Introduction
An elevated blood glucose level in critically ill patients is termed stress hyperglycemia. Several studies have documented that stress hyperglycemia affects outcomes in patients with various clinical conditions requiring intensive care, such as myocardial infarction, cerebrovascular disorders, or traumatic brain injury [
1‐
3]. Hyperglycemia has also been indicated to have deleterious effects on patients with sepsis [
4‐
6]. The Surviving Sepsis Campaign guidelines have therefore consistently recommended blood glucose control with a goal of < 150 mg/dl, since the publication of the original version in 2004 [
7‐
9].
In the pathophysiology of sepsis, proinflammatory cytokines including TNFα, IL-1, and IL-6 are known to play a pivotal role, and overproduced cytokines enter into the bloodstream causing hypercytokinemia, which leads to organ failure via humoral mediator network activation and vascular endothelial damage [
10,
11].
We have routinely measured blood levels of IL-6 in all patients with sepsis since 2000 to assess the severity of hypercytokinemia, and we have reported the clinical usefulness of the routine measurement of IL-6 levels in septic patients [
12,
13]. Several other clinical studies have also demonstrated that measurement of the blood IL-6 level is useful as a biomarker of hypercytokinemia [
14‐
16]. We have also implemented blood glucose control in patients with sepsis according to the Surviving Sepsis Campaign guidelines since 2004 [
7]. Based on our daily clinical experience in critical care of septic patients, we hypothesize that a correlation exists between the blood IL-6 level and hyperglycemia in sepsis.
With this background, we retrospectively investigated the correlation between blood IL-6 level and blood glucose level in patients with sepsis. We also investigated the relationship between the blood IL-6 level and glucose control in the same patients.
Materials and methods
Patients
This retrospective observational study was conducted in patients with sepsis who were admitted to the ICU of Chiba University Hospital (an eight-bed general ICU annually admitting approximately 800 to 1,000 medical/surgical patients, both adult and pediatric) from 2005 to 2010. The inclusion criteria for this study were: age 18 years or older, and diagnosis with sepsis on ICU admission according to the definition of the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference [
17]. The exclusion criteria were: ICU stay shorter than 7 days (too short for sufficient observation of changes in blood glucose level or appropriate determination of effects of glucose control), and steroid therapy during the ICU stay (potentially influencing blood glucose level and inflammatory reaction).
Prior to the initiation of critical care in the ICU, each patient or his/her relative provided written informed consent to possible research use of imaging and blood testing data required for clinical practice as well as information regarding the effectiveness of various treatments, with confidentiality of personal information secured. This retrospective study was approved by the Chiba University Hospital Clinical Research Center Ethics Committee before initiation of the research.
Patient management for glucose control
We have implemented blood glucose control in all septic patients, adopting the goal of < 150 mg/dl recommended in the Surviving Sepsis Campaign guidelines [
7‐
9]. In addition to an upper-limit blood glucose goal of < 150 mg/dl, a lower-limit blood glucose goal of > 100 mg/dl was adopted to minimize the risk of hypoglycemia. Arterial blood samples for blood glucose monitoring were collected via an intra-arterial pressure monitoring line, and blood glucose was measured with a glucose analyzer built in a blood gas analysis system (Stat Profile pHOx Plus M; NOVA Biomedical Japan, Tokyo, Japan).
Parenteral nutrition was initiated with glucose, amino acids, and vitamins at 100 to 400 kcal/day. Early transition to enteral nutrition was attempted as soon as possible if the cardiopulmonary function of the patient was stable, and unless there was bowel dysfunction in the patient. In patients with septic shock, enteral nutrition was initiated after recovery from shock. Total energy intake was gradually increased by increasing doses of enteral nutrition. Energy intake was also increased with special care to avoid hyperglycemia. To reduce the blood glucose level, regular insulin was continuously administered intravenously using an infusion pump. Measurement of the blood glucose level and adjustment of the infusion rate of intravenous nutritional solution and insulin were all implemented by intensivists only, not by nurses and non-intensivist doctors. The intervals of measurement of blood glucose level and adjustment of infusion rate of intravenous nutrition solution and insulin were basically 3 to 6 hours. However, intensivists shortened measurement intervals, and changed glucose/insulin doses more frequently, if they judged it necessary.
According to the management protocol of the present study, blood glucose control was performed in all patients, aiming to achieve the blood glucose goals within 24 hours after ICU admission. Successful blood glucose control was defined as a blood glucose level of < 150 mg/dl achieved within 24 hours after ICU admission and maintained thereafter within a range between 100 and 150 mg/dl for at least 6 days (that is, up to ICU day 7).
Blood IL-6 measurement
The blood IL-6 level was determined by chemiluminescent enzyme immunoassay method. A rapid assay system (Human IL-6 CLEIA; Fujirebio, Tokyo, Japan), requiring an assay time of approximately 30 minutes, was used for monitoring the blood IL-6 level with an automated device (Lumipulse f
®; Fujirebio) [
12,
13]. The blood IL-6 level thus obtained was used as an index of hypercytokinemia.
Comparative parameters and statistical analysis
All continuous variables are expressed as the mean ± standard deviation. Normal distributions were confirmed for all continuous variables other than the blood IL-6 level. The blood IL-6 level was used for statistical analysis after logarithmic conversion, since a normal distribution was confirmed for logarithmically converted values of this parameter.
Background parameters were compared between two patient groups, those with successful glucose control and those with failed glucose control, using the unpaired t test (continuous variables) and the chi-square test (dichotomous variables). The correlation between blood glucose level and blood IL-6 level on ICU admission (that is, at baseline) was examined by calculating Pearson's product-moment correlation coefficient. The correlations between the blood glucose level and other biomarkers (lactate, C-reactive protein, and so forth) or severity scores, such as the Acute Physiology and Chronic Health Evaluation (APACHE) II or the Sequential Organ Failure Assessment score on ICU admission (that is, at baseline), were also examined by calculating Pearson's product-moment correlation coefficient.
The rate of successful blood glucose control was compared among three patient groups stratified by blood IL-6 level on ICU admission using the chi-square test. Time-course changes in the blood glucose level, the blood IL-6 level, energy intake, and insulin dose during the ICU stay were compared between the successful and failed blood control groups by repeated-measures analysis of variance. For patients in the successful glucose control group, the correlation between insulin dose on ICU day 1 (required for achievement of blood glucose goal < 150 mg/dl) and blood IL-6 level on ICU admission was further examined by calculating Pearson's product-moment correlation coefficient. The 28-day and 60-day survival rates were compared between the successful and failed glucose control groups using the chi-square test. Statistical package software (PASW Statistics 18 for Windows; SPSS Japan Inc., Tokyo, Japan) was used for all statistical analyses, with P < 0.05 considered significant.
Discussion
In the present study we investigated the relationship between the IL-6 level, the glucose level, and glucose control in septic patients. The main findings of the present study are as follows. First, a significant correlation was observed between the blood IL-6 level and the blood glucose level on ICU admission (Table
2). Second, the rate of successful glucose control decreased with an increase in the blood IL-6 level on ICU admission (Figure
1). Third, in the failed glucose control group the insulin dose per 100 kcal energy intake was higher (Figure
2c) and glucose control was more difficult (Figure
2d), with the blood IL-6 level remaining higher (Figure
2e). Finally, a significant positive correlation was observed between the blood IL-6 level and daily insulin dose (required for achievement of the blood glucose goal) on ICU day 1 in the successful glucose control group (Figure
3). These results suggest that a high IL-6 level is associated with hyperglycemia and difficulties in glucose control. Furthermore, in the comparison of each parameter on ICU admission between the successful and failed glucose control groups, the APACHE II score, lactate level, and IL-6 level were significantly different.
P values concerning comparison of the IL-6 level were highest when comparing all parameters (Table
1). In the comparison of correlation coefficients between the blood glucose level and various parameters, the correlation coefficients between IL-6 and glucose levels in nondiabetic patients were the highest of all of the correlation coefficients, as shown in Table
2. From these results, it is suggested that high IL-6 level is an independent marker of severity of sepsis.
Proinflammatory cytokines play a pivotal role in the pathophysiology of sepsis, and overproduced cytokines enter the bloodstream causing hypercytokinemia, which induces various pathophysiological changes and eventually results in organ failure [
10,
11]. We have measured the blood level of IL-6 for real-time monitoring of the severity of hypercytokinemia, and previously reported the usefulness of blood IL-6 measurements in patients with sepsis [
12,
13]. Although a limitation has existed in single cytokine measurement, several other studies also demonstrated that the blood level of IL-6 was useful as a biomarker of hypercytokinemia [
14‐
16].
While involvement of stress hormones (for example, corticosteroids, glucagon, catecholamines) in the mechanism of stress hyperglycemia has been repeatedly described, proinflammatory cytokines can cause hyperglycemia by various mechanisms [
4‐
6]. For example, TNFα is known to induce insulin resistance by enhancing secretion of stress hormones [
18]. In a rat model of zymosan-induced inflammatory response, Petit and colleague reported that an increase in the blood TNFα level induced a reduction in glucose uptake in skeletal muscle [
19]. IL-6 is known to inhibit tyrosine phosphorylation of insulin receptor substrate-1 in isolated human adipocytes, and thereby to suppress glycogen synthesis [
20]. In several clinical studies, the serum level of TNFα or IL-6 was increased and glucose tolerance was impaired in diabetics compared with nondiabetics [
21,
22]. Combining these results of previous basic and clinical research with our findings, these results suggest that hypercytokinemia might be involved in the development of hyperglycemia in sepsis and thereby might affect the success of glucose control.
Although two large-scale clinical studies on glucose control in septic patients previously failed to demonstrate clinical significance of glucose control due to high incidence of hypoglycemic events [
23,
24], recent review articles suggest that stabilization of the blood glucose level is important in the management of sepsis [
4‐
6]. In our present study, the patients with successful glucose control showed better outcome than the patients with failed glucose control (Table
3). These results from our study also support the importance of glucose control in septic patients.
While the clinical efficacy of glucose control in sepsis has been demonstrated, our clinical experience suggests that achievement of glucose control may remain difficult. Recently, three strategies of successful glucose control have been reported [
25,
26]. One strategy is statistical process control using a newly developed glucose control protocol on the basis of statistical analysis of previous blood glucose level data [
25]. Another is a computer-generated alert system that involves an electronic blood glucose alert in the bedside computer [
26]. The third strategy involves the application of an artificial endocrine pancreas, a mechanical glucose control system that is capable of automatic blood sampling at short intervals, measurement of blood glucose levels, and automatic adjustment of insulin and glucose dose [
27].
Considering that a high IL-6 level correlated with glucose control in the present study, there is a possibility that improving hypercytokinemia is a new strategy for successful glucose control in addition to above three strategies. Blood purification such as continuous hemodiafiltration is one of the most promising treatments available for the treatment of hypercytokinemia [
28,
29]. We have reported that continuous hemodiafiltration using a polymethylmethacrylate membrane hemofilter with cytokine-adsorbing capacity successfully improved hypercytokinemia [
29]. Although there is little evidence, these blood purifications might be expected to control hypercytokinemia and thus to facilitate glucose control.
The present study has several limitations. First, although we found correlation between a high IL-6 level and hyperglycemia in the present study, it is difficult to elucidate the interaction between high IL-6 level and hyperglycemia. Previous studies suggest that high proinflammatory response might be one of the causes of hyperglycemia. However, additional study is needed to prove these mechanisms. Second, because there was a weak correlation between the APACHE II score and glucose level on ICU admission, we could not completely deny that IL-6-related hyperglycemia might be an epiphenomenon of severe illness. Furthermore, the observed difference in survival rate between the successful and failed glucose control groups does not necessarily demonstrate effectiveness of blood control in sepsis, since there was a substantial difference in severity scores between these two groups. Future studies employing prospective designs may be needed to further elucidate the relationship between hypercytokinemia and stress hyperglycemia and the relationship between hypercytokinemia and glucose control.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MN carried out the data acquisition, secondary database construction, and drafted the manuscript. SO and HH participated in the design of the study, helped to draft the manuscript, and performed the statistical analysis. EW, RA, TN, and YM conceived of the study, participated in its design and coordination, and helped to draft the manuscript. SO and HH helped to draft, read, and approve the final manuscript. All authors approved the final manuscript for publication.