Objective To assess the association between fasting plasma glucose (FPG) levels on admission and mortality in older patients with acute myocardial infarction (AMI), and compare the effects of FPG levels on outcomes in the context of contemporary treatments, including drug treatment, percutaneous coronary intervention and coronary artery bypass grafting.
Methods From April 2004 to October 2006, 1854 older (age ≥65 years) patients with AMI were enrolled in the Beijing Elderly Acute Myocardial Infarction Study (BEAMIS) consecutively. Patients were categorised into 4 groups: hypoglycaemia group (N=443, 23.9%), FPG≤5 mmol/l; euglycaemia group (N=812, 43.8%), FPG≥ 5.1 to≤7.0 mmol/l (5–7 mmol/l); mild hyperglycaemia group (N=308, 16.6%), FPG≥ 7.1 to≤9.0 mmol/l (7–9 mmol/l); and severe hyperglycaemia group (N=291, 15.7%), FPG≥9.1 mmol/l. The primary end point was in-hospital and 3-year all-cause mortality from the day of admission.
Results Compared with the euglycaemia group, hypoglycaemia or hyperglycaemia groups were all associated with higher in-hospital and 3-year all-cause mortality. There was a U-shaped relationship between admission FPG levels and short- and long-term all-cause mortality. This U-shaped relationship applied equally to subgroups in the context of contemporary treatments.
Conclusions In older patients with AMI, increased as well as decreased admission FPG levels could predict higher in-hospital and 3-year mortality. There was a striking U-shaped relationship between admission FPG levels and short- and long-term mortality. An initial admission FPG level ≥ 5.1 to≤7.0 mmol/l may be desirable because it was associated with better clinical outcomes.
- Fasting plasma glucose
- all-cause mortality
- acute myocardial infarction
Statistics from Altmetric.com
Although numerous studies have established an association between admission randomised glucose levels and poor outcomes in patients with acute myocardial infarction (AMI), less is known about the correlation between the initial fasting plasma glucose (FPG) on admission and major adverse cardiovascular events.1–7 Also, many gaps in knowledge remain in our understanding of the association between glucose levels and prognosis in patients with AMI.8 First, the optimal glucose value (or range of values) that should be considered normal on admission has not been established. Second, most previous studies did not directly compare the effects of glucose levels on outcomes in the context of contemporary treatments, including drug treatment, percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG). Third, evidence for an association between decreased glucose and clinical outcomes is scarce. Finally, although the elderly represent a growing majority of patients with AMI, few data are available about this patient population. The aim of this study was to assess whether initial FPG levels on admission were associated with all-cause mortality in older patients with AMI by analysing data from the Beijing Elderly Acute Myocardial Infarction Study (BEAMIS).
BEAMIS Study and patients population
BEAMIS is an observational multicentre study conducted at five medical centres. Detailed clinical data were abstracted from medical records by trained medical record reviewers. From April 2004 to October 2006, 2016 consecutive older (age ≥65 years) patients presenting with AMI were screened consecutively. All patients underwent coronary angiography. Patients with unstable angina pectoris (myocardial biomarkers ≤2 times the normal upper limit) were excluded. To enhance homogeneity and ensure examination of a representative cohort in the context of contemporary treatment modalities, 22 (1.1%) patients who underwent percutaneous transluminal coronary angioplasty without stent implantation and 135 (6.7%) patients who underwent bare-metal stent implantation were excluded. Five (0.2%) patients with a terminal illness were also excluded. A total of 1854 patients therefore constituted the study group, and FPG was measured the morning after admission. in all subjects. Patients were divided into three groups according to their treatments: drug treatment group (39.5%, including 71 thrombolysis), PCI group (47.5%, including 26 thrombolysis) and CABG group (13.0%, including seven thrombolysis). The drug treatment group included patients receiving contemporary drug treatment. For the PCI and CABG group, drugs (such as aspirin, clopidogrel, β blockers, angiotensin converting enzyme inhibitors, angiotensin receptor blockers, calcium channel blockers, nitrates, antidiabetic agents, statins, etc) were given according to American College of Cardiology/American Heart Association guidelines in addition to revascularisation. The study protocol was reviewed and approved by the ethics committee of each participating institution.
AMI was defined by a combination of two of the three following characteristics: chest pain consistent with ongoing myocardial ischaemia persisting for >30 min, ischaemic electrocardiographic changes and peak myocardial biomarker (including creatine kinase (CK)-MB and cardiac troponin I (cTnI)/cTnT) levels >2 times the corresponding normal upper limits.9 Patients were classified as having recognised diabetes mellitus (DM) if their medical records contained documentation of a previous history of DM, they were diagnosed with DM on admission, or were receiving an antidiabetic agent at the time of admission.
The primary end point was in-hospital and 3-year all-cause mortality from the day of admission.
All case record form data were entered into two Epidata 3.1 databases (the Epidata Association, Denmark) by different people. Analyses were conducted with the SPSS 16.0 software package (SPSS, Chicago, Illinois, USA). Baseline demographic and clinical characteristics were compared by the Pearson χ2 test for categorical variables and the analysis of variance test for continuous variables. The unadjusted association between groups of admission FPG and in-hospital and 3-year mortality was tested with the Pearson χ2 test. Cox proportional models were used to assess whether the association between admission FPG levels and in-hospital and 3-year mortality was independent of other characteristics. Variables clinically considered or previously demonstrated to be of prognostic importance10–12 and those identified in bivariate analyses as predictors of in-hospital and 3-year mortality were entered into the models. The following items were included as covariates: demographic factors (age, gender); medical history (previous myocardial infarction, hypertension, diabetes, stroke, current smoking); admission clinical characteristics, including vital signs (heart rate, systolic and diastolic blood pressure), Killip class, left ventricular ejection fraction, presence of ST-segment elevation; laboratory tests on admission (creatinine level, low-density lipoprotein cholesterol level); peak CK and cTn levels; and drugs during admission (aspirin, clopidogrel, β blockers, angiotensin converting enzyme inhibitors, angiotensin receptor blockers, calcium channel blockers, nitrates, antidiabetic agents, statins). To assess whether FPG-associated mortality risks differed among subgroups (ie, patients with and without known DM, or patients receiving different treatments), the Mantel–Haenzel test for heterogeneity was used to compare in-hospital and 3-year mortality.
Baseline demographic and clinical characteristics
A total of 1854 older patients presented with AMI in this study: men accounted for 62.6% of the population, while ST-segment elevation myocardial infarction accounted for 79.5% of presentations. Table 1 shows the demographic and clinical data. In the overall cohort, the mean age was 72.5±5.5 years (range 65–92 years). Only a minority of patients (24.6%) had a history of recognised DM. The prevalence of hypertension, hyperlipidaemia, prior myocardial infarction, prior stroke and current smoking was 57.0%, 4.9%, 10.6%, 14.1% and 29.1%, respectively. There were 46.6% patients belonging to Killip I class with the rest meeting criteria for Killip II–IV class. The median admission FPG was 6.0 mmol/l (range 2.5–26.8 mmol/l) with an average value of 7.0±3.2 mmol/l. According to the admission FPG level, patients were categorised into four groups: hypoglycaemia group (N=443, 23.9%), FPG≤5 mmol/l; euglycaemia group (N=812, 43.8%), FPG≥5.1 to≤7.0 mmol/l (5–7 mmol/l); mild hyperglycaemia group (N=308, 16.6%), FPG≥7.1 to≤9.0 mmol/l (7–9 mmol/l); and severe hyperglycaemia group (N=291, 15.7%), FPG≥9.1 mmol/l. Compared with patients who had lower admission FPG, patients with higher FPG tended to be more frequently women, with history of documented DM and higher Killip class and low-density lipoprotein cholesterol level. There were no significant differences among the four FPG level groups in other characteristics or in patients' treatment except for frequency of use of antidiabetic agents (for FPG ≤5 mmol/l, 15.1%; FPG 5–7 mmol/l, 18.9%; FPG 7–9 mmol/l, 35.1%; and FPG ≥9.1 mmol/l, 46.7%; p<0.001). A total of 104 (5.6%) patients received thrombolysis and most of them (83/104) used urokinase. There was no significant differences in the administration rate of thrombolytic agents because of FPG levels.
Association between admission FPG and mortality
In the overall cohort, a total of 190 patients (10.2%) died in hospital and 455 patients (24.5%) died during the 3-year follow-up (table 1). The probability of survival of patients in the different FPG level groups is shown by in-hospital and 3-year (figure 1) Kaplan–Meier curves. We specify the numbers at risk and list the cumulative probability of all cause mortality with 95% CI below the Kaplan–Meier curves. Compared with the euglycaemia group, both the hypo- and hyperglycaemia groups were associated with higher in-hospital and 3-year all-cause mortality. Patients in the FPG 5–7 mmol/l group had the best outcome (in-hospital mortality for FPG ≤5 mmol/l, 10.0%; FPG 5–7 mmol/l, 6.6%; FPG 7–9 mmol/l, 11.3%; and FPG ≥9.1 mmol/l, 19.6%; 3-year mortality for FPG ≤5 mmol/l, 23.7%; FPG 5–7 mmol/l, 15.8%; FPG 7–9 mmol/l, 30.5%; FPG ≥9.1 mmol/l, 44.0%). There was a striking U-shaped relationship between admission FPG levels and short- and long-term all-cause mortality in older patients with AMI (figure 2A).
FPG-associated in-hospital mortality risks differed between patients with and without recognised DM (figure 2B). In-hospital mortality of diabetic patients with hypoglycaemia (12.5%) was higher than that of diabetic patients with either mild hyperglycaemia (10.5%, 19.0% relative increase) or severe hyperglycaemia (10.7%, 16.8% relative increase). In contrast, in patients without DM, hyperglycaemia (FPG 7–9 mmol/l, 11.9%; FPG ≥9.1 mmol/l, 33.3%) had a more significant impact on in-hospital mortality than hypoglycaemia (9.7%). Regarding 3-year mortality, hyper- or hypoglycaemia had a similar effect in patients with or without DM (figure 2C). The FPG 5–7 mmol/l group had the lowest in-hospital and 3-year mortality.
In patients receiving different treatments, a U-shaped relationship persisted between admission FPG levels and mortality (figure 2D,E). In-hospital mortality in the drug treatment group was for FPG ≤5 mmol/l, 18.4%; FPG 5–7 mmol/l, 13.0%; FPG 7–9 mmol/l, 22.9%; and FPG ≥9.1 mmol/l, 27.7%. In-hospital mortality in the PCI group was: for FPG ≤5 mmol/l, 2.8%; FPG 5–7 mmol/l, 1.6%; FPG 7–9 mmol/l, 4.2%; and FPG ≥9.1 mmol/l, 12.1%. In-hospital mortality in the CABG group was: for FPG ≤5 mmol/l, 15.0%; FPG 5–7 mmol/l, 6.7%; FPG 7–9 mmol/l, 7.1%; and FPG ≥9.1 mmol/l, 8.3%. Three-year mortality in the drug treatment group was: for FPG ≤5 mmol/l, 44.5%; FPG 5–7 mmol/l, 27.4%; FPG 7–9 mmol/l, 49.5%; and FPG ≥9.1 mmol/l, 51.7%. Three-year mortality in the PCI group was: for FPG ≤5 mmol/l, 7.1%; FPG 5–7 mmol/l, 4.3%; FPG 7–9 mmol/l, 11.2%; FPG ≥9.1 mmol/l, 27.9%. Three-year mortality in the in CABG group was: for FPG ≤5 mmol/l, 31.7%; FPG 5–7 mmol/l, 25.3%; FPG 7–9 mmol/l, 48.9%; and FPG ≥9.1 mmol/l, 57.9%.
After multivariable adjustment, patients with hyper- or hypoglycaemia tended to have increased risks for both in-hospital and 3-year mortality compared with the euglycaemic population (table 2). Hazard ratios (HRs) and 95% CI for groups with FPG ≤5 mmol/l or FPG>7 mmol/l were significantly higher than 1.0. Accordingly, there remained a U-shaped relationship between FPG levels and mortality after adjustment for confounding factors.
Hypoglycaemia was associated with a greater increase in relative risk of in-hospital and 3-year mortality in patients with recognised DM compared with non-diabetic subjects. In patients without established DM, there was a graded increase in the risk of in-hospital and 3-year mortality as admission FPG became progressively higher. In contrast, among patients with DM, the hyperglycaemia-associated mortality risk increase was seen only at 3 years' follow-up. Differences in FPG-associated mortality risks persisted when analyses were repeated among patients who received different treatments.
Hyperglycaemia has been found to be an independent predictor of mortality.1–7 In a meta-analysis of 15 relatively small and mostly older studies that evaluated the association between admission glucose level and death, Capes et al7 demonstrated that the relative risk of in-hospital death in non-diabetic patients with AMI with admission glucose ≥6.1 mmol/l was 3.9 compared with non-diabetic patients with AMI who were normoglycaemic. Among patients with AMI with diabetes, those with admission glucose ≥10 mmol/l had a 70% relative increase in the risk of in-hospital death compared with diabetic patients with normal admission glucose values. Similarly, Foo et al5 demonstrated a near-linear relationship between higher admission glucose levels and higher rates of cardiac death among 2127 patients with acute coronary syndromes. The Cooperative Cardiovascular Project,4 the largest retrospective study of this subject to date, which examined the outcomes of 141 680 elderly patients with AMI, demonstrated a significant 13–77% relative increase in 30-day mortality and a 7–46% relative increase in 1-year mortality depending on the degree of hyperglycaemia. Nonetheless, less is known about the correlation between the initial FPG on admission and the occurrence of major adverse cardiovascular events. This study demonstrated an independent graded association between elevated levels of admission FPG and increasing in-hospital and 3-year mortality, consistent with the above studies. This higher risk of both short- and long-term mortality persisted after controlling for higher burden of comorbidities (such as previous myocardial infarction and hypertension) and greater disease severity (higher Killip class, higher peak CK and creatinine levels, and lower ejection fraction) seen in patients with raised FPG levels.
The prognostic significance of hypoglycaemia after AMI is controversial. Svensson et al3 conducted a study in 713 diabetic patients with unstable angina or non-Q-wave myocardial infarction and found a significantly higher mortality at 2 years in subjects with hypoglycaemia (admission glucose ≤3.0 mmol/l) than in those with euglycaemia; however, a causal link between in-hospital hypoglycaemia and clinical outcomes ascertained 2 years later is difficult to establish. Kosiborod et al13 showed that hypoglycaemia (admission glucose <3.3 mmol/l) was associated with increased mortality in patients with AMI, but this risk was confined to patients who had spontaneous hypoglycaemia (and did not include iatrogenic hypoglycaemia). A more recent analysis by Goyal et al14 showed that both admission and postadmission hyperglycaemia (admission glucose ≤3.8 mmol/l) could predict 30-day death in patients with AMI. However, only hypoglycaemia on admission predicted death, and this relationship dissipated after admission. Another report from the DIGAMI 2 (Diabetes mellitus, Insulin Glucose infusion in Acute Myocardial Infarction 2) trial15 showed that hypoglycaemia (admission glucose <3.0 mmol/l) during the initial hospitalisation was not an independent risk factor for future morbidity or mortality in patients with type 2 diabetes and myocardial infarction. Hypoglycaemic episodes were, however, more prevalent in patients at high risk for other reasons.
All the trials mentioned above used a much lower threshold than this study to define hypoglycaemia, and non-diabetic subjects were not included in some trials. In our study, we observed that mild to moderate decreasing FPG levels (≤5 mmol/l) were associated with a relative increase in risk of mortality. There was a U-shaped relationship between admission FPG levels and short- and long-term mortality. This U-shaped relationship was not restricted to patients with pre-existing diabetes. The prognostic correlates of admission FPG levels applied equally also to subgroups receiving contemporary treatments (drug treatment, PCI or CABG). In each subgroup, the risk of short- and long-term all-cause death increased in the hyperglycaemia and hypoglycaemia groups. Our result adds to the ongoing debate about admission plasma glucose level in patients presenting with AMI.
The exact mechanisms behind the association of hyperglycaemia/hypoglycaemia and higher mortality have not been definitively established. However, previous physiological studies show that higher glucose levels in patients with AMI are associated with higher free fatty acid concentrations (which may induce cardiac arrhythmias), insulin resistance and impaired myocardial glucose use, thus increasing the consumption of oxygen and potentially worsening ischaemia.16 Hyperglycaemia also has been associated with microvascular dysfunction,17 a prothrombotic state,18 vascular inflammation,19 endothelial dysfunction20 and generation of reactive oxygen species.21 All these mechanisms may potentiate tissue injury in the setting of AMI. Hypoglycaemia and rapid changes in blood glucose have been shown to increase levels of counter-regulatory hormones such as epinephrine and norepinephrine, which may induce vasoconstriction, platelet aggregation and thereby ischaemia.22 Furthermore, in the presence of hypokalaemia and raised serum catecholamine levels, which often simultaneously present during hypoglycaemia, cardiac repolarisation might be sufficiently prolonged to induce cardiac arrythmias.23
Previous randomised clinical trials of glucose control in AMI have been limited primarily to patients with recognised diabetes, and their results have been inconsistent.24–29 The DIGAMI study25 has been to date the only randomised trial of glucose control in AMI which achieved a significantly lower glucose level in the intervention arm than in the control arm; it is also the only randomised trial to have demonstrated a survival benefit associated with better glucose control. In the DIGAMI 2 trial,26 there was no difference in outcomes among the 1253 randomised patients with AMI (intensive treatment vs routine metabolic management). Furthermore, none of the most recently published trials (including BARI 2D,27 ADVANCE28 and ACCORD29), which compared different glycaemic–control targets, showed a significant reduction in cardiovascular events. Thus, it remains to be established if patients would benefit if their admission glucose was adjusted after arrival, and this needs to be further investigated in prospective randomised clinical trials.
Several possible limitations should be taken into account when interpreting the results of our study. First, given the retrospective nature of the analysis, the possibility of residual unmeasured confounding factors cannot be entirely excluded, and whether hyperglycaemia/hypoglycaemia are markers or mediators of death cannot be definitively determined in this observational study. Second, we were unable to determine how many patients with unrecognised DM before admission were diagnosed with DM after discharge. Finally, although causes of in-hospital death could be classified as cardiac and non-cardiac death, during follow-up some patients died out of hospital and we could not determine the accurate causes of death.
In older patients with AMI, increased as well as decreased admission FPG levels predicted higher in-hospital and 3-year mortality. There was a striking U-shaped relationship between admission FPG levels and short- and long-term mortality. An initial admission FPG level≥5.1 to≤7.0 mmol/l may be desirable because it was associated with better clinical outcomes.
The authors sincerely thank Roberto Patarca, MD for revising the manuscript.
Funding The Beijing Elderly Acute Myocardial Infarction Study (BEAMIS) was supported by the grant from Beijing Municipal Science and Technology Commission (No Z0005190042811).
Competing interests None.
Patient consent Obtained.
Ethics approval The study protocol was reviewed and approved by the ethical committee of each participating institution, including Beijing Anzhen Hospital Affiliated to Capital Medical University, People's Hospital Affiliated to Peking University, Beijing Xuanwu Hospital Affiliated to Capital Medical University, General Hospital of Beijing Millitary and Beijing Friendship Hospital Affiliated to Capital Medical University.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.