Early glucose variability in cardiogenic shock following acute myocardial infarction: a pilot study

Introduction

Glycaemic variability estimates glucose fluctuations which occur in a specific time frame [Kim et al. 2014], and patients with the same mean glucose levels may have different glycaemic variability patterns.

In critically ill patients, it was reported that the association between increased glycaemic variability and adverse outcomes, mainly increased mortality and morbidity (such as nosocomial infections and increased length of stay) [Krinsley, 2008], but no data are so far available on the association between glycaemic variability and outcomes in patients with cardiogenic shock (CS) following ST elevation myocardial infarction (STEMI).

The present investigation was therefore aimed at assessing the relationship between glycaemic variability and mortality, both short term and long term, in 67 consecutive patients with CS following STEMI admitted to our Intensive Cardiac Care Unit (ICCU).

Methods

The study population comprised 67 consecutive patients with CS following STEMI treated with primary percutaneous coronary intervention (PCI) and admitted to our ICCU from 1 January 2011 to 31 December 2013. All case data were collected prospectively, but retrospectively analysed.

A clinical diagnosis of CS was made in the presence of all the following criteria: (a) systolic blood pressure persistently lower than 90 mmHg or vasopressors required to maintain a systolic blood pressure of more than 90 mmHg; (b) signs of hypoperfusion (e.g. urine output less than 30 ml/h or cold/diaphoretic extremities or altered mental status); (c) clinical evidence of elevated left-ventricular filling pressure (e.g. pulmonary congestion on physical examination or chest X-ray) [Hochman, 2003; Richards and Wilcox, 2014]. Pulmonary artery catheterization was not required when clinical criteria and echocardiographic evidence of the left-ventricular dysfunction without mechanical complications were both present. The treatment of shock was according to international recommendations [Richards and Wilcox, 2014].

The Acute Physiology and Chronic Health Evaluation II (APACHE II) score was calculated on admission.

According to guidelines [Deedwania et al. 2008], intensive insulin therapy was administered in patients with significant hyperglycaemia (i.e. plasma glucose greater than 180 g/L).

Glucose values were measured every 12 h during the first 48 h after ICCU admission. Glycaemic variability was measured in the first 48 h by means of standard deviation (SD) of glucose values [Krinsley, 2008; Hermanides et al. 2010], and the mean absolute glucose change per hour (MAGC) defined as the sum of all absolute glucose change divided by the time in hours [Hermainides et al. 2010].

Primary outcome was all-cause death, during ICCU and at follow up.

Follow up was available in all patients discharged alive (8.8 months [0.1–17.8]). Ten patients died (10/38, 26%) during follow up.

The study protocol was in accordance with the Declaration of Helsinki and approved by the local ethics committee. Informed consent was obtained from all patients before enrolment. In patients mechanically ventilated, informed consent was obtained from their relatives.

Results

Our series included 67 CS patients (mean age 72.9 ± 12.4 years), mainly males (M:F, 40/27 (59.7/40.3). A history of diabetes was present in 19 patients (28.4%), while 37 patients were hypertensive (55.2%). A total of 29 patients died (43.3%) during ICCU stay,. In our series, length of stay was 96 (36–173) h (survivors: 120 h [60–206], nonsurvivors: 48 h [20–144]). When compared with survivors, nonsurvivors were older (dead: 76.8 ± 10.9 years versus survivors: 70.0 ± 12.8 years, p = 0.024), and showed a higher APACHE II score (dead: 26 [18–28] versus survivors: 17 [12–22], p < 0.001). No difference was observed in admission ejection fraction values between the two subgroups (dead: 28.3 ± 11.6 versus survivors: 32.5 ± 8.0, p = 0.085). In our series, no difference was detectable in the severity of coronary artery disease between survivors and dead patients. Intra-aortic balloon pumps were implanted in 67.2% (45/67) and revascularization was more frequently completed in survivors (p = 0.030).

Table 1 shows glucose values (mean ± SD) measured every 12 h for the first 48 h in all patients as well as in survivors and dead patients. Mean absolute glucose change per hour during the first 48 h is also shown in all patients as well as in survivors and dead patients. As depicted in Table 1, a significant and progressive decrease in glucose values was observed in all patients during the first 48 h from admission during which nonsurvivors showed significantly higher glucose levels than survivors. Lower glycaemic variability was observed in survivors when compared with nonsurvivors, as indicated by lower values of SD and MAGC, respectively. No difference was observed in nadir glycaemia between survivors and nonsurvivors (survivors: 0.91 [0.80–0.95], versus nonsurvivors: 0.97 [0.65–1.20] g/L, p = 0.375), while nonsurvivors showed significantly higher values of peak glycaemia (nonsurvivors: 2.24 [2.03–3.00] versus survivors: 1.89 [1.60–2.32] g/L, p = 0.008). Values of glycaemia < 0.70 g/L were observed in 10 patients: 3 patients among survivors (3/ 27, 11.1%) and 7 patients among nonsurvivors (7/39, 17.9%); (p = 0.446).

OPEN IN VIEWER

Table 1.

Glucose values and glucose variability during the first 48 h.

Glucose values	During the first 48 h
Mean ± standard deviation	Admission glycaemia	12 h	24 h	36 h	48 h	Within groups p
All	1.98 ± 1.11	1.40 ± 0.46	1.29 ± 0.40	1.34 ± 0.34	1.20 ± 0.28	< 0.001
Survivors	1.65 ± 0.72	1.41 ± 0.48	1.31 ± 0.39	1.31 ± 0.35	1.16 ± 0.25	< 0.001
Dead	2.40 ± 1.38	1.39 ± 0.45	1.26 ± 0.43	1.41 ± 0.33	1.32 ± 0.34
Between groups p	0.019
Glucose variability during the first 48 h			All patients (n = 67)	Survivors (n = 38, 56.7%)	Dead (n = 29, 43.3%)	p
Mean absolute glucose change per hour			0.42 (0.20–0.68)	0.37 (0.18–0.55)	0.63 (0.30–0.97)	0.011
Standard deviation			0.36 (0.23–0.58)	0.31 (0.18–0.43)	0.56 (0.29–0.91)	0.008

All values are median (interquartile range) and are reported as g/L.

At linear regression analysis, MAGC and SD were significantly related to mean glucose values, respectively (MAGC: coefficient 0.54, Pearson’s R² 0.34, p < 0.001; SD: coefficient 0.64, Pearson’s R² 0.38, p < 0.001)

Cox regression analysis is shown in Table 2. Different models are depicted to assess the predictive value of different glucose measurements for long-term mortality. In Cox regression analysis, MAGC and SD were independent predictors of death (MAGC: adjusted hazard ratio [HR]: 8.60, 95% confidence interval [CI]: 2.21–33.41, p = 0.002; SD: adjusted HR: 6.64, 95% CI: 1.92–22.99, p = 0.003), as well as peak glycaemia (adjusted HR: 1.95, 95% CI: 1.20–3.15, p = 0.007) .

OPEN IN VIEWER

Table 2.

Cox regression analysis.

	Adjusted HR	95% CI	p value
Model 1
Age (1-year step)	1.03	1.00–1.07	0.086
Admission LVEF (1% step)	0.98	0.93–1.03	0.393
APACHE II (1-unit step)	1.03	1.00–1.05	0.077
Admission eGFR (5 ml/min/1.73 m2 step)	0.91	0.85–0.97	0.003
MAGC (1 g/L step)	8.60	2.21–33.41	0.002
Model 2
Age (1-year step)	1.04	1.00–1.08	0.040
Admission LVEF (1% step)	0.98	0.93–1.03	0.380
APACHE II (1-unit step)	1.02	1.00–1.05	0.097
Admission eGFR (5 ml/min/1.73 m2 step)	0.92	0.86–0.97	0.006
SD (1 g/L step)	6.64	1.92–22.99	0.003
Model 3
Age (1 year step)	1.05	1.01–1.09	0.023
Admission LVEF (1% step)	0.98	0.94–1.03	0.442
APACHE II (1-unit step)	1.02	1.00–1.05	0.071
Admission eGFR (5 ml/min/1.73 m2 step)	0.92	0.87–0.98	0.007
Peak glucose (1 g/L step)	1.95	1.20–3.15	0.007
Model 4
Age (1-year step)	1.04	1.01–1.08	0.027
Admission LVEF (1% step)	0.99	0.94–1.03	0.530
APACHE II (1-unit step)	1.03	1.01–1.05	0.019
Admission eGFR (5 ml/min/1.73 m2 step)	0.92	0.86–0.97	0.003
Nadir glucose (1 g/L step)	2.98	0.79–11.30	0.108

APACHE II: Acute Physiology and Chronic Health Evaluation II; CI, confidence interval; eGFR, estimated glomerular filtration rate; HR, hazard ratio; LVEF, left-ventricular ejection fraction; MAGC, mean absolute glucose change per hour; SD, standard deviation.

Discussion

The main finding of the present investigation is that early glycaemic variability (as indicated by MAGC and SD) is an independent predictor of mortality in consecutive CS patients following STEMI.

The association between high glycaemic variability and mortality has been described in different populations of critically ill patients [Ali et al. 2008]. Dossett and colleagues investigated the impact of glycaemic variability in a cohort of ventilated surgical intensive care unit (ICU) patients [Dossett et al. 2008]. In this series, survivors and nonsurvivors had similar mean glucose levels during ICU stay, but nonsurvivors showed increased values of several measures of glycaemic variability. In 4084 critically ill patients, glycaemic variability was strongly and independently associated with mortality among critically ill nondiabetic patients, but not among patients with diabetes [Krinsley, 2009]. Egi and colleagues analyzed data on glucose values from 7049 Australian patients admitted to five different ICUs over 4 years and observed that glycaemic variability was a significant and independent predictor of ICU and hospital mortality, and that it was a stronger predictor of ICU mortality than mean glucose concentration [Egi et al. 2006]. More recently, in 297 medical and surgical patients receiving total parental nutrition [Farrokhi et al. 2013], high glycaemic variability was associated with increased mortality, independently of hypoglycaemia or hyperglycaemia. Few studies investigated the role of glycaemic variability in patients with acute myocardial infarction [Su et al. 2013; Zang et al. 2014; Wang et al. 2014]. Su and colleagues observed that elevated admission glycaemic variability was more important than admission glucose and prior long-term abnormal glycometabolic status (glycosylated haemoglobin) in predicting 1-year major cardiac adverse events (MACE) in 222 patients with acute myocardial infarction [Su et al. 2013]. In STEMI patients glycaemic variability remained an independent prognostic factor for composite MACE in STEMI patients undergoing primary PCI [Zang et al. 2014]. Similarly, in a small series of 34 patients glycaemic variability was an independent predictor of MACE in diabetic patients with acute myocardial infarction [Wang et al. 2014].

The present study assessed for the first time the impact of glucose variability in patients with CS following STEMI, and we documented that survivors showed lower glycaemic variability than nonsurvivors and that early glycaemic variability is an independent predictor of early mortality in our series.

There are several possible explanations for this association. First, lower glycaemic variability may indicate more attention in medical and nursing care. Second, higher glycaemic variability may be associated with more severe illness. Third glycaemic variability may have a true deleterious effect since there is growing evidence from experimental models [Piconi et al. 2004], and patient studies in diabetes mellitus that fluctuations in blood glucose may increase the risk of hyperglycaemia-induced oxidative stress [Monnier et al. 2006].

In the available literature [Yang et al. 2013; Vis et al. 2007], only the prognostic impact of admission glycaemia has been investigated in patients with CS following acute myocardial infarction, and it was reported that admission glucose levels were independently associated with mortality in these patients. In the present investigation, we further extended these findings observing that significantly higher glucose values are detectable in nonsurvivors during the first 48 h and that peak glycaemia is an independent predictor of mortality.

A possible limitation of our study is represented by the small number of enrolled patients. However, our series comprises consecutive CS patients, all managed by the same medical team according to implemented protocols for glucose management [Richards and Wilcox, 2014; Lazzeri et al. 2012]. Further studies are needed to confirm our results in larger cohorts and eventually to assess the effect of strategies specifically targeting glucose variability reduction on mortality. Another potential limitation may be the fact that in our study we measured glucose values every 12 h during the first 48 h after ICCU admission while recent papers [Su et al. 2013; Zang et al. 2014; Wang et al. 2014], assessing the role of glycaemic variability in patients with acute myocardial infarction, used a continuous glucose monitoring system, which is supposed to provide a more accurate measurement of glycaemic changes (since it provides a higher number of detection data). However, so far, there is no acknowledged gold standard for numerous indicators reflecting glycaemic variability, especially in critically ill patients [Meynaar et al. 2012].