (*) :W. G. Wieringa :B. Hiemstra :V. M. van Deursen : E. Lipsic : P. van der Harst :D. J. van Veldhuisen : I. C. C. van der HorstDepartment of Cardiology, University of Groningen, UniversityMedical Center Groningen, Hanzeplein 1, 9700 RBPO Box 30.001,Groningen, The Netherlandse-mail: [email protected]

I. C. C. van der HorstDepartment of Critical Care, University of Groningen, UniversityMedical Center Groningen, Groningen, The Netherlands

Cardiovasc Drugs TherDOI 10.1007/s10557-013-6504-7

Acute Myocardial Infarction (DIGAMI) 2 study (n =1,145),metformin treatment was associated with improved long-termoutcome [7]. More recently, Zhao and colleagues showed thatTIMI flow and myocardial blush grade, markers of epicardialblood flow and myocardial perfusion respectively, werehigher in MI patients (n =154) on chronic metformin [8](Table 1). Whether these beneficial effect are related to re-duced MI size has not been studied in humans yet.

In several experimental studies the effect of metformin onMI size was observed when started before or after ischaemia-reperfusion (Table 1). Our group demonstrated that metforminadministration started 2 days prior to ischaemia-reperfusionand continued for 12 weeks reducedMI size by 22% in a non-diabetic rat model of MI [9]. Other preclinical studies inrodents found similar results, demonstrating that metforminreduced MI size between 22 % and 65 % (Table 1) [10–14].Metformin has also been demonstrated to improve left ven-tricular function and inhibit progression of HF in both ischae-mic and non-ischaemic models [9, 13–15]. The effect in non-ischaemic models suggests metformin acts also through othermechanisms, such as improved cardiac remodelling. In arecent study chronic metformin treatment in both diabeticand non-diabetic rats augmented myocardial resistance toischaemia-reperfusion injury [14].

Taken together, metformin associated cardioprotective ef-fects may at least partially be caused by reduction in MI size.Therefore we investigated whether chronic metformin in pa-tients presenting with ST-segment elevation myocardial infarc-tion (STEMI) is associatedwith reduction ofMI size comparedto non-metformin using diabetics and to non-diabetics.

Methods

Population

We performed a retrospective cohort study of all consecutivepatients presenting with STEMI (n =3,288) to the UniversityMedical Center Groningen between January 2004 and De-cember 2010 undergoing primary percutaneous coronary in-tervention. Records of patients presenting with symptomssuggesting acute myocardial ischaemia of more than 30 min,time from onset of symptom of less than 12 hours and ST-segment elevation of more than 0.1 mV in two or more leadson the electrocardiogram were collected for the registry. Med-ical history and information recorded during hospitalization,including electrocardiographic, angiographic, clinical, andlaboratory data, were obtained from chart review. Patients onantihyperglycemic treatment at hospital admission, patientswith a previous diagnosis of diabetes mellitus, and patientswith a level of glycosylated hemoglobin (HbA1c) of ≥ 6.5 %(determined at hospital admission) were considered to be

diabetic [16]. The study was approved by the Medical EthicsCommittee of the University Medical Center Groningen.

Angiographic and Laboratory Analyses

The ischaemic time was defined as the time between onset ofsymptoms and the first percutaneous coronary intervention(PCI), i.e. either thrombus aspiration (n =1,943), balloon in-flation (n =1,023), or direct stenting (n =322). The culpritsegment of the MI was classified using the classificationadapted from the 1975 American Heart Association Commit-tee Report by Austen and colleagues [17]. Patients weredivided into anterior vs. non-anterior MI. A culprit lesion insegment 1, 2, 5, 6, and 11 was considered a proximal coronarylesion, whereas a culprit lesion in segment 3, 4, 7, 8, 9, 10, 12,13, 14, 15, and 16 was considered a distal coronary lesion.Thrombolysis in Myocardial Infarction (TIMI) flow gradewas scored before PCI, and TIMI flow grade and myocardialblush grade (MBG) were scored after PCI [18, 19]. TIMI flowgrade was classified as 0: no antegrade flow, 1: minimalantegrade flow into the obstructed segment, 2: slow antegradeflow into the distal bed, and 3: normal antegrade flow into thedistal bed [18]. MBG was assessed for the infarct relatedartery and was defined as previously described by van’t Hofand colleagues: 0, no myocardial blush; 1, minimal myocar-dial blush; 2, moderate myocardial blush; and 3, normalmyocardial blush or contrast density [19]. Persistent myocar-dial blush suggesting leakage of contrast medium into extra-vascular space was given grade 0. The coronary angiogramswere analyzed independently after the procedure.

Serum creatine kinase (CK), myocardial band of CK (CK-MB), and troponin-T measurements were collected in allpatients during their stay in our hospital. The first laboratoryanalysis was performed at the emergency department or thecoronary care unit before coronary intervention or before theprimary PCI from the arterial access site. Following the PCI,blood samples were taken at the coronary care unit accordingto a standardized protocol at 3, 6, 12, 24, 36, and 48 h afterprimary PCI. Marker levels of CK and CK-MB were deter-mined as a UV assay and immunologic UV assay (Mega,Merck, Darmstadt, Germany) from January 2004 till March2006 and thereafter as a UV assay and an immunologic UVassay (Modular P, Roche, Mannheim, Germany). Troponin-Twas determined during the study period as an immunoassay(Modular E, Roche, Mannheim, Germany). Differences ob-served in cardiac markers in the study had no association withdifferences in the analytic system used. The peak levels of CK,CK-MB, and troponin-Twere recorded as an estimation ofMIsize, which have been shown to have a strong correlation toscar tissue in human setting [20]. Further, the admission levelsof serum creatinine, plasma glucose, and glycosylated hemo-globin were also recorded.

Cardiovasc Drugs Ther

Tab

le1

Effectsof

metform

inon

MIsize

andprognosisin

human

(A)andexperimentalsettin

gs(B)

Ref.

Subjects

Studydesign

Nr.of

subjects

Follo

w-up

Effecto

fmetform

inon

blood

glucoselevels

Effecto

fmetform

inon

endpoints

A)Hum

ansetting

[7]

Diabetic

STEMIpatients

Post-hoc

analysisof

RCT

1,145

4.1years(m

edian)

Com

parableto

otherstrategies

Improved

survival(H

Rdeath:

0.65,

0.47–0.90,p=0.01)

[8]

Diabetic

STEMIpatients,

firstM

IRetrospectiv

ecohort

154

Hospitalization

Com

parableto

otherstrategies

Low

erincidenceof

no-reflow

phenom

enon

(4.2vs.14.6%,

P<0.05).

Ref.

Model

Tim

ingof

metform

inadministration

Doseof

metform

inAdm

inistrationroute

InducedI/Rinjury

Effectsof

metform

inon

myocardialinfarctsize

B)Experim

entalsettin

g

[9]

Non-diabetic

rats

2days

priorto

reperfusion,

continuously

250mg/kg

Oral

LADlig

ation,perm

anent

Reduced

infarctsizeby

22%

(29.6±3.2%

vs.38.0±2.2%;

P<0.05)

[10]

Non-diabetic

mice

Duringreperfusion

125μg/kg

Intracardiac

injection

LCAlig

ation60

min;R

eperfusion

4wReduced

Inf/AARby

29%

(48.92

±2.51

%vs.68.63

±2.34

%,P

<0.001)

[11]

RatLangendorff

Perfusionduring

reperfusion

for15

min

50μM

Perfusion

LADlig

ation35

min,R

eperfusion

120min

Reduced

Inf/AARby

55%

(19±4%

vs.42±2%,P

<0.01)

Ratin

vivo

5min

priorto

reperfusion

5mg/kg

Intravenousinfusion

LADlig

ation30

min;R

eperfusion

120min

Reduced

infarctsizeby

45%

(34±6%

vs.62±5%;P

<0.01)

[12]

RatLangendorff

24hpriorto

ischaemia

250mg/kg

Oralg

avage

LCAlig

ation45

min;R

eperfusion

120min

Reduced

Inf/AARby

46%

(I/R:

19.9±3.9%

vs.36.7±3.6%,

P<0.01),

[13]

Non-diabetic

mice

18hpriorto

ischaemia

125μg/kg

Intraperito

neal

injection

LCAlig

ation30

min;R

eperfusion

24h

Reduced

Inf/AARby

62%

(19.35

±1.92

%vs.50.64

±0.86

%,P

<0.001)

Duringreperfusion

125μg/kg

Intracardiac

injection

LCAlig

ation30

min;R

eperfusion

24h

Reduced

Inf/AARby

49%

(25.9±2.73

%vs.

50.64±0.86

%,P

<0.001).

All

125μg/kg

Bothroutes

LCAlig

ation30

min;R

eperfusion

24h

Decreased

troponin-T

rise

by56

%(5.99±0.49

ng/m

lvs.

13.63±3.12

ng/m

l,P=0.026).

[14]

Diabetic

rats

4wpriorto

ligation

300mg/kg

Oral

LADlig

ation35

min,R

eperfusion

60min

Reduced

Inf/AARby

65.4%

(16.6±2.0%

vs.

47.8±2.0%,P

<0.0005)

Non-diabetic

rats

4wpriorto

ligation

300mg/kg

Oral

LADlig

ation35

min,R

eperfusion

60min

Reduced

Inf/AARby

30.7%

(40.8±4.5%

vs.

58.9±3.5%,P

<0.05)

STEMIST

-segmentelevatio

nmyocardialinfarction,RCTrandom

ized

controlledtrial,MImyocardialinfarction,HRhazard

ratio

,ORodds

ratio

,LCAleftcoronary

artery,Inf/AARinfarctsizeperarea

atrisk,L

AD

leftanterior

descending

coronary

artery


Statistical Analysis

Baseline clinical and angiographic data were collected retro-spectively. Data are presented as mean ± standard deviationwhen normally distributed, as median and interquartile rangewhen non-normally distributed, and as frequencies and per-centages for categorical variables. Differences between base-line variables were evaluated by Student’s t test, Kruskal-Wallis test, Mann–Whitney U, Wilcoxon, chi-square, or Fish-er exact tests, when appropriate. Correlations were assessedusing Pearson’s or Spearman’s rank correlation coefficients,when appropriate. Hazard ratios (HRs) for MI size werecalculated with logistic and Cox proportional univariableand multivariable hazard regression analyses, respectively.Logistic regression analysis was used to adjust the correlationsfor potential confounders. In multi linear regression we usedfour known potential variables unrelated to metformin use thatcould be confounding variables for the effect of theantihyperglycaemic treatment on MI size: age, gender, TIMIflow post PCI, and total ischaemic time. To investigate wheth-er outcome was influenced by changes in guidelines andprotocols over time we analyzed the primary outcome of MIsize over time (years). P values≤0.05 were considered toindicate statistical significance. Statistical analyses were per-formed using R version 2.15.1.

Role of the Funding Source

The funding source had no role in the study design, datacollection, data analysis, data interpretation, or writing of thereport. The corresponding author had full access to all the datain the study and had final responsibility for the decision tosubmit the publication.

Results

Patients with Diabetes vs. Patients Without Diabetes

A total of 677 (20.6 %) of 3,288 STEMI patients had diabetes.The baseline characteristics are displayed in Table 2. Patientswith diabetes more often were female, were older, and hadhigher heart rates on admission compared to non-diabetics.Diabetics more often had a history of hypertension, hypercho-lesterolaemia, more often had had a prior MI, had more oftenundergone a PCI, more often had a history of renal dysfunc-tion, peripheral artery disease, and stroke compared to non-diabetics.

The median ischaemic time in patients with diabetes waslonger than in patients without diabetes. We observed nodifferences between both groups regarding culprit lesion –anterior vs. non-anterior or proximal vs. distal culprit lesion–and TIMI flow pre PCI or post PCI. There were no differences

between both groups regarding the first coronary intervention,number of stents and total length of stents. Patients withoutdiabetes had better MBG compared to patients with diabetes.The levels of the first recorded troponin-T and CK-MB weresignificantly higher in patients with diabetes compared to non-diabetics, while there were no differences in peak levels ofCK, CK-MB, or troponin-T between both groups.

Patients with Diabetes on Metformin Versus Patientswith Diabetes Treated Otherwise

Data regarding the antihyperglycaemic treatment during pre-sentation with STEMI was available for 660 of a total of 677(97 %) diabetic patients. Metformin was used by 185 patientswith diabetes (28%). In this group 43% of patients also used asulfonylureum derivate, 16 % also used insulin, and 3 % usedother antihyperglycaemic agents. Of the diabetic patients onother antihyperglycaemic strategies than metformin, 19 % ofpatients used insulin, 9 % were on sulfonylureum derivates,and 70 % of patients did not use an antihyperglycaemic agent(Table 2).

Diabetics on metformin had a higher BMI, less often weresmokers, and more often had hypertension.

Patients on metformin had higher levels of admission glu-cose and higher levels of glycosylated hemoglobin thanpatients treated otherwise.

The median ischaemic time was longer in patients onmetformin than in those treated otherwise. There were nodifferences between both groups in the percentage of patientspresenting with an anterior MI or a proximal culprit lesion. Inpatients on metformin the TIMI flow pre PCI was less often 0compared to those treated otherwise. There were no differ-ences between both groups regarding the first coronary inter-vention, number of stents, and total length of stents. Therewere no differences in TIMI flow post PCI or in MBG.

The first recorded levels of CK, CK-MB, and troponin-Twere comparable between groups (Table 3). In the diabeticpatients treated with metformin peak levels of CK, CK-MB,and troponin-T were all significantly lower compared to thepatients with diabetes treated otherwise (Table 3 and Fig. 1).

The use of metformin was univariately associated withlower peak values of either CK-MB or troponin-T. Whencorrected for age, sex, ischaemic time, TIMI flow post PCI,and history ofMI, the use of metformin remained significantlyassociated with lower peak values of both markers of MI sizein multivariate analysis (Table 4).

The observation that metformin is associated with smallerMI size remained significant even when metformin as mono-therapy was compared to each group of treatment, i.e. eitherdiet, sulphonylurea, insulin, and other antihyperglycaemictreatment (data not shown). We considered these subgroupsin our cohort too small for extensive analyses and conclusions.


Table 2 Basal study subject characteristics displayed for patients without diabetes (NoDM) versus with diabetes (DM), and for patients with diabetes onmetformin (DM + metformin) versus patients with diabetes not on metformin (DM - metformin)

No DM DM P-value DM + Metformin DM - Metformin P-valuen =2611 n=677 n =185 n =475

Demographics

Female (%) 27 37 < 0.001 36 37 0.704

Age (years) 63±13 67±12 < 0.001 66±12 67±13 0.696

BMI (kg/m2) 27±4 28±5 < 0.001 29±5 28±4 0.003

Blood pressure (mmHg)

Systolic 127±27 128±28 0.389 128±25 129±29 0.727

Diastolic 74±15 73±16 0.145 73±13 73±17 0.987

Heart rate (bpm) 77±19 80±21 0.003 84±19 79±22 0.01

Antihyperglycaemic medication

Metformin (%) 0 28 < 0.001 100 0 <0.001

Sulfonylurea derivates (%) 0 19 < 0.001 43 9 <0.001

Insulin (%) 0 18 < 0.001 16 19 0.449

Other (%) 0 1 < 0.001 3 0 <0.001

Diet (%) 0 51 < 0.001 0 70 <0.001

Risk factors and comorbidities

Current smoking (%) 50 48 0.309 40 50 0.034

Positive family history(%) 46 44 0.381 45 44 0.965

Hypertension (%) 36 52 < 0.001 61 49 0.009

Dyslipidemia (%) 26 40 < 0.001 53 35 <0.001

Myocardial infarction (%) 9 12 0.013 15 11 0.162

PCI (%) 7 11 < 0.001 13 10 0.311

CABG (%) 3 3 0.513 4 3 0.362

Renal dysfunction (%) 5 9 < 0.001 9 8 0.916

Peripheral artery disease (%) 2 3 0.004 3 3 0.947

Stroke (%) 4 8 < 0.001 10 8 0.404

Procedural and angiographic characteristics

Ischaemic time (min) 182 215 <0.001 252 200 0.020

LAD culprit laesion (%) 42 44 0.513 49 42 0.132

Proximal laesion (%) 58 61 0.172 58 63 0.289

TIMI flow pre PCI (%) 0.284 0.008

0 52 51 40 55

1 11 12 15 11

2 20 17 20 17

3 18 20 26 17

TIMI flow post PCI (%) 0.621 0.189

0 4 5 2 6

1 8 10 9 11

2 26 24 20 26

3 61 61 69 57

Myocardial blush grade (%) 0.02 0.13

0 7 8 12 6

1 21 25 27 24

2 39 39 36 40

3 33 28 25 29

BMI body-mass index, PCI percutaneous coronary intervention, CABG coronary artery bypass grafting, LAD left anterior descending, TIMIthrombolysis in myocardial infarction


Patients on Metformin vs. Patients Without Diabetes

Large differences in baseline characteristics between patientswith diabetes using metformin and non-diabetics were ob-served, with a higher cardiovascular risk profile in diabeticson metformin. Despite these differences patients with diabetesusing metformin had significantly lower peak levels of CK-MB (152 vs. 169 U/l, p =0.015) and troponin-T (2.5 vs.3.3 ng/l, p =0.021), and non-significant lower peak levels ofCK (1,101 vs. 1,314 U/l, p =0.103) compared to patientswithout diabetes (Table 3).

Discussion

In a large cohort of consecutive patients presenting withSTEMI we observed that chronic metformin use is associated

with smaller MI size. The effect was independent of knownrisk factors for MI size. This effect of metformin on MI sizehas not been studied in clinical setting before and thereforeshould be confirmed by others. What has been shown is thatmetformin treatment has a favorable long-term outcome afterMI [5, 8].

Our observations confirm the advantageous effects on MIsize observed in experimental studies (Table 1) [9–13]. Wefound lower peak values of CK, CK-MB, and troponin-Twhere preclinical studies observed effects of metformin onhistopathologic changes like decreased scar tissue or reducedtroponin-T levels.

The group of patients with diabetes treated otherwise in-cludes patients who were treated for diabetes with insulin,sulfonylurea derivates, or diet, and also includes patients whowere classified as having diabetes based on elevated HbA1clevels. Patients classified as diabetic based on HbA1c, and

Table 3 Laboratory measurements at admission for ST segment eleva-tion myocardial infarction and peak levels during hospital admission ofpatients without (no DM) vs. with diabetes (DM), and of diabetic patients

with metformin (DM + metformin) vs. diabetic patients treated otherwise(DM - metformin)

No DM DM P-value DM + Metformin DM - Metformin P-valuen =2611 n =677 n =185 n=475

Admission

Creatinine (μmol/L) 78 (67–91) 78 (65–100) 0.145 75 (64–96) 80 (67–101) 0.056

Glucose (mmol/L) 8.4±2.7 11.2±4.3 < 0.001 11.9±3.7 10.8±4.5 0.002

HbA1c (%) 5.7 (5.5–6.0) 6.8 (6.5–7.7) < 0.001 7.1 (6.5–7.8) 6.8 (6.5–7.5) 0.038

CK first (U/L) 136 (86–272) 14.6 (88–328) 0.101 155 (90–434) 144 (87–297) 0.359

CK-MB first (U/L) 21 (15–40) 24 (17–49) < 0.001 26 (17–51) 23 (17–46) 0.489

Troponin T first (μg/L) 0.05 (0–0.18) 0.06 (0.01–0.33) 0.002 0.07 (0.02–0.45) 0.06 (0–0.26) 0.059

Peak levels

CK peak (U/L) 1314 (56–2896) 1319 (527–2951) 0.854 1101 (336–2602) 1422 (623–3042) 0.05

CK-MB peak (U/L) 169 (79–322) 171 (72–297) 0.284 152 (58–267) 182 (83–313) 0.018

Troponin T peak (μg/L) 3.3 (1.1–7.0) 3.7 (1.1–8.4) 0.306 2.5 (0.6–7.7) 4.0 (1.4–8.7) 0.021

HbA1c glycosylated hemoglobin, CK creatin kinase, CK-MB myocardial band of creatin kinase

0

200

400

600

DM + Metformin DM - Metformin

b) Peak levels of CK-MB (U/L)

0

2000

4000

6000


a) Peak levels of CK (U/L)

0

5

10

15

20


c) Peak levels of Troponin-T (µg/L)

P=0.05 P=0.018 P=0.021

Fig. 1 Effect of metformin in patients with diabetes on markers of MI size. Box-and-whisker plot representing the p5, p25, p50, p75 and p95 of peak levels ofCK (a), CK-MB (b), and troponin-T (c) for diabetic patients with and without metformin. CK, creatin kinase; CK-MB, myocardial band of creatin kinase


patients on diet are likely to have had shorter duration ofdiabetes. Patients on insulin are likely to have a longer dura-tion of diabetes. The duration of diabetes may influenceoutcome in general and also the response to ischaemia-reperfusion injury [21]. Therefore, we performed subgroupanalyses where metformin as monotherapy was compared toeach separate groups according to treatment and observed thatmetformin remained associated with smaller MI size.

Patients on chronic metformin had higher admission glu-cose levels and higher glycosylated hemoglobin levels com-pared to diabetics on other antihyperglycaemic strategies.Timmer and colleagues demonstrated that higher admissionglucose is strongly associated with larger MI size and resultsin left ventricular dysfunction [22, 23]. In a study usingcardiac magnetic resonance imaging, Mather and colleaguesalso demonstrated that admission glucose levels were associ-ated with myocardial infarct size measured with late gadolin-ium contrast-enhanced imaging [24]. Ultimately, higher ad-mission glucose is consequently associated with impairedoutcome after MI, likely due to increased MI size [22, 23].In our study, the metformin group had higher mean admissionglucose than the patients with diabetes treated otherwise.Thus, it might have been expected that the metformin groupshould have larger MI size, possibly even underestimating theobserved effect size of metformin. Moreover, the observedeffects seem to go beyond the effect of metformin on glucoseregulation.

In our study, patients with diabetes had significantly longerischaemic time, which logically explains that we recordedhigher CK-MB and troponin-T levels at admission. Remark-ably, the longer ischaemic time and also the worse MBG indiabetics did not result higher peak values of CK, CK-MB,and troponin-Tcompared to non-diabetics. This effect resultedentirely from the difference in the metformin group.

Experimental studies suggested that the effects of metfor-min on MI size reduction may be independent ofglycometabolic state [9, 10]. In our study, when comparingpatients with diabetes on metformin treatment to patientswithout diabetes, we found lower peak values of CK-MBand troponin-T in the metformin group. Moreover, in all otherpatients than the patients on metformin, the peak levels were

comparable. Thus, the MI sizes in the non-metforminpopulation irrespective of diabetic state were comparable.Therefore, it can be hypothesized that the MI size reduc-ing effects of metformin may be independent of glucosemetabolism.

Several mechanisms may be involved in metformin asso-ciated MI size reduction, but the exact mechanisms remainelusive [25]. Metformin is associated with enhanced phos-phorylation of AMP activated kinase (AMPK) [9–13, 26–28].AMPK phosphorylation is associated with several targetsassociated with reduction of I/R injury, and leads to activationof the Reperfusion Injury Salvage Kinase (RISK) pathwayincluding phospatidylinositol-3-kinase (PI3K) and Akt path-ways, upregulation of the tumor suppressor gene p53, inhibi-tion of mammalian target of rapamycin (mTOR), and upreg-ulation of endothelial nitric oxide synthase (eNOS) [12, 13,26–28]. Another target of metformin may be the increase ofglucose utilisation of the heart, by facilitating glucose metab-olism via increase of glucose transporters (GLUT-1 andGLUT-4) [9]. Further, metformin activates the adenosine re-ceptor via increased intracellular formation of adenosine, pos-sibly reducing MI size [11]. Also, metformin is associatedwith a decrease in dipeptidyl peptidase-4 activity and anincrease in circulating levels of glucagon-like peptide 1, whichare associatedwith a reduction ofMI size in both experimentaland clinical setting [29–31]. Collectively, these metformininduced changes in myocardial gene and energy program,especially the activation of AMPK, are associated with de-creased MI size [25, 28].

Interestingly, in non-ischaemic canine models of heart fail-ure, metformin improved left ventricular function as well [15].Recently, it was suggested that metformin may improve leftventricular function in heart failure patients with reducedejection fraction, suggesting that the observed post MI im-provement of left ventricular function may be caused bymechanisms other than MI size reduction [32]. Combinedwith our analyses, several concurrent mechanisms may playa role in preserving ejection fraction, both in ischaemic andnon-ischaemic aetiologies. Further, several animal experi-ments have shown that the timing of metformin administrationin ischaemia-reperfusion models is associated with MI size.

Table 4 Relation of metformin on myocardial infarct size. Logartitmic Beta’s and P-values shown for relation of metformin on myocardial infarct size

Log(CKMB-max) Log(Troponin-T-max)

Beta (C.I.) P-value Beta (C.I.) P-value

Metformina −0.21 [−0.39–0.04] 0.018 −0.43 [−0.79–0.15] 0.020

Model 1b −0.20 [−0.39–0.03] 0.021 −0.40 [−0.80–0.15] 0.024

CK-MB myocardial band of creatin kinase, C.I. confidence intervala Univariate analysisbMultivariate analysis adjusted for age, sex, ischaemic time, TIMI flow post PCI, history of myocardial infarction


Metformin started 24 to 48 prior to ligation resulted in evensmaller MI size (albeit non-signifcant) compared tometforminstarted after reperfusion [13]. This suggests that chronic met-formin treatment leads to additional cardioprotective effectsleading to improved reperfusion or improved resilience toischaemia. Since metformin is associated with inhibition ofplasminogen activator inhibitor 1 (PAI-1) and thrombocytemitochondrial complex 1, metformin therapy might reducethrombocyte aggregation [33–36]. In a smaller group of pa-tients Zhao and colleagues demonstrated that chronic metfor-min treatment reduces the no-reflow phenomenon [8]. There-fore, it can be speculated that the observed effect of metforminon MI size may therefore be partly mediated by the effect oncoronary flow. Since TIMI flow post PCI is a better predictorof MI size than TIMI flow pre PCI, we added TIMI flow postPCI to the multivariate model. The effect of metformin on MIsize remained significant in the multivariate analysis.

For future perspectives, prospective trials should investi-gate the effects of metformin treatment on MI size reduction.The current ongoing Glycometabolic Intervention in Patientswith ST elevation myocardial infarction Study (GIPS-III)studies the effect of metformin treatment on left ventricularfunction in non-diabetic patients presenting with STEMI un-dergoing primary PCI [28]. The left ventricular function willbe measured after 4 months by magnetic resonance imaging,as well as the MI size using late gadolineum enhancement. Apossible pitfall may be the timing of metformin administra-tion. Hence, in a prospective design metformin can only bestarted after diagnosis of STEMI and primary PCI, possiblyreducing its effect size. The GIPS-III trial is registered atClinicalTrials.gov, number NCT01217307.

Before conclusions can be drawn from this analysis of alarge cohort of patients, several reservations have to be ad-dressed. First, this analysis was performed using retrospec-tively collected data. Therefore data was not available for allpatients, e.g. data on use of antihyperglycaemic drugs wasavailable for 97 % of patients with diabetes. Second, due tonature of the retrospective design the observed effects ofmetformin may be biased; physicians are likely to prescribemetformin to patients with a higher body mass index. On theother hand, the characteristics of diabetics treated with andwithout metformin were equal for most variables in our co-hort. Third, we used peak levels of CK, CK-MB, andtroponin-T as an estimation of MI size. The use of peak levelsof CK, CK-MB, and troponin-T as enzymatic infarct sizecorrelates well to scar tissue in human setting using cardiacmagnetic resonance imaging with late gadolinium enhance-ment [20]. Ideally we would use area under curve of thesemarkers, however due to our function as a tertiary hospital forthe region and thus fast transfers of patients there was a widevariety in the duration of admission in our hospital. In dailycare, patients were only transferred from the Coronary CareUnit to referring hospitals after peak levels were reached, even

within 24 h after admission. Therefore, peak levels wereavailable for all patients which we decided to use as estimateof MI size.

Conclusion

Our analysis demonstrates that chronic metformin treatment isassociated with reduced MI size compared to non-metforminbased strategies in diabetic patients presenting with STEMI.Our findings suggest that metformin might have additionalbeneficial effects beyond glucose lowering efficacy. Futureprospective studies are needed to elucidate the effect of met-formin on cardioprotection and myocardial function.

Funding Source This study was supported by grant 95103007 fromZonMW, the Netherlands Organization for Health Research and Devel-opment, The Hague, the Netherlands.

Conflict of Interest The authors declare that they have no conflict ofinterest.

References

1. Fokkema ML, James SK, Albertsson P, et al. Population trends inpercutaneous coronary intervention: 20-year results from theSCAAR (Swedish Coronary Angiography and AngioplastyRegistry). J Am Coll Cardiol. 2013;61:1222–30.

2. Task Force on the management of ST-segment elevation acute myo-cardial infarction of the European Society of Cardiology (ESC), StegPG, James SK, Atar D, et al. ESC Guidelines for the management ofacute myocardial infarction in patients presenting with ST-segmentelevation. Eur Heart J. 2012;33:2569–619.

3. Burns RJ, Gibbons RJ, Yi Q, CORE Study Investigators, et al. Therelationship of left ventricular ejection fraction, end-systolic volumeindex and infarct size to six-month mortality after hospital dischargefollowing myocardial infarction treated by thrombolysis. J Am CollCardiol. 2002;39:30–6.

4. van der Vleuten PA, Rasoul S, Huurnink W, et al. The importance ofleft ventricular function for long-term outcome after primary percu-taneous coronary intervention. BMC Cardiovasc Disord. 2008;8:4.

5. Bailey CJ. Metformin: effects on micro and macrovascular com-plications in type 2 diabetes. Cardiovasc Drugs Ther. 2008;22:215–24.

6. UK Prospective Diabetes Study (UKPDS) group. Effect of intensiveblood-glucose control with metformin on complications in over-weight patients with type 2 diabetes (UKPDS 34). Lancet.1998;352:854–65.

7. Mellbin LG, Malmberg K, Norhammar A, Wedel H, Rydén L,DIGAMI 2 Investigators. Prognostic implications of glucose-lowering treatment in patients with acute myocardial infarction anddiabetes: experiences from an extended follow-up of the DiabetesMellitus Insulin-Glucose Infusion in Acute Myocardial Infarction(DIGAMI) 2 study. Diabetologia. 2011;54:1308–17.

8. Zhao JL, Fan CM, Yang YJ, et al. Chronic pretreatment of metforminis associated with the reduction of the no-reflow phenomenon inpatients with diabetes mellitus after primary angioplasty for acutemyocardial infarction. Cardiovasc Ther. 2013;31:60–4.


9. Yin M, van der Horst IC, van Melle JP, et al. Metformin improvescardiac function in a nondiabetic rat model of post-MI heart failure.Am J Physiol Heart Circ Physiol. 2011;301:H459–68.

10. Gundewar S, Calvert JW, Jha S, et al. Activation of AMP-activatedprotein kinase by metformin improves left ventricular function andsurvival in heart failure. Circ Res. 2009;104:403–11.

11. Paiva M, Riksen NP, Davidson SM, et al. Metformin prevents myo-cardial reperfusion injury by activating the adenosine receptor. JCardiovasc Pharmacol. 2009;53:373–8.

12. Solskov L, Løfgren B, Kristiansen SB, et al. Metformin inducescardioprotection against ischaemia/reperfusion injury in the rat heart24 h after administration. Basic Clin Pharmacol Toxicol. 2008;103:82–7.

13. Calvert JW, Gundewar S, Jha S, et al. Acute metformin therapyconfers cardioprotection against myocardial infarction via AMPK-eNOS-mediated signaling. Diabetes. 2008;57:696–705.

14. Whittington HJ, Hall AR, McLaughlin CP, Hausenloy DJ, YellonDM, Mocanu MM. Chronic metformin associated cardioprotectionagainst infarction: not just a glucose lowering phenomenon.Cardiovasc Drugs Ther. 2013;27:5–15.

15. Cittadini A, Napoli R, Monti MG, et al. Metformin prevents thedevelopment if chronic heart failure in the SHHF rat model.Diabetes. 2012;61:944–53.

16. Standards of medical care. Diabetes Care. 2011;34(S1):S11–61.17. Austen WG, Edwards JE, Frye RL, et al. A reporting system on

patients evaluated for coronary artery disease. Report of the Ad HocCommittee for Grading of Coronary Artery Disease, Council onCardiovascular Surgery, American Heart Association. Circulation.1975;51:5–40.

18. Chesebro JH, Knatterud G, Roberts R, et al. Thrombolyis in myocar-dial infarction (TIMI) trial, phase I: a comparison between intravenoustissue plasminogen activator and intravenous streptokinase. Clinicalfindings through hospital discharge. Circulation. 1987;76:142–54.

19. van’t Hof AW, Liem A, Suryapranata H, Hoorntje JC, de Boer MJ,Zijlstra F. Angiographic assessment of myocardial reperfusionin patients treated with primary angioplasty for acute myocar-dial infarction: myocardial blush grade. Circulation. 1998;97:2302–6.

20. Ingkanisorn WP, Rhoads KL, Aletras AH, Kellman P, Arai AE.Gadolineum delayed enhancement cardiovascular magnetic reso-nance correlates with clinical measures of myocardial infarction. JAm Coll Cardiol. 2004;43:2253–9.

21. Whittington HJ, Babu GG, Mocanu MM, et al. The diabetic heart:too sweet for its own good? Cardiol Res Pract. 2012;2012:845698.

22. Timmer JR, van der Horst IC, Ottervanger JP, et al. Prognostic valueof admission glucose in non-diabetic patients myocardial infarction.Am Heart J. 2004;148:399–404.

23. Timmer JR, Hoekstra M, Nijsten MW, et al. Prognostic value ofadmission glycosylated hemoglobin and glucose in nondiabetic pa-tients with ST-segment-elevation myocardial infarction treated withpercutaneous coronary intervention. Circulation. 2011;124:704–11.

24. Mather AN, Creab A, Abidin N, et al. Relationship of dysglycemia toacute myocardial infarct size and cardiovascular outcome as deter-mined by cardiovascular magnetic resonance. J Cardiovasc MagnReson. 2010;12:61.

25. El Messaoudi S, Rongen GA, de Boer RA, Riksen NP. Thecardioprotective effects of metformin. Curr Opin Lipidol. 2011;22:445–53.

26. Jalving M, Gietema JA, Lefrandt JD, et al. Metformin: taking awaythe candy for cancer? Eur J Cancer. 2010;46:2369–80.

27. Bhamra GS, Hausenloy DJ, Davidson SM, et al. Metformin protects theischemic heart by the Akt-mediated inhibition of mitochondrial perme-ability transition pore opening. Basic Res Cardiol. 2008;103:274–84.

28. Lexis CP, van der Horst IC, Lipsic E, for the GIPS-III Investigators,et al. Metformin in non-diabetic patients presenting with STelevationmyocardial infarction: rationale and design of the glycometabolicintervention as adjunct to primary percutaneous intervention in STelevation myocardial infarction (GIPS)-III trial. Cardiovasc DrugsTher. 2012;26:417–26.

29. Cuthbertson J, Patterson S, O’Harte FP, Bell PM. Investigation of theeffect of oral metformin on dipeptidylpeptidase-4 (DPP-4) activity inType 2 diabetes. Diabet Med. 2009;26:649–54.

30. Timmers L, Henriques JP, de Kleijn DP, et al. Exenatide reducesinfarct size and improves cardiac function in a porcine model ofischemia and reperfusion injury. J AmColl Cardiol. 2009;53:501–10.

31. Lønborg J, Kelbæk H, Vejlstrup N, et al. Exenatide reduces finalinfarct size in patients with ST-segment-elevation myocardial infarc-tion and short-duration of ischemia. Circ Cardiovasc Interv. 2012;5:288–95.

32. Lexis CP, van der Horst IC, Lipsic E. Effects of metformin on insulinresistance in heart failure. Which came first: the chicken or the egg?Eur J Heart Fail. 2012;14:1197–8.

33. Tsikouris JP, Suarez JA, Meyerrose GE. Plasminogen activatorinhibitor-1: physiologic role, regulation, and the influence of com-mon pharmacologic agents. J Clin Pharmacol. 2002;42:1187–99.

34. Ersoy C, Kiyici S, Budak F, et al. The effect of metformin treatmenton VEGF and PAI-1 levels in obese type 2 diabetic patients. DiabetesRes Clin Pract. 2008;81:56–60.

35. Protti A, Lecchi A, Fortunato F, et al. Metformin overdose causesplatelet mitochondrial dysfunction in humans. Crit Care. 2012;16:R180.

36. Owen MR, Doran E, Halestrap AP. Evidence that metformin exertsits anti-diabetic effects through inhibition of complex 1 of the respi-ratory chain. Biochem J. 2000;348:607–14.


metformina iam

Documents