Impact of hyperglycemia on neuropathological alterations during critical illness.

Context: Although preventing excessive hyperglycemia during critical illness may provide clinical neuroprotection, it remains debated whether normoglycemia is without risk for the brain.

Objective: To address this question, we compared the neuropathological alterations in microglia, astrocytes, and neurons, with uncontrolled hyperglycemia, moderately controlled hyperglycemia, and normoglycemia during human critical illness. We further investigated the time course in an animal model.

Design And Setting: We analyzed brain specimens from patients who died in the intensive care unit and from critically ill rabbits randomized to hyper- or normoglycemia. PATIENTS/OTHER PARTICIPANTS: We compared 10 critically ill patients randomized to normoglycemia (104 ±9 mg/dl) or moderate hyperglycemia (173 ±32 mg/dl), and five patients with uncontrolled hyperglycemia (254 ±83 mg/dl) with 16 controls (out of hospital sudden deaths). Critically ill rabbits were randomized to hyperglycemia (315 ±32 mg/dl) or normoglycemia (85 ±13 mg/dl) and studied after 3 and 7 d.

Interventions: Insulin was infused to control blood glucose.

Main Outcome Measures And Results: Patients with uncontrolled hyperglycemia showed 3.7-6-fold increased microglial activation, 54-95% reduced number and activation of astrocytes, more than 9-fold increased neuronal and glial apoptosis, and a 1.5-2-fold increase in damaged neurons in hippocampus and frontal cortex (all P ≤ 0.05). Most of these abnormalities were attenuated with moderate hyperglycemia and virtually absent with normoglycemia. Frontal cortex of hyperglycemic rabbits that had been critically ill for 3 d only revealed microglial activation, followed after 7 d by astrocyte and neuronal abnormalities similar to those observed in patients, all prevented by normoglycemia.

Conclusions: Preventing hyperglycemia with insulin during critical illness reduced neuropathological abnormalities, with microglial activation being the earliest preventable event. Whether these pathological findings associate with neurological outcome remains unknown.