Introduction
Hyperglycemia, defined as serum glucose > 11 mmol/l after subarachnoid hemorrhage, is a common finding and is an independent predictor of long term poor outcome [
1,
2]. It is believed that the surge in serum glucose immediately after injury is related to a complex interrelationship between sympathetic and inflammatory mediators, acquired insulin resistance and administration of dexamethasone [
3].
While such observational data argue for tight glycemic control, recent studies investigating the relationship between serum and brain glucose levels have found that intensive insulin protocols targeted to a tight serum glucose level (4.4 to 6.1 mmol/l) may compromise cerebral metabolism by leading to concomitant critical reductions in interstitial brain glucose levels [
4,
5].
Each of these studies did not take into account the impact that nutritional support may have upon outcomes related to tight glycemic control. The adequate delivery of non-protein calories was rigorously maintained in randomized controlled trials of surgical and medical ICU patients [
6‐
8], and a recent single center study of SAH patients found that a bolus delivery of enteral nutrition significantly impacted brain glucose levels [
9].
We sought to understand the impact enteral nutrition has upon the relationship between serum and brain interstitial glucose levels in a cohort of subarachnoid hemorrhage patients treated with institutional protocol targeting serum glucose levels between 5.5 to 7.8 mmol/l. We hypothesized that higher non-protein caloric delivery would result in higher interstitial brain glucose levels and reduce the incidence of brain metabolic crisis.
Discussion
In this study of 50 poor-grade SAH patients, we found that our insulin protocol maintained serum glucose levels ≤ 7.8 mmol/l and similar to previous observations that low serum glucose levels were associated with brain metabolic crisis. We were unable; however, to directly demonstrate our central hypothesis that higher non- protein caloric delivery results in higher interstitial brain glucose levels. In contrast, Kinoshita
et al. [
9] demonstrated increases in serum and brain glucose concentrations two hours after a 250-calorie bolus of ENSURE
® (Abbott Laboratories, Inc., Columbus, Ohio, USA
{). This discrepancy is likely due to a significantly lower caloric delivery in our patients (0.442 calories/kg/hour) in combination with our intravenous insulin protocol that is designed to maintain serum glucose levels within a range of 5.5 to 7.8 mmol/l. Kinosheta
et al. provided bolus feeds that resulted in higher serum glucose levels (183.8 ± 26.2 mg/dL) that were actively controlled in our observational study. This is a limitation experimentally but does suggest that brain metabolism is not supported by enteral nutrition under normal clinical circumstances. Underfeeding is a common problem in critically ill patients and as we and others have shown, it is associated with higher rates of medical and infectious complications as well as poor long term outcome [
23,
24].
As expected, we observed greater administration of insulin as serum glucose levels increased. However, we were surprised to observe a different relationship between brain glucose levels and insulin doses. In 2011, Zetterling
et al. [
25] reported that insulin administration lowered cerebral microdialysis brain glucose and pyruvate concentrations, often to low levels, despite plasma glucose remaining above 6 mmol/L. While cerebral metabolism monitoring is used to detect acute events, in practice severity of the initial brain injury is the primary determinate of observed metabolic status [
26]. An LPR greater than 40 has been established in the literature to represent a state of brain metabolic crisis (for example, [
27]). We speculated that perhaps the effect of insulin on brain glucose concentration is altered by the severity of injury of the tissue being monitored. We found that when metabolic status was not critical (LPR < 40) that the brain glucose-insulin administration relationship was similar to the serum glucose-insulin administration relationship. In contrast, during metabolic crisis (LPR ≥ 40) a negative relationship was observed, such that higher insulin doses (> 2 IU/hr) were associated with a lower brain glucose level. While our observational study design restricts our ability to infer causal relationships between insulin and brain glucose, our observations do appear congruent with the findings of Zetterling
et al.
It is physiologically plausible that the effect of insulin on brain glucose metabolism may be altered by brain injury. Previous experimental and clinical studies have shown that insulin increases glucose transport across the blood-brain barrier [
28], promotes glucose storage as glycogen in astrocytes [
29], and stimulates protein and RNA and DNA synthase in the brain [
30,
31]. High concentrations of endogenous insulin may enable brain cells to divert glucose metabolism to secondary pathways [
32,
33] that may be important to neuronal repair, including fatty acid and amino acid production, and the pentose phosphate shunt pathway, which may be important to protect against oxidative damage [
34].
It is conceivable that protocol-driven clinical administration of insulin to maintain serum glucose concentrations may provide a dangerous false signal to injured brain tissue that excess glucose is available and can be diverted from ATP production to secondary pathways for tissue repair functions. Vespa
et al., [
4] conducted a dual microdialysis positron emission tomography (PET) scan study in traumatic brain injury patients demonstrating that even though intensive insulin therapy resulted in lower brain glucose concentrations, the global metabolic rate of glucose did not change and that this corresponded with signs of energy failure, including oxygen extraction increases to near-ischemia level and brain glutamate and lactate pyruvate ratio increases. Two other studies found significant insulin-related reductions in brain glucose concentrations but did not find evidence that this resulted in increased metabolic distress [
25,
35]. However, in both studies the reported mean LPR was approximately 30. No study has looked at the relationship between brain glucose levels and insulin during cerebral metabolic distress (LPR > 40). Glucose metabolism of healthy brain tissue is stable and relatively unaffected by normal alterations of cerebral perfusion in contrast to that of injured brain tissue, which is very sensitive to such changes [
36].
It is currently unknown if providing additional nutritional support to brain injured patients in combination with insulin therapy may be important to support increased cerebral energy demands and brain tissue recovery. The current cohort of patients, on average, received approximately half the amount of calories recommended by the American College of Chest Physicians' guidelines for critically ill patients (approximately 25 calories/kg/day) [
20], but is consistent with reports of underfeeding of neurological critically ill patients elsewhere [
37]. We speculate that the effect of insulin on brain glucose metabolism in the context of brain injury remains poorly understood and is in need of further study.
Our study has several limitations. The true impact of enteral nutrition on brain metabolism is difficult to determine due to the fact that serum glucose concentrations were actively controlled with insulin. Evaluating relatively small-to-moderate differences in calories received over time may also mask the effects of enteral nutrition on brain metabolism. Individual responses to increases in calories received via enteral nutrition and the impact of insulin infusions were not analyzed in this study. Although our findings are consistent with the few studies available in this area [
4,
9,
25,
35,
38], a patient-specific analysis might lead to different conclusions. This study does not provide any data to clarify a mechanism for insulin-related reductions of brain glucose concentrations or suggest whether adequate caloric intake (for example, 25 calories/kg/day) might mitigate this effect. Prospective studies with controlled amounts of caloric delivery are required to adequately address these limitations.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JMS participated in the design of the study, directed data collection, performed the data analyses, led discussion of results interpretation and drafted the manuscript. JC participated in the collection of data, results interpretation and made critical revisions to the manuscript. SBK collected cerebral metabolism data and participated in drafting the manuscript. HL collected the clinical patient data and participated in results interpretation. MP collected ICU patient data and participated in results interpretation. KL, ESC and SAM participated in data collection, results interpretation and made critical revisions to the manuscript. DSS provided expertise in interpretation of nutrition results and made critical revisions to the manuscript. NB conceived of the study, participated in the collection of data, results interpretation, and drafting of the manuscript. All authors have read and approved the final manuscript.