Prognostic Value of the Acute-to-Chronic Glycemic Ratio at Admission in Acute Myocardial Infarction: A Prospective Study

Elevated levels of plasma glucose at hospital admission (acute hyperglycemia) are common among patients with acute myocardial infarction (AMI) (1,2). Acute hyperglycemia has been recognized as an independent determinant of adverse outcomes, both in patients with and in patients without diabetes (3,4). Acute hyperglycemia results in a prothrombotic state, modulates inflammatory response and oxidative stress, and is the cause of endothelial dysfunction and impaired microcirculatory function (5–7), leading to larger infarct size (8,9). These phenomena may explain the association between elevated plasma glucose and poor prognosis in AMI. Indeed, patients with acute hyperglycemia typically have a more complicated in-hospital clinical course, including a higher incidence of heart failure, cardiogenic shock, and death (1–9).

In AMI, patients with diabetes have a worse outcome than those without diabetes (10,11). However, acute hyperglycemia has been shown to be a powerful predictor of poor prognosis, particularly in patients without diabetes (12–14). This emphasizes the role of an acute rise of glucose level, as compared with a chronic elevation, in predisposing patients toward a worse prognosis. Chronic elevation of glucose levels cannot be determined in patients admitted with AMI, but it can be estimated by assessing the glycosylated hemoglobin (HbA_1c) value (15). Therefore, in AMI patients, the combined information provided by acute (measured at hospital admission) and chronic (estimated by HbA_1c) glycemic value assessment may be a better prognostic predictor than glycemic value at admission or diabetes status alone. Indeed, it represents the “true” acute glycemic increase. This may be particularly relevant in patients with diabetes, in whom elevated glucose levels at admission do not necessarily indicate the occurrence of acute hyperglycemia.

Thus, the purpose of our study was to investigate the possible association between the ratio of acute to chronic (A/C) glycemic values and in-hospital outcomes in an unselected cohort of consecutive AMI patients. In particular, we hypothesized that A/C glycemic ratio, as compared with admission glycemic value, is more closely associated with infarct size and the most clinically relevant hemodynamic consequences of AMI, such as acute pulmonary edema, cardiogenic shock, and death.

This was a prospective, observational study. We enrolled all consecutive patients with AMI, both ST-elevation myocardial infarction (STEMI) and non–ST-elevation myocardial infarction (NSTEMI), admitted to the Intensive Cardiac Care Unit of Centro Cardiologico Monzino in Milan between 1 June 2010 and 29 June 2016. Patients experiencing AMI as a complication of elective percutaneous coronary intervention (PCI) (type 4a AMI) and those with a history of hemoglobinopathy were excluded. The study complied with the Declaration of Helsinki, and the ethics committee of Centro Cardiologico Monzino approved the research protocol (no. R520-CCM549). Written informed consent was obtained from all participants. No extramural funding was used to support this work.

Blood glucose and HbA_1c levels were measured in all patients at hospital admission. A diagnosis of diabetes was made if this disease or antidiabetes treatment, including oral agents or insulin, was recorded in the admission history. A diagnosis of unknown diabetes was made when patients had ≥6.5% (48 mmol/mol) HbA_1c despite no previous history of the disease (16). These patients were considered to have diabetes. Acute hyperglycemia was defined as a blood glucose at admission >198 mg/dL (>11 mmol/L) according to the definition used in previous studies (17). As currently no uniform definition of acute hyperglycemia in the setting of AMI exists, we also considered a cutoff value of >144 mg/dL (18). Average chronic glucose levels were estimated by HbA_1c, expressed as percent value, according to the following formula (15):

The A/C glycemic ratio was calculated in all patients with measurement of blood glucose at admission and estimation of chronic glucose levels.

Study patients received standard medical treatment and coronary revascularization at the discretion of the attending physician based on the current standards of care recommended by published guidelines. In all patients with diabetes, antidiabetes medications were withheld at hospital admission. In patients with acute hyperglycemia (>198 mg/dL), insulin was administered, with a glucose level target range of 140–180 mg/dL (19).

Demographical, clinical, biochemical, echocardiographic, and angiographic data were obtained. Troponin I (Beckman Coulter, Fullerton, CA) was measured every 6 h from hospital admission to 24 h after it reached peak value. Left ventricular ejection fraction (LVEF) was measured with echocardiography in all patients within 24 h from hospital admission. The TIMI (Thrombolysis In Myocardial Infarction) risk score was calculated in STEMI and NSTEMI patients.

The primary end point of the study was the combination of in-hospital mortality, nonfatal acute pulmonary edema, and cardiogenic shock. Acute pulmonary edema was defined as severe respiratory distress, tachypnea, and orthopnea with rales over the lung fields and arterial oxygen saturation <90% on room air prior to treatment with oxygen. Cardiogenic shock was defined as prolonged hypotension (systolic blood pressure ≤85 mmHg) with evidence of decreased organ perfusion caused by severe left ventricular dysfunction, right ventricular infarction, or mechanical complications of infarction requiring intra-aortic balloon pump and/or inotropic agents. Infarct size, estimated by troponin I peak value, was the secondary end point of the study.

A sample size of 1,500 patients was calculated under the following assumptions: 10% overall incidence of the primary end point, with an expected 7% and 14% incidence in patients with the lowest and the highest A/C glycemic ratio tertile, respectively (odds ratio [OR] 2.16). This sample size allowed a 95% statistical power in assessing a significant difference (α error of 0.05) of the combined end point between the three A/C glycemic ratio tertiles.

Continuous variables are presented as mean ± SD. Variables with a skewed distribution are presented as median and interquartile ranges. Categorical data are presented as n (%). Trends across A/C glycemic ratio tertiles were assessed by ANCOVA and by Mantel-Haenszel χ² as appropriate. The association between A/C glycemic ratio tertiles and the primary end point and troponin I peak value (below or above the median value) was assessed by logistic regression analysis. Analyses were adjusted for the baseline risk profile of the patient, as assessed by the TIMI risk score, and for a model including independent predictors of both the primary end point and troponin I peak value, identified by performing a logistic regression analysis with stepwise selection of variables. Patients were grouped into tertiles according to A/C glycemic ratio levels, using the lowest tertile as a reference. Results are presented as OR with 95% CIs. To formally check whether the effects of acute glycemia and A/C glycemic ratio differed between patients with and without diabetes, we tested the appropriate interaction terms. The main analysis was repeated by using a Cox proportional hazards model, in order to take into account the exact event times, although most of our clinical end points occurred in the first few days of hospitalization. Hazard ratios and 95% CI for the primary end point associated with A/C glycemic ratio tertiles were then evaluated.

Receiver operating characteristic (ROC) curves were calculated, and the areas under the ROC curves (AUCs) with 95% CI were used to measure the ability of the considered variables to predict the primary end point. AUCs were compared as recommended by DeLong et al. (20). ROC curve was used to calculate a cutoff value for the A/C glycemic ratio that would enable balancing the sensitivity and specificity of predicting a case. This was done by choosing the value that minimized the Euclidean distance between the ROC curve and the point with coordinates 1 − specificity = 1 and sensitivity = 1 (top-left corner of the ROC graph).

Net reclassification improvement was used to identify the possible additional prognostic value of A/C glycemic ratio when added to acute glycemia. All tests were two tailed, and P < 0.05 was required for statistical significance. All analyses were performed using SAS, version 9.4 (SAS Institute, Cary, NC). Reclassification statistics were assessed with the SAS macros published by Cook and Ridker (21).

In total, 1,553 consecutive AMI (747 STEMI and 806 NSTEMI) patients (mean ± SD age 67 ± 13 years [1,150 men]) were included in the study. Of these, 417 (27%) had diabetes, while 233 (15%) and 583 (37%) had acute hyperglycemia (when a cutoff of 198 mg/dL or 144 mg/dL was considered, respectively). Overall, median acute glucose level was 131 mg/dL (interquartile range 110–165). Median acute glucose level was 192 mg/dL (146–262) and 123 mg/dL (107–147) in patients with and without diabetes, respectively (P < 0.001). Overall median estimated chronic glucose level was 115 mg/dL (107–132). Acute glucose level was 152 mg/dL (133–183) and 112 mg/dL (105–123) in patients with and without diabetes (P < 0.001).

Table 1 shows the baseline characteristics and outcomes of patients stratified according to A/C glycemic ratio tertiles. Patients in the highest tertile were older and more likely to have STEMI, diabetes, lower estimated glomerular filtration rate, and LVEF and higher TIMI risk score than patients in the lower tertiles. They also had a more complicated in-hospital clinical course, with significantly higher mortality. Figure 1 shows primary end point incidence and mean troponin I peak value in the three A/C glycemic ratio tertiles. Patients in the second and third tertiles had a significantly higher risk of experiencing the primary end point and developing a larger infarct size, even after adjustment for the baseline risk profile (TIMI risk score) and for major clinical confounders (Supplementary Table 1). A similar result was observed when a Cox model was built for the primary end point: adjusted hazard ratio for second versus first A/C glycemic ratio tertile 1.49 (95% CI 0.92–2.43) and for third versus first tertile 3.55 (2.34–5.41) (P = 0.10 and P < 0.0001, respectively).

Supplementary Table 2 shows AUC for acute glycemia and A/C glycemic ratio in predicting the primary end point in the entire population and in patients with and without diabetes. At reclassification analysis, the A/C glycemic ratio provided the best prognostic power compared with acute glycemia, particularly in patients with diabetes.

Figure 2 shows the OR for the primary end point of acute glycemia and A/C glycemic ratio tertiles in patients with and patients without diabetes.

In the entire population, the cutoff value of A/C glycemic ratio that maximized sensitivity and specificity for primary end point prediction was 1.3. Overall, 34% of patients had a ratio above the cutoff; the incidence of the primary end point was 22% and 7% (P < 0.001) in patients with an A/C glycemic ratio above and below the cutoff, respectively. Figure 3 shows the OR, adjusted for the TIMI risk score, for the primary end point of acute hyperglycemia (>144 mg/dL and >198 mg/dL) and A/C glycemic ratio above the cutoff value (≥1.3) in the entire population and in patients with and patients without diabetes.

The primary finding of the study is that the ability of glycemia at admission to predict in-hospital mortality, morbidity, and infarct size in AMI patients significantly improves when the average chronic glucose level, as estimated by HbA_1c, is taken into account. The prognostic power of the A/C glycemic ratio is particularly evident in patients with diabetes, in whom a high glucose value at admission is not always an index of an acute glycemic rise. Finally, we identified a cutoff value (≥1.3) for the A/C glycemic ratio that is able to discriminate patients at high risk.

Acute hyperglycemia is frequently observed in the early phase of AMI, irrespective of diabetes presence, and it has been constantly associated with a poor outcome and a larger infarct size (1–9). The impact of acute hyperglycemia seems to be more pronounced in patients without diabetes than in those with diabetes, suggesting that the magnitude of the acute glycemic rise from chronic levels, rather than the absolute admission glycemic level per se, can be detrimental (12–14). From a practical point of view, AMI patients with similar acute hyperglycemia may have different risk profiles according to chronic glycemic values. Therefore, we hypothesized that the assessment of the A/C glycemic ratio could better identify true stress hyperglycemia than the glycemic value measured at hospital admission. In order to estimate the average chronic glycemia, we used the formula proposed by Nathan et al. (15).

To the best of our knowledge, this is one of the first studies exploring the A/C glycemic ratio in AMI patients. While the prognostic impact of glycemia at admission has been widely evaluated in AMI, the clinical relevance of the A/C glycemic ratio has never been fully investigated. A very recent study by Yang et al. (22) analyzed the parameter of relative hyperglycemia (defined in their study as stress hyperglycemia ratio) in a large registry of patients undergoing PCI. They found that this ratio is a strong predictor of short-term and long-term major adverse cardiovascular and cerebrovascular events. Differently from our study, they included all spectrums of coronary artery disease, only 30% of which involved AMI. Moreover, owing to the retrospective design of the study, their index was based not on admission glycemia but, rather, on the first-measured random glycemia during hospitalization. Another study, by Fujino et al. (23), considered the possible additional role of acute and chronic hyperglycemia in AMI and showed that patients with acute hyperglycemia had worse in-hospital outcome. However, in patients with chronic hyperglycemia who showed acute hyperglycemia at admission, mortality was significantly lower. A limitation of this study was that chronic glycemia was defined in a dichotomic way, according to HbA_1c value (<6.5% or ≥6.5% [< or ≥48 mmol/mol]) indicating diabetes status. Thus, they could not detect acute glycemic rise, i.e., the occurrence of true stress hyperglycemia, and were not able to evaluate its magnitude.

It is unclear whether an acute rise of glucose level directly contributes to myocardial injury, thus affecting patient outcome, or is only a marker of disease severity. Although no causal link can be inferred from our data, the relationships between acute glycemic rise, troponin I peak increase, and worse clinical outcome remained significant after adjustment for major clinical confounders. In agreement with this hypothesis, experimental evidence and clinical evidence have shown that an acute increase of plasma glucose triggers oxidative stress, inflammation, and endothelial dysfunction; activates coagulation; and abolishes ischemic preconditioning (5–9). All these factors may further increase myocardial damage in the setting of acute ischemia. Indeed, acute hyperglycemia has been associated with a lower myocardial salvage index evaluated by cardiac MRI (24). Interestingly, this association was not found in AMI patients with acute hyperglycemia and diabetes. This may be due to the ≥180 mg/dL glycemic threshold that was used to define acute hyperglycemia, a value that seems low in patients with diabetes with chronically elevated glycemic levels.

In our study, the prognostic power of the A/C glycemic ratio was particularly robust in patients with diabetes and, to a lesser extent, in patients with unknown diabetes, for whom it allowed reclassification properly of ∼30% and ∼10%, respectively. Conversely, in patients without diabetes and in patients with prediabetes, A/C glycemic ratio and acute glycemia had a similar prognostic accuracy. These findings are not unexpected, as the magnitude of acute glycemic elevation may be small in patients with known and unknown diabetes and an impaired chronic glyco-metabolic profile. In these patients, the A/C glycemic ratio may better identify the presence of a true stress hyperglycemia. Consistent with this idea, Roberts et al. (25) recently found that relative hyperglycemia, defined as admission glucose divided by estimated average glucose, was more strongly associated with critical illness than absolute hyperglycemia in patients acutely admitted to medical or surgical services.

The strengths of the current study include its large sample size, the prospective design, a well-characterized population, adjustment for a variety of risk factors, and a special focus on infarct size. Some limitations warrant mention. Firstly, we evaluated an AMI population admitted to a single center and treated, in most cases, with PCI. As this therapeutic strategy may have influenced the results of our study, the overall applicability of our findings to AMI patients not undergoing coronary revascularization needs to be clarified. Secondly, because this was an observational study, a cause-effect relationship between plasma glucose and outcomes cannot be established. Thirdly, impact on outcomes of in-hospital glycemic fluctuations, therapeutic management of acute hyperglycemia, glycemic target choice, and diabetes type (1 vs. 2) was not investigated and should be taken into account as a possible bias. Finally, the A/C glycemic ratio was calculated on the average chronic glycemic value estimated from HbA_1c. Thus, we cannot exclude that the calculated ratio does not fully reflect acute glycemic changes occurring during the index event. Furthermore, in patients with low admission hemoglobin value, average chronic glucose level might have been underestimated.

Our study may have some potential clinical implications. In AMI patients without diabetes, high glucose levels at admission reflect stress hyperglycemia and may be used to guide intensive glycemic control. Conversely, in patients with diabetes and high glucose levels at admission, assessment of A/C glycemic ratio may identify true stress hyperglycemia and may help physicians to better discriminate high-risk from low-risk AMI patients and to tailor treatment. Kosiborod et al. (26) have shown that glucose normalization after admission is associated with better survival in hyperglycemic patients hospitalized with AMI. However, in patients undergoing cardiothoracic surgery, intensive glycemic control improved outcomes in patients without known diabetes but not in those with a previous diagnosis of diabetes (27). Thus, the role of a tight control of hyperglycemia as a strategy for improving prognosis in AMI patients, and in particular in those with diabetes, is still under debate. Patients with diabetes often show high glycemic levels at admission that are not always associated with acute hyperglycemia. In these patients, an intensive lowering of glucose levels may not be beneficial, as the detrimental effects of the glycemic disorder are not limited to stress hyperglycemia but, rather, also include fluctuations of glycemic values, with acute glucose changes in both directions (28). Accordingly, previous studies have indicated that glucose variability in patients with diabetes has a more pronounced effect on oxidative stress (28), platelet activation, and aggregation (29), and on macrovascular and microvascular complications (30), than chronically elevated glucose levels. Moreover, acute variability of glucose values, assessed by measuring the mean amplitude of glycemic excursion with a continuous glucose monitoring system, negatively correlated with the myocardial salvage index in AMI (31). Therefore, future multicenter studies are needed to confirm our results and to investigate whether a strategy based on glucose normalization in patients with a high A/C glycemic ratio may have a greater impact on infarct size reduction and outcome improvement than an approach centered on the treatment of hyperglycemia at admission only.

In conclusion, we have demonstrated that the A/C glycemic ratio in AMI patients is closely associated with in-hospital morbidity and mortality. Use of the A/C glycemic ratio may be particularly valuable in patients with diabetes with chronically elevated glycemic levels because it may identify true stress hyperglycemia, which has been associated with larger infarct size and worse in-hospital outcome.

Acknowledgments. The authors acknowledge Michela Palmieri, Centro Cardiologico Monzino (Milan, Italy) for her precious help in revising the manuscript.

Funding. This study was funded by Centro Cardiologico Monzino, Istituto di Ricovero e Cura a Carattere Scientifico.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. G.M. and N.C. contributed to the study concept and design. V.M., M.D.M., M.C., S.M., M.R., J.C., M.M., I.M., M.G., and G.L. acquired data. G.M., N.C., F.V., R.M., and A.L.B. analyzed and interpreted data. G.M., N.C., and A.L.B. drafted the manuscript. G.M., M.D.M., M.C., S.M., M.R., J.C., M.M., I.M., M.G., G.L., R.M., and A.L.B. critically revised the manuscript for important intellectual content. A.B. and F.V. performed statistical analysis. M.C. and S.M. provided administrative, technical, or material support. G.M., N.C., and A.L.B. supervised the study. G.M. and A.L.B. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.