Diabetes Care Volume 41, April 2018 847

Giancarlo Marenzi,1 Nicola Cosentino,1 Prognostic Value of the Acute-to- Valentina Milazzo,1 Monica De Metrio,1 Milena Cecere,1 Susanna Mosca,1 Chronic Glycemic Ratio at Mara Rubino,1 Jeness Campodonico,1 Marco Moltrasio,1 Ivana Marana,1 Admission in Acute Myocardial Marco Grazi,1 Gianfranco Lauri,1 Alice Bonomi,1 Fabrizio Veglia,1 Infarction: A Prospective Study Roberto Manfrini,1 and 1,2 Care 2018;41:847–853 | https://doi.org/10.2337/dc17-1732 Antonio L. Bartorelli

OBJECTIVE Acute hyperglycemia is a powerful predictor of poor prognosis in acute myocardial infarction (AMI), particularly in patients without diabetes. This emphasizes the im- portance of an acute glycemic rise rather than glycemia level at admission alone. We investigated in AMI whether the combined evaluation of acute and chronic glycemic levels, as compared with admission glycemia alone, may have a better prognostic value.

RESEARCH DESIGN AND METHODS We prospectively measured admission glycemia and estimated average chronic levels (mg/dL) by the following formula: [(28.7 3 glycosylated hemoglobin %) 2 46.7], and calculated the acute-to-chronic (A/C) glycemic ratio in 1,553 consecutive AMI patients (mean 6 SD age 67 6 13 years). The primary end point was the combination of in-hospital mortality, acute pulmonary edema, and cardiogenic .

RESULTS RISK METABOLIC AND CARDIOVASCULAR The primary end point rate increased in parallel with A/C glycemic ratio tertiles (5%, 8%, and 20%, respectively; P for trend <0.0001). A parallel increase was observed in troponin I peak value (15 6 34 ng/mL, 34 6 66 ng/mL, and 68 6 131 ng/mL; P < 0.0001). At multivariable analysis, A/C glycemic ratio remained an independent pre- dictor of the primary end point and of troponin I peak value, even after adjustment for major confounders. At reclassification analyses, A/C glycemic ratio showed the best prognostic power in predicting the primary end point as compared with glycemia at 1Centro Cardiologico Monzino, Istituto di Rico- admission in the entire population (net reclassification improvement 12% [95% CI 4–20]; P = vero e Cura a Carattere Scientifico, Milan, Italy 2 0.003) and, particularly, in patients with diabetes (27% [95% CI 14–40]; P < 0.0001). Department of Biomedical and Clinical Sciences “Luigi Sacco,” University of Milan, Milan, Italy CONCLUSIONS Corresponding author: Giancarlo Marenzi, giancarlo In AMI patients with diabetes, A/C glycemic ratio is a better predictor of in-hospital [email protected]. morbidity and mortality than glycemia at admission. Received 18 August 2017 and accepted 19 December 2017. This article contains Supplementary Data online Elevated levels of plasma glucose at hospital admission (acute hyperglycemia) are at http://care.diabetesjournals.org/lookup/ common among patients with acute myocardial infarction (AMI) (1,2). Acute hyper- suppl/doi:10.2337/dc17-1732/-/DC1. glycemia has been recognized as an independent determinant of adverse outcomes, © 2018 by the American Diabetes Association. both in patients with and in patients without diabetes (3,4). Acute hyperglycemia Readers may use this article as long as the work fl is properly cited, the use is educational and not results in a prothrombotic state, modulates in ammatory response and oxidative for profit, and the work is not altered. More infor- , and is the cause of endothelial dysfunction and impaired microcirculatory func- mation is available at http://www.diabetesjournals tion (5–7), leading to larger infarct size (8,9). These phenomena may explain the .org/content/license. 848 Acute Hyperglycemia and AMI Diabetes Care Volume 41, April 2018

association between elevated plasma glu- with a history of hemoglobinopathy were with echocardiography in all patients cose and poor prognosis in AMI. Indeed, excluded. The study complied with the within 24 h from hospital admission. patients with acute hyperglycemia typi- Declaration of Helsinki, and the ethics The TIMI (Thrombolysis In Myocardial In- cally have a more complicated in-hospital committee of Centro Cardiologico Monzino farction) risk score was calculated in clinical course, including a higher incidence approved the research protocol (no. STEMI and NSTEMI patients. of heart failure, , and R520-CCM549). Written informed con- death (1–9). sent was obtained from all participants. Study End Points In AMI, patients with diabetes have a No extramural funding was used to sup- The primary end point of the study was worse outcome than those without dia- port this work. the combination of in-hospital mortality, betes (10,11). However, acute hypergly- nonfatal acute pulmonary edema, and cemia has been shown to be a powerful Study Protocol cardiogenic shock. Acute pulmonary predictor of poor prognosis, particularly Blood glucose and HbA1c levels were edema was defined as severe respiratory in patients without diabetes (12–14). measured in all patients at hospital ad- distress, tachypnea, and orthopnea with This emphasizes the role of an acute rise mission. A diagnosis of diabetes was rales over the lung fields and arterial ox- of glucose level, as compared with a made if this disease or antidiabetes treat- ygen saturation ,90% on room air prior chronic elevation, in predisposing patients ment, including oral agents or insulin, was to treatment with oxygen. Cardiogenic toward a worse prognosis. Chronic eleva- recorded in the admission history. A di- shock was defined as prolonged hypoten- tion of glucose levels cannot be deter- agnosis of unknown diabetes was made sion (systolic blood pressure #85 mmHg) minedinpatientsadmittedwithAMI, when patients had $6.5% (48 mmol/mol) with evidence of decreased organ perfu- but it can be estimated by assessing the HbA1c despite no previous history of the sion caused by severe left ventricular dys- glycosylated hemoglobin (HbA1c)value disease (16). These patients were consid- function, right ventricular infarction, or (15). Therefore, in AMI patients, the com- ered to have diabetes. Acute hyperglyce- mechanical complications of infarction re- bined information provided by acute mia was defined as a blood glucose at quiring intra-aortic balloon pump and/or (measured at hospital admission) and admission .198 mg/dL (.11 mmol/L) inotropic agents. Infarct size, estimated chronic (estimated by HbA1c) glycemic according to the definition used in previ- by troponin I peak value, was the second- value assessment may be a better prog- ous studies (17). As currently no uniform ary end point of the study. nostic predictor than glycemic value at definition of acute hyperglycemia in admission or diabetes status alone. Indeed, the setting of AMI exists, we also con- Statistical Analysis it represents the “true” acute glycemic in- sidered a cutoff value of .144 mg/dL A sample size of 1,500 patients was cal- crease. This may be particularly relevant (18). Average chronic glucose levels culated under the following assumptions: in patients with diabetes, in whom ele- 10% overall incidence of the primary end were estimated by HbA1c, expressed as vated glucose levels at admission do not percent value, according to the following point, with an expected 7% and 14% in- necessarily indicate the occurrence of formula (15): cidence in patients with the lowest and acute hyperglycemia. the highest A/C glycemic ratio tertile, re- Thus, the purpose of our study was to Estimated chronic glucose levels ðmg=dLÞ spectively (odds ratio [OR] 2.16). This investigate the possible association be- sample size allowed a 95% statistical 5½ð28:7 3 HbA1c %Þ 2 46:7 tween the ratio of acute to chronic (A/C) power in assessing a significant difference glycemic values and in-hospital outcomes (a error of 0.05) of the combined end in an unselected cohort of consecutive The A/C glycemic ratio was calculated in point between the three A/C glycemic ra- AMI patients. In particular, we hypothe- all patients with measurement of blood tio tertiles. sized that A/C glycemic ratio, as compared glucose at admission and estimation of Continuous variables are presented as with admission glycemic value, is more chronic glucose levels. mean 6 SD. Variables with a skewed dis- closelyassociatedwithinfarctsizeandthe Study patients received standard med- tribution are presented as median and most clinically relevant hemodynamic con- ical treatment and coronary revasculari- interquartile ranges. Categorical data are sequences of AMI, such as acute pulmonary zation at the discretion of the attending presented as n (%). Trends across A/C edema, cardiogenic shock, and death. physician based on the current standards glycemic ratio tertiles were assessed by of care recommended by published ANCOVA and by Mantel-Haenszel x2 as guidelines. In all patients with diabetes, appropriate. The association between A/C RESEARCH DESIGN AND METHODS antidiabetes medications were withheld glycemic ratio tertiles and the primary This was a prospective, observational study. at hospital admission. In patients with end point and troponin I peak value (be- We enrolled all consecutive patients with acute hyperglycemia (.198 mg/dL), insu- low or above the median value) was as- AMI, both ST-elevation myocardial infarc- lin was administered, with a glucose level sessed by logistic regression analysis. tion (STEMI) and non–ST-elevation myo- target range of 140–180 mg/dL (19). Analyses were adjusted for the baseline cardial infarction (NSTEMI), admitted Demographical, clinical, biochemical, risk profile of the patient, as assessed by to the Intensive Cardiac Care Unit of echocardiographic, and angiographic the TIMI risk score, and for a model in- Centro Cardiologico Monzino in Milan data were obtained. Troponin I (Beckman cluding independent predictors of both between 1 June 2010 and 29 June 2016. Coulter, Fullerton, CA) was measured ev- the primary end point and troponin I peak Patients experiencing AMI as a complica- ery 6 h from hospital admission to 24 h value, identified by performing a logistic tion of elective percutaneous coronary in- after it reached peak value. Left ventricu- regression analysis with stepwise selection tervention (PCI) (type 4a AMI) and those lar ejection fraction (LVEF) was measured of variables. Patients were grouped into care.diabetesjournals.org Marenzi and Associates 849

tertiles according to A/C glycemic ratio lev- 115 mg/dL (107–132). Acute glucose level CONCLUSIONS els, using the lowest tertile as a reference. was 152 mg/dL (133–183) and 112 mg/dL The primary finding of the study is that Results are presented as OR with 95% CIs. (105–123) in patients with and without the ability of glycemia at admission to To formally check whether the effects of diabetes (P , 0.001). predict in-hospital mortality, morbidity, acute glycemia and A/C glycemic ratio dif- Table 1 shows the baseline character- and infarct size in AMI patients signifi- fered between patients with and without istics and outcomes of patients stratified cantly improves when the average chronic diabetes, we tested the appropriate inter- according to A/C glycemic ratio tertiles. glucose level, as estimated by HbA1c,is action terms. The main analysis was re- Patients in the highest tertile were older taken into account. The prognostic power peated by using a Cox proportional and more likely to have STEMI, diabetes, of the A/C glycemic ratio is particularly fi hazards model, in order to take into ac- lower estimated glomerular ltration rate, evident in patients with diabetes, in count the exact event times, although and LVEF and higher TIMI risk score than whom a high glucose value at admission most of our clinical end points occurred patients in the lower tertiles. They also is not always an index of an acute glycemic fi in the rst few days of hospitalization. Haz- had a more complicated in-hospital clinical rise. Finally, we identified a cutoff value fi ard ratios and 95% CI for the primary end course, with signi cantly higher mortality. ($1.3) for the A/C glycemic ratio that is point associated with A/C glycemic ratio Figure 1 shows primary end point inci- able to discriminate patients at high risk. tertiles were then evaluated. dence and mean troponin I peak value Acute hyperglycemia is frequently ob- Receiver operating characteristic (ROC) in the three A/C glycemic ratio tertiles. served in the early phase of AMI, irrespec- curves were calculated, and the areas un- Patients in the second and third tertiles tive of diabetes presence, and it has been fi der the ROC curves (AUCs) with 95% CI had a signi cantly higher risk of experienc- constantly associated with a poor out- were used to measure the ability of the ing the primary end point and developing a come and a larger infarct size (1–9). The considered variables to predict the pri- larger infarct size, even after adjustment impact of acute hyperglycemia seems to mary end point. AUCs were compared as for the baseline risk profile (TIMI risk score) be more pronounced in patients without recommended by DeLong et al. (20). ROC and for major clinical confounders (Sup- diabetes than in those with diabetes, sug- curve was used to calculate a cutoff value plementary Table 1). A similar result was gesting that the magnitude of the acute for the A/C glycemic ratio that would en- observed when a Cox model was built for glycemic rise from chronic levels, rather able balancing the sensitivity and spec- the primary end point: adjusted hazard than the absolute admission glycemic ificity of predicting a case. This was ratio for second versus first A/C glyce- level per se, can be detrimental (12–14). done by choosing the value that mini- mic ratio tertile 1.49 (95% CI 0.92–2.43) From a practical point of view, AMI pa- mized the Euclidean distance between and for third versus first tertile 3.55 tients with similar acute hyperglycemia the ROC curve and the point with coor- (2.34–5.41) (P =0.10andP , 0.0001, may have different risk profiles according dinates 1 2 specificity= 1andsensitivity= respectively). 1 (top-left corner of the ROC graph). Supplementary Table 2 shows AUC for to chronic glycemic values. Therefore, we Net reclassification improvement was acute glycemia and A/C glycemic ratio in hypothesized that the assessment of the used to identify the possible additional predicting the primary end point in the A/C glycemic ratio could better identify prognostic value of A/C glycemic ratio entire population and in patients with true stress hyperglycemia than the glyce- when added to acute glycemia. All tests and without diabetes. At reclassification mic value measured at hospital admis- were two tailed, and P , 0.05 was re- analysis, the A/C glycemic ratio provided sion. In order to estimate the average quired for statistical significance. All the best prognostic power compared chronic glycemia, we used the formula analyses were performed using SAS, with acute glycemia, particularly in pa- proposed by Nathan et al. (15). version 9.4 (SAS Institute, Cary, NC). Re- tients with diabetes. To the best of our knowledge, this is fi classification statistics were assessed Figure 2 shows the OR for the primary one of the rst studies exploring the A/C with the SAS macros published by Cook end point of acute glycemia and A/C gly- glycemic ratio in AMI patients. While the and Ridker (21). cemic ratio tertiles in patients with and prognostic impact of glycemia at admis- patients without diabetes. sion has been widely evaluated in AMI, RESULTS In the entire population, the cutoff the clinical relevance of the A/C glycemic In total, 1,553 consecutive AMI (747 value of A/C glycemic ratio that maxi- ratio has never been fully investigated. A STEMI and 806 NSTEMI) patients (mean 6 mized sensitivity and specificity for pri- very recent study by Yang et al. (22) ana- SD age 67 6 13 years [1,150 men]) were mary end point prediction was 1.3. lyzed the parameter of relative hypergly- included in the study. Of these, 417 Overall, 34% of patients had a ratio above cemia (defined in their study as stress (27%) had diabetes, while 233 (15%) the cutoff; the incidence of the primary hyperglycemia ratio) in a large registry and 583 (37%) had acute hyperglycemia end point was 22% and 7% (P , 0.001) in of patients undergoing PCI. They found (when a cutoff of 198 mg/dL or 144 mg/dL patients with an A/C glycemic ratio above that this ratio is a strong predictor of was considered, respectively). Overall, and below the cutoff, respectively. Figure 3 short-term and long-term major ad- median acute glucose level was 131 mg/dL shows the OR, adjusted for the TIMI verse cardiovascular and cerebrovascular (interquartile range 110–165). Median risk score, for the primary end point of events. Differently from our study, they acute glucose level was 192 mg/dL acute hyperglycemia (.144 mg/dL and included all spectrums of coronary artery (146–262) and 123 mg/dL (107–147) in .198 mg/dL) and A/C glycemic ratio disease, only 30% of which involved AMI. patients with and without diabetes, re- above the cutoff value ($1.3) in the en- Moreover, owing to the retrospective de- spectively (P , 0.001). Overall median tire population and in patients with and sign of the study, their index was based estimated chronic glucose level was patients without diabetes. not on admission glycemia but, rather, on 850 Acute Hyperglycemia and AMI Diabetes Care Volume 41, April 2018

Table 1—Baseline characteristics and outcomes of the study patients according to A/C glycemic ratio tertiles A/C glycemic ratio tertile First (0.30–1.00) Second (1.01–1.25) Third (1.26–4.57) P* n 518 517 518 Age (years) 66 6 12 67 6 13 68 6 11 0.02 Male sex, n (%) 383 (74) 398 (77) 369 (71) 0.32 Body weight (kg) 76 6 14 77 6 15 75 6 15 0.32 Height (cm) 170 6 91706 81686 90.07 STEMI, n (%) 166 (32) 255 (49) 326 (63) ,0.0001 Diabetes, n (%) 116 (22) 110 (21) 191 (37) ,0.0001 Hypertension, n (%) 342 (66) 334 (65) 318 (61) 0.12 Smokers, n (%) 294 (57) 283 (55) 245 (47) 0.001 Hyperlipidemia, n (%) 281 (54) 248 (48) 249 (48) 0.04 Prior MI, n (%) 132 (25) 126 (24) 129 (25) 0.92 Prior CABG, n (%) 70 (13) 64 (12) 58 (11) 0.53 Prior PCI, n (%) 142 (27) 122 (24) 124 (24) 0.29 LVEF (%) 53 6 10 52 6 10 48 6 13 ,0.0001 Culprit vessel (LAD), n (%) 199 (38) 177 (34) 198 (37) 0.55 TIMI risk score 3.6 6 1.5 3.8 6 1.4 4.0 6 1.4 ,0.0001 Laboratory values at hospital admission Blood glucose (mg/dL) 106 (97–118) 127 (115–140) 179 (152–230) ,0.0001

HbA1c (mmol/mol) 45 6 13 42 6 11 45 6 14 0.45 HbA1c (%) 6.1 6 1.2 5.9 6 1.0 6.1 6 1.3 0.45 Serum creatinine (mg/dL) 0.88 (0.75–1.07) 0.92 (0.79–1.10) 0.98 (0.82–1.2) ,0.0001 eGFR (mL/min/1.73 m2)856 28 80 6 27 75 6 57 ,0.0001 Hemoglobin (g/dL) 13.7 6 1.6 13.8 6 1.7 13.5 6 1.9 0.07 Troponin I (ng/mL) 0.49 (0.09–2.65) 0.60 (0.22–3.09) 0.62 (0.11–2.42) 0.03 Medication before MI, n (%) Statins 199 (39) 161 (31) 172 (34) 0.09 ACE inhibitors/ARB 225 (43) 198 (38) 190 (37) 0.03 b-Blockers 198 (38) 182 (35) 163 (32) 0.005 Aspirin 195 (38) 171 (33) 182 (35) 0.44 In-hospital procedures PCI, n (%) 419 (81) 416 (80) 427 (82) 0.78 CABG, n (%) 27 (5) 21 (4) 18 (4) 0.51 In-hospital complications Death, n (%) 3 (0.6) 8 (1.5) 17 (3) 0.001 APE, n (%) 25 (5) 30 (6) 85 (16) ,0.0001 Cardiogenic shock, n (%) 7 (1.3) 18 (3) 46 (9) ,0.0001 MV, n (%) 7 (1.3) 6 (1.2) 41 (8) ,0.0001 Atrial fibrillation, n (%) 47 (9) 57 (11) 91 (18) 0.001 VT/VF, n (%) 7 (1.3) 19 (4) 74 (14) ,0.0001 AV block, n (%) 15 (3) 12 (2) 25 (5) 0.08 Blood transfusions, n (%) 15 (3) 14 (3) 28 (5) 0.03 AKI requiring RRT, n (%) 4 (0.8) 3 (0.6) 15 (3) 0.004 CCU LOS (days) 4 6 246 256 3 ,0.0001 Data are mean 6 SD or median (interquartile range) unless otherwise indicated. Hypertension was defined as previous treatment with antihypertensive drugs. Hyperlipidemia was defined as serum LDL cholesterol $140 mg/dL, triglycerides $150 mg/dL, HDL cholesterol ,40 mg/dL, or treatment with lipid- lowering drugs. AKI, ; APE, acute pulmonary edema; ARB, angiotensin II receptor blocker; AV, atrioventricular; CABG, coronary artery bypass graft; CCU LOS, length of stay; eGFR, estimated glomerular filtration rate; LAD, left anterior descending coronary artery; MI, myocardial infarction; MV, ; RRT, renal replacement therapy; VT/VF, ventricular tachycardia/ventricular fibrillation. *P for trend.

the first-measured random glycemia dur- mortality was significantly lower. A limita- It is unclear whether an acute rise of ing hospitalization. Another study, by tion of this study was that chronic glycemia glucose level directly contributes to myo- Fujino et al. (23), considered the possible was defined in a dichotomic way, accord- cardial injury, thus affecting patient out- additional role of acute and chronic hy- ing to HbA1c value (,6.5% or $6.5% come, or is only a marker of disease perglycemia in AMI and showed that pa- [, or $48 mmol/mol]) indicating diabe- severity. Although no causal link can be in- tients with acute hyperglycemia had tes status. Thus, they could not detect ferred from our data, the relationships be- worse in-hospital outcome. However, in acute glycemic rise, i.e., the occurrence tween acute glycemic rise, troponin I peak patients with chronic hyperglycemia who of true stress hyperglycemia, and were increase, and worse clinical outcome re- showed acute hyperglycemia at admission, not able to evaluate its magnitude. mained significant after adjustment for care.diabetesjournals.org Marenzi and Associates 851

Figure 1—Incidence of the combined end point (in-hospital mortality, acute pulmonary edema, and cardiogenic shock) and troponin I (cTnI) peak value in the study population stratified according to the A/C glycemic ratio tertiles. major clinical confounders. In agreement glycemic threshold that was used to de- elevation may be small in patients with with this hypothesis, experimental evi- fine acute hyperglycemia, a value that known and unknown diabetes and an im- dence and clinical evidence have shown seems low in patients with diabetes paired chronic glyco-metabolic profile. In that an acute increase of plasma glucose with chronically elevated glycemic levels. these patients, the A/C glycemic ratio triggers oxidative stress, inflammation, In our study, the prognostic power of may better identify the presence of a and endothelial dysfunction; activates co- the A/C glycemic ratio was particularly true stress hyperglycemia. Consistent agulation; and abolishes ischemic pre- robust in patients with diabetes and, with this idea, Roberts et al. (25) recently conditioning (5–9). All these factors may to a lesser extent, in patients with un- found that relative hyperglycemia, de- further increase myocardial damage in the known diabetes, for whom it allowed re- fined as admission glucose divided by es- setting of acute ischemia. Indeed, acute hy- classification properly of ;30% and ;10%, timated average glucose, was more perglycemia has been associated with a respectively. Conversely, in patients with- strongly associated with critical illness lower myocardial salvage index evaluated out diabetes and in patients with predia- than absolute hyperglycemia in patients by cardiac MRI (24). Interestingly, this as- betes, A/C glycemic ratio and acute acutely admitted to medical or surgical sociation was not found in AMI patients glycemia had a similar prognostic accu- services. with acute hyperglycemia and diabetes. racy. These findings are not unexpected, The strengths of the current study in- This may be due to the $180 mg/dL as the magnitude of acute glycemic cludeitslargesamplesize,theprospective

Figure 2—OR (95% CI) for the primary end point (in-hospital mortality, acute pulmonary edema, and cardiogenic shock) grouped according to tertiles of acute glycemia (left panel) and of A/C glycemic ratio (right panel) in patients with and without diabetes. 852 Acute Hyperglycemia and AMI Diabetes Care Volume 41, April 2018

Figure 3—OR (95% CI) of the combined end point (in-hospital mortality, acute pulmonary edema, and cardiogenic shock) in patients with acute hyperglycemia (.144 mg/dL and .198 mg/dL) and an A/C glycemic ratio above the identified cutoff value ($1.3). Analyses were adjusted for the TIMI risk score.

design, a well-characterized population, be used to guide intensive glycemic con- activation, and aggregation (29), and on adjustment for a variety of risk factors, trol. Conversely, in patients with diabetes macrovascular and microvascular compli- and a special focus on infarct size. Some and high glucose levels at admission, as- cations (30), than chronically elevated limitations warrant mention. Firstly, we sessment of A/C glycemic ratio may iden- glucose levels. Moreover, acute variability evaluated an AMI population admitted tify true stress hyperglycemia and may of glucose values, assessed by measuring to a single center and treated, in most help physicians to better discriminate the mean amplitude of glycemic excur- cases, with PCI. As this therapeutic strat- high-risk from low-risk AMI patients and sion with a continuous glucose monitor- egy may have influenced the results of to tailor treatment. Kosiborod et al. (26) ing system, negatively correlated with the our study, the overall applicability of our have shown that glucose normalization myocardial salvage index in AMI (31). findings to AMI patients not undergoing after admission is associated with better Therefore, future multicenter studies are coronary revascularization needs to be clar- survival in hyperglycemic patients hospi- needed to confirm our results and to in- ified. Secondly, because this was an obser- talized with AMI. However, in patients vestigate whether a strategy based on glu- vational study, a cause-effect relationship undergoing cardiothoracic , inten- cose normalization in patients with a high between plasma glucose and outcomes sive glycemic control improved outcomes A/C glycemic ratio may have a greater im- cannot be established. Thirdly, impact on in patients without known diabetes but not pact on infarct size reduction and outcome outcomes of in-hospital glycemic fluctua- in those with a previous diagnosis of diabe- improvement than an approach centered tions, therapeutic management of acute tes (27). Thus, the role of a tight control of on the treatment of hyperglycemia at ad- hyperglycemia, glycemic target choice, hyperglycemia as a strategy for improving mission only. anddiabetestype(1vs.2)wasnotinvesti- prognosis in AMI patients, and in particu- In conclusion, we have demonstrated gated and should be taken into account lar in those with diabetes, is still under that the A/C glycemic ratio in AMI patients as a possible bias. Finally, the A/C glycemic debate. Patients with diabetes often is closely associated with in-hospital mor- ratio was calculated on the average chronic show high glycemic levels at admission bidity and mortality. Use of the A/C glyce- glycemic value estimated from HbA1c. that are not always associated with acute mic ratio may be particularly valuable in Thus, we cannot exclude that the calcu- hyperglycemia. In these patients, an inten- patients with diabetes with chronically el- lated ratio does not fully reflect acute gly- sive lowering of glucose levels may not be evated glycemic levels because it may cemic changes occurring during the index beneficial, as the detrimental effects of the identify true stress hyperglycemia, which event. Furthermore, in patients with low glycemic disorder are not limited to stress has been associated with larger infarct size admission hemoglobin value, average hyperglycemia but, rather, also include and worse in-hospital outcome. chronic glucose level might have been fluctuations of glycemic values, with acute underestimated. glucose changes in both directions (28). Our study may have some potential Accordingly, previous studies have indi- Acknowledgments. The authors acknowledge clinical implications. In AMI patients with- cated that glucose variability in patients Michela Palmieri, Centro Cardiologico Monzino out diabetes, high glucose levels at admis- with diabetes has a more pronounced ef- (Milan, Italy) for her precious help in revising the sion reflect stress hyperglycemia and may fect on oxidative stress (28), platelet manuscript. care.diabetesjournals.org Marenzi and Associates 853

Funding. This study was funded by Centro Car- 7. Williams SB, Goldfine AB, Timimi FK, et al. and American Diabetes Association consensus diologico Monzino, Istituto di Ricovero e Cura a Acute hyperglycemia attenuates endothelium- statement on inpatient glycemic control. Diabetes Carattere Scientifico. dependent vasodilation in humans in vivo. Circu- Care 2009;32:1119–1131 Duality of Interest. No potential conflicts of in- lation 1998;97:1695–1701 20. DeLong ER, DeLong DM, Clarke-Pearson DL. terest relevant to this article were reported. 8. Esposito K, Nappo F, Marfella R, et al. Inflam- Comparing the areas under two or more corre- Author Contributions. G.M. and N.C. contrib- matory cytokine concentrations are acutely in- lated receiver operating characteristic curves: utedtothestudyconceptanddesign.V.M.,M.D.M., creased by hyperglycemia in humans: role of a nonparametric approach. Biometrics 1988;44: M.C., S.M., M.R., J.C., M.M., I.M., M.G., and G.L. oxidative stress. Circulation 2002;106:2067–2072 837–845 acquired data. G.M., N.C., F.V., R.M., and A.L.B. 9. Jensen CJ, Eberle HC, Nassenstein K, et al. Im- 21. Cook NR, Ridker PM. Advances in measuring analyzed and interpreted data. G.M., N.C., and pact of hyperglycemia at admission in patients the effect of individual predictors of cardiovascu- A.L.B.draftedthemanuscript.G.M.,M.D.M.,M.C., with acute ST-segment elevation myocardial in- lar risk: the role of reclassification measures. Ann S.M., M.R., J.C., M.M., I.M., M.G., G.L., R.M., and farction as assessed by contrast-enhanced MRI. Intern Med 2009;150:795–802 A.L.B. critically revised the manuscript for impor- Clin Res Cardiol 2011;100:649–659 22. Yang Y, Kim TH, Yoon KH, et al. The stress tant intellectual content. A.B. and F.V. performed 10. Timmer JR, Hoekstra M, Nijsten MW, et al. hyperglycemia ratio, an index of relative hypergly- statistical analysis. M.C. and S.M. provided admin- Prognostic value of admission glycosylated hemo- cemia, as a predictor of clinical outcomes after istrative,technical,ormaterialsupport.G.M.,N.C., globin and glucose in nondiabetic patients with percutaneous coronary intervention. Int J Cardiol and A.L.B. supervised the study. G.M. and A.L.B. ST-segment-elevation myocardial infarction trea- 2017;241:57–63 are the guarantors of this work and, as such, had ted with percutaneous coronary intervention. Cir- 23. Fujino M, Ishihara M, Honda S, et al. Impact of full access to all the data in the study and take culation 2011;124:704–711 acute and chronic hyperglycemia on in-hospital responsibility for the integrity of the data and the 11. Mak KH, Moliterno DJ, Granger CB, et al. In- outcomes of patients with acute myocardial in- accuracy of the data analysis. fluence of diabetes mellitus on clinical outcome in the farction. Am J Cardiol 2014;114:1789–1793 thrombolytic era of acute myocardial infarction. GUSTO- 24. Teraguchi I, Imanishi T, Ozaki Y, et al. Impact I Investigators. Global Utilization of Streptokinase of stress hyperglycemia on myocardial salvage References and Tissue Plasminogen Activator for Occluded Cor- following successfully recanalized primary acute 1. Oswald GA, Corcoran S, Yudkin JS. Prevalence onary Arteries. J Am Coll Cardiol 1997;30:171–179 myocardial infarction. Circ J 2012;76:2690–2696 and risks of hyperglycaemia and undiagnosed di- 12. Umpierrez GE, Isaacs SD, Bazargan N, You X, 25. Roberts GW, Quinn SJ, Valentine N, et al. Rel- abetes in patients with acute myocardial infarc- Thaler LM, Kitabchi AE. Hyperglycemia: an inde- ative hyperglycemia, a marker of critical illness: tion. Lancet 1984;1:1264–1267 pendent marker of in-hospital mortality in pa- introducing the stress hyperglycemia ratio. J Clin 2. Ishihara M, Kagawa E, Inoue I, et al. Impact of tients with undiagnosed diabetes. J Clin Endocrinol Endocrinol Metab 2015;100:4490–4497 admission hyperglycemia and diabetes mellitus Metab 2002;87:978–982 26. Kosiborod M, Inzucchi SE, Krumholz HM, et al. on short- and long-term mortality after acute 13. Krinsley JS, Egi M, Kiss A, et al. Diabetes status Glucose normalization and outcomes in patients myocardial infarction in the coronary intervention and the relation of the three domains of glycemic with acute myocardial infarction. Arch Intern Med era. Am J Cardiol 2007;99:1674–1679 control to mortality in critically ill patients: an in- 2009;169:438–446 3. Eitel I, Hintze S, de Waha S, et al. Prognostic ternational multicenter cohort study. Crit Care 27. Umpierrez G, Cardona S, Pasquel F, et al. Ran- impact of hyperglycemia in nondiabetic and dia- 2013;7:R37 domized Controlled Trial of Intensive Versus Con- betic patients with ST-elevation myocardial infarc- 14. Egi M, Bellomo R, Stachowski E, et al. The servative Glucose Control in Patients Undergoing tion: insights from contrast-enhanced magnetic interaction of chronic and acute glycemia with Coronary Artery Bypass Graft Surgery: GLUCO-CABG resonance imaging. Circ Cardiovasc Imaging mortality in critically ill patients with diabetes. Trial. Diabetes Care 2015;38:1665–1672 2012;5:708–718 Crit Care Med 2011;39:105–111 28. Monnier L, Mas E, Ginet C, et al. Activation of 4. Planer D, Witzenbichler B, Guagliumi G, et al. 15. Nathan DM, Kuenen J, Borg R, et al. Trans- oxidative stress by acute glucose fluctuations Impact of hyperglycemia in patients with ST-segment lating the A1C assay into estimated average glu- compared with sustained chronic hyperglycemia elevation myocardial infarction undergoing percuta- cose values. Diabetes Care 2008;31:1473–1478 in patients with type 2 diabetes. JAMA 2006;295: neous coronary intervention: the HORIZONS-AMI 16. American Diabetes Association. Execu- 1681–1687 trial. Int J Cardiol 2013;167:2572–2579 tive summary: Standards of Medical Care in 29. Monnier LH, Lachkar H, Richard JL, et al. 5. Stranders I, Diamant M, van Gelder RE, et al. Diabetesd2014. Diabetes Care 2014;37(Suppl. Plasma beta-thromboglobulin response to insulin- Admission blood glucose level as risk indicator of 1):S5–S13 induced in type I diabetic patients. death after myocardial infarction in patients with 17. Marenzi G, De Metrio M, Rubino M, et al. Diabetes 1984;33:907–909 and without diabetes mellitus. Arch Intern Med Acute hyperglycemia and contrast-induced ne- 30. Gorst C, Kwok CS, Aslam S, et al. Long-term 2004;164:982–988 phropathy in primary percutaneous coronary in- glycemic variability and risk of adverse outcomes: 6. Worthley MI, Holmes AS, Willoughby SR, et al. tervention. Am Heart J 2010;160:1170–1177 a systematic review and meta-analysis. Diabetes The deleterious effects of hyperglycemia on plate- 18. Oliver MF. Stress hyperglycaemia and death Care 2015;38:2354–2369 let function in diabetic patients with acute co- after myocardial infarction. Lancet 2000;355: 31. Teraguchi I, Imanishi T, Ozaki Y, et al. Acute- ronary syndromes mediation by superoxide 1647–1648 phase glucose fluctuation is negatively correlated production, resolution with intensive insulin ad- 19. MoghissiES, Korytkowski MT, DiNardoM, et al. with myocardial salvage after acute myocardial ministration. J Am Coll Cardiol 2007;49:304–310 American Association of Clinical Endocrinologists infarction. Circ J 2014;78:170–179