Insulin-related decrease in cerebral glucose despite normoglycemia in aneurysmal subarachnoid hemorrhage

This study was approved by the local research ethics committee at Charité Virchow Medical Center, in accordance with the Declaration of Helsinki as revised in Edinburgh in October 2000. Written informed consent was obtained from each patient or their closest family relative.

This study forms part of an ongoing prospective study on cerebral metabolism monitored by bedside microdialysis in aneurysmal SAH patients. A total of 31 patients were enrolled, who were admitted to the neurosurgical department of a university hospital between August 2005 and June 2007 and who met the following criteria: confirmation of SAH by head computed tomography; presence of intracranial aneurysm(s), as demonstrated by cerebral angiogram; and surgical therapy. All patients considered surgical candidates were managed in accordance with a uniform protocol detailed previously [5]. Depending on their neurological course, the patients were classified as asymptomatic (n = 6) or as exhibiting symptoms of acute focal deficit (AFND; n = 14) or delayed ischaemic neurological deficit (DIND; n = 11), with detailed definitions described previously [5]. Information compiled for each patient included the following: haemodynamic parameters, respiratory values, laboratory test results, fluid balance data and chest radiograph findings. Glucose levels were targeted to be under 140 mg/dl using intravenous insulin if necessary. Routine glycaemic control consisted of blood glucose checks every 4 hours (6 hours in patients without insulin treatment) and goal-directed adjustment of insulin infusion by the bedside nurse.

A microdialysis catheter (CMA 70; CMA Microdialysis, Solna, Sweden; length 10 mm, molecular weight limit of 100 kDa) was inserted immediately after aneurysm clipping into brain parenchyma of the respective vascular territory of the aneurysm (for instance, the right frontal lobe in patients with an anterior communicating artery aneurysm). Care was taken to avoid insertion into macroscopically lesioned brain tissue or into an intracerebral haemorrhage. Catheters were perfused with sterile Ringer's solution at a flow rate of 0.3 μl/min. On the outlet tube, perfusates were collected in microvials, exchanged hourly and analyzed immediately at bedside in a mobile photometric, enzyme-kinetic analyzer (CMA 600; CMA Microdialysis). The estimated recovery for the system is 0.65 to 0.72 [10]. Parameters of energy metabolism (glucose, pyruvate, lactate, lactate:pyruvate ratio), glycerol (a marker of cell membrane degradation) and glutamate (a marker of ischaemia) were analyzed [6, 11, 12]. In a series of studies, changes in the lactate:pyruvate ratio and glutamate were shown to indicate early the onset of delayed neurological deterioration and to be in good accordance with the clinical course of SAH patients [4, 5, 11‐15]. The lactate:pyruvate ratio was the best metabolic prognostic marker of 12-month outcome in SAH [16].

Microdialysis data are presented as microdialysate concentrations. The occurrence of critical decreases in cerebral glucose (defined as decrease to <0.6 mmol/l, corresponding to the mean level minus 1 standard deviation of our asymptomatic SAH patients [8]) was recorded.

Demographic and clinical characteristics of the 31 patients are summarized in Table 1. Patients were classified as an insulin treatment group (n = 24) and a noninsulin treatment group (n = 7). All patients (n = 31) underwent early surgery (within 72 hours after initial bleeding). Age, admission World Federation of Neurological Surgeons (WFNS) score and Fisher score were significantly higher in the insulin treatment group (Table 1).

In 92% of the insulin-treated patients, hyperglycaemia was present on admission (86% in the noninsulin treatment group) and in all patients on at least one of the following 10 days in the intensive care unit. Continuous intravenous insulin treatment started 2.6 ± 3.0 days after admission. The median blood glucose at onset of insulin infusion was 142.0 ± 7.6 mg/dl. No episodes of insulin-induced hypoglycaemia (lowest value 80 mg/dl) occurred during 10 days after SAH. A critical decrease in cerebral glucose (to <0.6 mmol/l) occurred in 19 patients (79%) of the insulin treatment group, beginning 99.1 ± 54.4 hours after initiation of insulin infusion (duration 6.9 ± 11.6 hours) and in three patients in the noninsulin treatment group. In older patients (P = 0.03) and males (P = 0.02), low cerebral glucose values during insulin infusion occurred more frequently than in younger patients and females. There was no difference in the incidence and duration of critically low cerebral glucose values (<0.6 mmol/l) between the insulin treatment and noninsulin treatment groups.

Results from Vespa and colleagues [9] in patients with TBI showed that microdialysis glucose reached its nadir at 11 ± 1 hours after onset of insulin infusion. Hence, in the present study the microdialysis parameters were analyzed hourly for 12 hours after initiation of insulin infusion. Interestingly, although blood glucose remained stable after insulin onset, cerebral glucose began to decrease significantly at 3 hours after the start of infusion and remained low until the end of our observation period (Figure 1). Extracellular values of pyruvate, lactate and lactate:pyruvate ratio did not change during insulin infusion (Figures 2 to 4). Glycerol, a marker of cellular membrane degradation, tended toward higher values at the end of the observation period (9 to 12 hours; Figure 5). This might either reflect tissue damage after SAH or indicate cellular distress following insulin-induced decrease in cerebral glucose. Extracellular concentrations of glutamate decreased during insulin infusion (hours 6 to 12; P < 0.05), indicating that no additional ischaemia was present during the observation period (Figure 6). Furthermore, the decrease in glutamate might reflect metabolic improvement caused by treatment of hyperglycaemia.

It must be considered that a decrease in cerebral glucose might not be caused by insulin treatment alone but can also reflect an increased consumption or decreased delivery caused by cerebral vasospasm. To evaluate whether the occurrence of vasospasm led to reductions in glucose that were independent of the use of insulin, we analyzed our data according to the onset of DIND.

Eighty-two per cent of the DIND patients were treated with insulin infusion (9/11), and in almost all of them (8/9) the insulin infusion was started before the onset of DIND. Cerebral glucose tended to decrease 1 day before clinical manifestation of DIND (P = 0.068; Figure 7), suggesting that the decrease in glucose may not be caused by DIND but possibly by insulin infusion, which started in average 2 days before DIND. The lactate:pyruvate ratio and glycerol level (not shown) did not change during this short period of time; however, this is not entirely unexpected (Figure 7).

Hyperglycaemia is common in critically ill patients. After stroke, the in-hospital mortality was shown to be lowest in normoglycaemic patients (1.7%), twice as high in diabetes patients (3%) and increased by a factor of 10 in patients with stress-induced hyperglycaemia (16%) [18]. In the latter group, insulin treatment became necessary thrice as frequently. Because hyperglycaemia appears to be injurious in conditions of brain ischaemia, the findings of others and our own group suggest that avoidance of hyperglycaemia should be a general strategy in SAH patients [1, 2, 16, 19, 20]. However, until now it was unclear whether this strategy could be endorsed for application in patients with low extracellular glucose levels. Data from TBI patients support the concern that a reduction in serum glucose could create substrate limitation in the injured brain [9]. In the present study, during continuous insulin infusion, there was a decrease in cerebral glucose levels in 79% of patients, similar to the percentage previously observed in TBI patients (79%) [9]. This decrease in cerebral glucose was associated with deterioration in homeostasis of cerebral metabolism in TBI patients.

No episodes of insulin-induced hypoglycaemia (lowest value 80 mg/dl) occurred during 10 days after SAH. It is difficult to determine what level of glucose might be too low for adequate brain function, especially because in these patients the demand for energy may be increased regionally. Dynamic PET scanning of labelled water (H₂¹⁵O) and deoxyglucose (¹⁸FDG) has revealed regional differences in cerebral metabolic capacity that may explain why cerebral cortex is more sensitive to an impaired glucose metabolism than other brain regions such as cerebellum [17]. It was shown that low cerebral glucose levels (<0.2 mmol/l) correlate with unfavourable outcome in TBI patients [7], which is supported by our clinical experience; hence, the suggested levels of 0.6 mmol/l might serve as an applicable threshold for a critical decrease in cerebral glucose [8]. In most patients (insulin treated: 83%; noninsulin treated: 77%) critically low cerebral glucose values were identified, but they occurred as late as 4.1 ± 2.3 days after the start of insulin infusion. The long time interval (>4 days) and the lack of any difference in their incidence between insulin-treated and noninsulin-treated patients suggest that these low cerebral glucose levels were unrelated to insulin treatment.

There are some limitations of the present study. Microdialysis is a regional method, and the volume of brain tissue monitored by the microdialysis catheter covers only a few millimeters from the membrane. Incomplete recovery affects the measured absolute concentrations and impairs the comparison with blood data [10]. Additionally, because blood glucose values were monitored discontinuously and not hourly, like the microdialysates, some effects of insulin on blood glucose might not have been detected in these patients.

Our results support the view of Vespa and colleagues [9] that forced normalization of elevated blood glucose using insulin might be deleterious for brain glucose concentrations and provoke further metabolic crisis. This, however, can only be evaluated in future studies; further research, ideally in combination with PET or cerebral blood flow studies, might investigate whether a normalization of critically low microdialysis glucose levels can improve outcome in patients with cerebral lesions.