DIABETES/METABOLISM RESEARCH AND REVIEWS RESEARCH ARTICLE || Diabetes Metab Res Rev 2005; 21: 143-149, Published online 28 July 2008 in Wiley interSelence (www interselence.wiley.com). DOK: 10.1002/dmrr.498 Role of previous treatment with sulfonylureas in diabetic patients with acute myocardial infarction: results from a nationwide French registry Nicolas Danchin!* Abstract Guillaume Charpentier? Frangois Ledru? Background The cardiovascular effects of sulfonylureas (SU) in diabetic Laurent Vaur? | patients are controversial and it has been suggested that diabetic patients Pascal Guéret* ‘with acute myocardial infarction while on SU were at increased risk. Guy Hanania® | Didier Blanchard® Jean-Marc Lablanche? | Nathalie Gens? | Jean-Pierre Cambou® Objectives To assess the in-hospital outcome of patients with acute myocardial infarction according to the use of SU at the time of the acute episode, Methods Of 443 intensive care units in France, 369 (83%) prospectively collected all cases of infarction admitted within 48 h of symptom onset in Service de Cardiologie, HEGP, Paris, | November 2000, France Corbeil, France diabetes, of whom 215 (44%) were on SU. Patients on SU were older and Laboratoire Aventis, Pars, France had a more frequent history of hyperlipidemia than those not receiving SU. CHU Henri Mondor, Créteil, France | Type and location of infarction were similar in the two groups, and there was no difference in Killip class on admission. In-hospital morality was lower in patients on SU (10.2%) than in those without SU (16.996) (p = 0.035), There was a trend toward less frequent ventricular fibrillation (2.3% vs 5.9%, p = 0.052). In two models of multivariate analyses, SU therapy was associated with decreased in-hospital mortality (model 1: relative risk: 0.44, = 0.012; model 2: relative risk: 0.37, p = 0.020). Scentre Hospitalier d’Aulnay, Aulnay, France Clinique St Gatien, Tours, France *CHU de Lille, Lille, France SINSERM U 558, Toulouse, France ee denen See de coolone Hapa Eropen Georges | ponpido, 20 ue Blane, 75015 enema ae Keywords myocar ifr: nes ae units sunray: diabetes peas Conclusions In this nationwide registry reflecting real-world practice, the use of sulfonylureas in diabetic patients was not associated with increased in-hospital morality. Copyright © 2005 John Wiley & Sons, Ltd, In diabetic patients, it has been suggested that sulfonylurea therapy (SU) might exert a deleterious role on cardiovascular events, because of its poten: tial impact on myocardial ischemic preconditioning (1,21: experimentally, it hhas been shown that some sulfonylureas that act on myocardial as well as pancreatic Kxrp channels are able to block preconditioning mechanisms (3,41 In the clinical setting, however, only limited data are available, and the results reported from different observational studies are conflicting [5-9]. The USIC Rec 18 Fray 2004 2000 survey isa prospective registry of all patients admitted to intensive care fee: 26 Apr 2004 tnt in Prance for acute myocardial infarction (AMD by the end of 2000: As Accepted: 14 May 2004 it collected the treatments currently used by the patients at the time their. unis ost: 6.04208 copyrgh© 208 Jn Wie & on Li eh UBSZ9SENGO {Aa14 infarction occurred, it offers the unique opportunity to assess the clinical course according to the differ- tent medication categories. The present study sought to assess the impact of current treatment with sul fonylureas on the in-hospital outcome in diabetic patients admitted for acute myocardial infarction. Methods Population The objective of the USIC 2000 study was to gather complete and representative data on the management and outcome of patients admitted to intensive care units for AMI over a one-month period in France, irrespective of the type of institution to which the patients were admitted (ie.: university hospitals, public hospitals or private clinics). The methodology of the registry has been described elsewhere [10]. Of the 443 centers with intensive care units that took care of patients with AMI in France in 2000, 369 participated in the study (83%). The participation rate was 88% (n = 43) for university hospitals, 85% (n = 229) for public hospitals and 79% (n = 97) for private clinics. One physician responsible for the study was recruited in each center and filled in a case record form for each patient meeting the inclusion criteria and admitted to the intensive care unit during the study recruitment period; the physicians in charge of the patients took care of them according to their usual practice and independently of che study. Patient selection All consecutive patients admitted to the participating centers from 1 November through 30 November 2000 were included in the study if they met the following criteria: = diagnosis of AMI on the basis of the presence of (1) serum markers elevation higher than twice the ‘upper limit ofthe normal for CK, CK-MB or troponins and (2) either chest pain lasting for at least 30 min and not relieved by nitrates or compatible symptoms ‘and/or ECG changes on at least two contiguous leads with pathologic Q waves (at least 0.04) and/or persisting ST elevation or depression >0.1 mV. ~ time from the beginning ofthe symptoms to admission to the intensive care unit less than 48 h. For the present study, only patients with diabetes mellitus were included. Patients were considered to have diabetes mellitus if they received insulin or oral antidiabetic medications, or if they had previously documented fasting hyperglycemia (> 1.26 g/1). \Coprighe © 2005 Jn Wily & Sons, i N.Danchin eta. Data-collection ‘The patients’ cardiovascular history, their current med- ications at the time of admission, including type of antidiabetic medications, their risk factors (smoking sta tus, history of hypertension or treated hypertension, cholesterol level >2.5 g/L or treated hyperlipidemia, family history, diabetes mellitus), their in-hospital clin- ical course, including maximal Killip class, as well as the initial diagnostic and therapeutic management were recorded for each patient. Furthermore, left ventricular jection fraction (LVEF), when assessed at any time dur ing the first 5 days, was recorded. The value for LVEF, which was used for the present analyses, was determined by the following priority ranking of the method used: ) LY contrast angiography, (2) radionuclide angiog. raphy, (3) echocardiography using Simpson's method, (4) echocardiography. Data were coffected by an indepen- dent contract research organization and analysis of the data used in the present article was made independently of the sponsor. Statistical analysis All continuous variables are described as their mean values + standard deviation. Comparisons between ‘groups were made using unpaired t-tests for continuous variables and chi square tests for discrete variables. Mul- tivariate stepwise logistic regression analysis was used to assess the independent prognostic value of baseline parameters on in-hospital outcome. Variables with a value of less than 0.20 on univariate analyses were inchided in the models. In addition, we added to the logistic regression model a variable reflecting the results of a propensity analysis of the prescription of sulfony- lureas before the index event: the propensity score for the prescription of sulfonylureas was calculated by using multivariate logistic regression analysis and the popu- lation was categorized into four quartiles of increased likelihood for the prescription of sulfonylureas. Two mt tivariate models were used, both including the propei score. The first model inchided demographic and medical history variables, as well as medications used at the time of infarction. The second model also included Killip class, heart rate, blood pressure on admission, and initial level of glycemia, which are classic prognostic indicators but which might have been influenced by the medications used at the time of onset of the acute episode (eg. sul fonylureas might have an effect on the size of infarction, and therefore on hemodynamic variables on admission). Foralltests,ap-value of <0.05 was considered significant. Results, Baseline parameters Of the 2320 patients included in the USIC 2000 study, 487 (21%) had diabetes mellitus. Among them, 215 Diabeves Meta Res Rey 2005; 21: 143-148.‘Suifonylureas and Acute Myocardial Infarction (44%) were on SU when their myocardial infarction occurred. Patients on SU were older, and had a more ‘common history of hyperlipidemia than patients without, SU. All other variables were similar between the two ‘groups (Table 1). Preadmission medications were similar as for antiplatelet agents, ACE-inhibitors, and beta-blockers. Patients on SU, however, were more likely to receive statins and metformin, and less likely to receive insulin, The percentages of Q.wave or ST-elevation Mls, of anterior Mis as well as Killip class on admission were Table 1, Baseline parameters and inital treatment during the first 48 h of the current ‘admission for all diabetic patients admitted for acute myocardial infarction No suifonjureas On sufonylreas Variable wean Nents pre ‘age yeas) e314 new (0.027 ‘ex women) 103 (37.9 74046) NS Body mas index > 30 61058) 54278) NS Risk factors Smoking 126 46.3) 103 47.9) NS. Hypetipidemia mar9) 16521) 0027 Hypertension| 1701625) 148 68.8) NS Previous history « Peripheral vascular dlsease 53.1195) 271128) NS Previous Mi e228) 7219) NS. car 261) 22(10.2) Ns Stroke 25 (82) 20183) NS cass 15655) 1161) NS ra 3114) 30 (140) Ns Renal nstficieney 23108) 16174) NS Curren episode of Anterior Mt 103379) 20377) Ns (Qeuave, T-clevation Mi or L668 2221818) 174 (603) NS Klip class on admission NS t 181,665) 140 65.1) oh 51 (188) 47219) 28103) 22 (102) v ial 6128) ‘Admission heart ate (opm) Bb 2t 835424 Ns Intl systolic Blood pressure (mm Hg) 133-4 30, 103 528 0.0001 Fist measured glycemia (gia) 240 + 105, 2504 108 NS mmott) 33458 iaos60 HbA (6) fo = 168), si20 Bis19 Ns Treatment before admission Insutin| 91035) 13460) 0.0001 Metformin 43178) 63283) 0.002 Antiplateet agents 93342) 760353) Ns Oral anticoagulants 12a) 1884) 007 totes 43180) 47219) NS ACEnhibitors 5721.0) e0a73) NS ‘RBs 2107) 20183) NS Beta blockers 67 a6) 5304) NS Statins 43158) 580270) 0.003 Fibates 1366) 281130) oor. Cakium antagonists 54238) 53293) NS Digoxin 9133) 1266) NS Diuretics 630232) 65.003) NS Treatment during te fist 48h Insulin 174 (640) 110651.) 0.008 Sulforylureas 1300) 92 (428), 0.0001 Metformin 1461) 11651) NS Unfractionated heparin 215 (780) 157073.) NS ow molecular weight heparin 78073) 66 (307), NS ‘Gycoprotein yl Blockers 53(195) 3518.1) NS Antiplateet agents 250(91.9) 201 35) NS (ral anticoagulants 403) 3114) NS Beta biockes 160158.) 1391687) NS ACEiohitors 120 (48.1) 109 60.7) Ns Statins | 102075) 96 (44.7) Ns Nitrates 176 (647) 161783) ors Dobutamineinorepnephrine 34) 20183) NS vas 140) 5102) NS ACE, angiotensin converting enzyme; ARBs, angiotensin receptor blockers, CABG, coronary bypass ‘grating: CHE, congestive heart fare; IABP, intasoricbaloon pump; LBS, let bundle Branch Bock, Hl, myocardial infarction; PCL, percutaneous coronary interventon Copyright © 2005 John Wy & Sons, Ld. labs eta Res Rev 2005; 21: 143-149146 similar in the two groups. Patients without SU had lower initial values of systolic arterial pressure. Glycemia at hospital admission, which was a significant predictor of ‘outcome in the whole population (Figure 1), was similar in the two groups (without SU vs with SU: 240 + 105 vs 250+ 109 mg/dL or 13.345.8 mmol/L vs 13.9 +6.0 mmol/L). Using multiple logistic regression. analysis, five variables were found independently correlated with sulfonylurea use before hospital admission and were used for calculating the propensity score: age, use of statins, ‘metformin, fibrates and absence of insulin treatment. The distribution of the propensity scores across quartiles is shown in Table 2. In-hospital management Reperfusion therapy was used in a similar number of patients with (35.3%) or without (34.7%) SU before ‘admission; 16% in each group received thrombolysis and 19% in each group had primary percutaneous coronary interventions. Medications administered during the first 48h are described in Table 1. Glycoprotein I1b/Illa inhibitors, heparin and aspirin were administered as Frequently in the two groups. Likewise, there were no differences in the use of beta-blocking agents or ACE- inhibitors. Fibrates were more frequently used in patients on SU, and there was a trend towards a more frequent use of statins. Insulin was used in §19% of patients initially on SU, significantly less than the 64% in patients without SU. Metformin was used in 5% of the patients in each group. 18 162 16 4 2 104 8) 7 | ‘ 2 4 Mortality ar hee 1IG135 136-176 on 2113 >176 Figure 1. tn-hospital mortality according to the quartile of lycemia on the first measurement following hospital admission inthe whole population (with or without sulfonylureas ‘on admission) Table 2. Propensity scores for sulfonylurea treatment by qua N.Danchin eta. Clinical course and in-hospital outcome (Table 3) Inchospital mortality was 10.2% in patients with versus 16.9% in patients without SU (p = 0.035). The main difference in outcome was observed in the first days, with a 5-day morality of 6.0 versus 12.1% (p = 0.023), ‘There was a trend toward more frequent ventricular fbrillaion (5.9 vs 2.3%, p ~0.052) and stroke (3.0 vs 0.9%, p = 0.11) in patients without SU. In patients on SU, in-hospital mortality was 10.0% in those who received insulin therapy within the fist 2 days of admission and 10.5% in those who did not. Maximal CK value and LVEF were not different inthe two groups. Kilip class Il or IV at any time during the hospital course was observed in 25.9 and 22.8%6, respectively (p = not significant) Multvariare ‘analysis using the first model (Le. excluding the inital variables chit might have been influenced by sulfonylurea therapy) showed that older age, history of renal insufficiency, and ST-elevation infarction were independent predictors of increased in- hospital mortality; in contrast, use of sulfonylureas was associated with a decreased risk (odds ratio: 0.44; 95% confidence interval: 0.23-0.84, p = 0.012). When the second model was used, higher Killip clas on admission, heart rate >90 beats/min on admission, older age and history of renal insufficiency predicted higher in-hospital mortality, while SU was associated with a decreased tisk (odds ratio: 0.37; p= 0.020) (Table 4). In the 94 patients who underwent primary angioplasty in-hospital mortality was 7.1% in patients on SU, versus 25.0% in patients without SU (p < 0.05); in the 245 patients who Underwenta percutaneous coronary intervention (PCI) at any time during their initial hospital stay, mortality was also lower in patients who were on SU (5.2 ¥8 11.8%, p <007) Discussion Representativeness of the study population ‘The present study analyses the current management of myocardial infarction in the intensive care unit on the scale of a whole country, irrespective of the type of institution 10 which the patients were admitted. Participation was high in all types of institutions (academic, general hospitals, private clinics) and no les Sulfonyureaweatment Quartile Number of patients Median [08] score No slfonjurea treatment Number of patients Median (10R] score 21 (-246, -1 44} 1 4 1.82 ~226, -1.08) 103 2 54 0.34[-045, -0.26), 8 [042 [-044, -0.29) 3 65 0.10[-008,0271 6 0.08|- 0.02, 0.25] 4 8 0,710.53, 1011 2 0630.52, 0.86) Copyright © 2005 Joka wey & Sons, Lit Diabetes Meta Res Rey 2005; 21: 143-149,Sulfonylureas and Acute Myocardial Infarction Table 3, Maximal Ck values, left ventricular ej Variable nat 1460 + 1678 9 = 251) Maximal Ck value a) 54130) Reintarcton 708) Ventricular bilaton 16158) ‘Attia ibelaio | 28 (105) 2nd or 3rd degree AV Block 210.8) Kilip dss tor 701253) SR, fee wal ruptare 5i19) Stroke 80 Irchospital death 461163) Irchosptal stoke or death 48178) AV, atoventiular, Ck, creatine Kinase; LVEF, left rupture Table 4, Multivariate logistic regression analysis of predictors of Inshospital death. The first model did not include variables that might be affected by antidiabetic agents, Both models included 2 propensity score with four categories of increasing probability of sulfonylureas prescription Relative 95% confidence risk interval paalue Mode! ‘ge > 70 years 288 148-548 0.002 Renal insufficiency Ser t61-812 0.002 Sulforyurea therapy before 0.44 0.23-084 0.012, samsson STelevation 37) 128110016 Model 2 Heart rate on 398 186-850 aot admission > 90 bpm Klip Cass 0.001 ass 261 105-651 lassi 235 079-702 Gass wer 21327 [age > 70 years 310 136-705 0007 feral insufficiency 403 Tarte 0009 Suifongurea therapy before «03716-0866 0.020 aémision imbalance was seen from one region to another. It may therefore be considered highly representative of the practice in France by the end of 2000. Compared with previous surveys carried out in France over the last ten years, the proportion of diabetic patients admitted for ‘AMI was higher in 2000 (219%) than in 1995 (1796) [11] and similar to the findings of a survey carried out in 1998 in patients with acute coronary syndromes (21%) (12] Role of previous sulfonylurea therapy on in-hospital outcome To our knowledge, our study is the largest to date that addresses the question of the impact of previous treatment with SU on early mortality after acute myocardial infarction. It does not bring information, however, on the incidence of myocardial infarction in diabetic patients on cor without sulfonylurea therapy. Its main finding is that the in-hospital outcome of diabetic patients who were ‘on SU at the time their myocardial infarction occurred Copyright © 2005 John Wiley & Sons, Ld No sulfonjiureas| 7 jection fraction and in-hospital complications (on sufonyureas pralue pais 1202 + 1854 (p= 204) NS 240) a7 12 (n= 195) NS 367) Ns 5123) 0.052 na) NS 13161) NS 49128) NS 2(03) Ns 2109) NS 221102) 0035 2412) o.oas ventculareection fraction; VSR, ventricular septal was at least as good (and in fact even better) than that of diabetic patients who were not treated with SU. The favorable effect of SU therapy persisted even after adjustment by multivariate analyses (including the propensity score that adjusted for the imbalances between ‘groups at hospital admission, such as the previous use of statins or fibrates) and was present whether the patients received early reperfusion therapy (with thrombolysis as ‘well as with PCI) or not. In keeping with these findings, early complications were less or as frequent in patients on SU; in particular, ventricular fibrillation was less common in patients on SU. The initial treatment (first 48 h) was similar in the two groups, except for the use of SU, which ‘was obviously higher in the initial SU group, and insulin, which was higher in the non-SU group; the percentage of patients not on insulin at baseline and who were subsequently put on insulin during the first 48 h, however, was similar in the two groups (no SU group: 49.7% vs SU group: 48.5%), suggesting that glycemic control at the acute stage was similar in both groups. In addition, insulin was used at the stage of hospital admission in approximately half of the patients who were on SU, and the mortality did not differ from that in patients in whom insulin was not added. Several studies, beginning with the now ancient UGDP trial have suggested that diabetic patients receiving SU were at increased cardiovascular risk compared with diabetic patients who did not receive such medications [5]. Their conclusions have been strengthened by the DIGAMI randomized trial [13], which has documented ‘a much poorer outcome in diabetic patients who did rot receive insulin at the acute stage of myocardial infarction. In the specific setting of acute myocardial infarction treated with primary balloon angioplasty, and in contrast with our findings, Garratt er al. [14] have also found a more adverse outcome in 67 patients who ‘were on SU, compared with 118 who were not. More recently, however, several reports have challenged these conclusions: in the UKPDS study [7], treatment with SU ‘was not related to higher cardiovascular complications. Klamann et al. [15] found no increase in hospital mortality in 76 patients admitted for acute myocardial infarction ‘while on SU, compared with 89 diabetic patients without Diobre Metab Res Rev 2005; 21: 249-149.