Insulin has recently been shown to ameliorate damage in models of glob
al brain ischemia. To determine whether insulin is also neuroprotectiv
e in focal ischemia, 20 rats were given 2 to 3 IU/kg insulin and 10 di
d not receive treatment prior to normothermic transient middle cerebra
l artery occlusion for 2 hours at a blood pressure of 60 mm Hg. To fur
ther elucidate whether infarction volume is influenced by variations i
n blood glucose levels within the physiological range, blood glucose w
as raised in 10 of the insulin-treated animals to levels comparable wi
th the untreated controls. At 1-week survival, damage was assessed usi
ng quantitative neuropathological examination of 25 coronal planes. It
was found that preischemic insulin lowered the mean intraischemic blo
od glucose level from 8.4 +/- 0.2 mM (mu +/- standard error of the mea
n) in the control group to 3.4 +/- 0.2 mM and reduced total damage (at
rophy plus cortical and striatal necrosis), expressed as the percentag
e of the normal hemisphere, from a control of 28.5% +/- 2.9% to 14.5%
+/- 1.6% (p < 0.005). Coadministration of glucose and insulin resulted
in a mean intraischemic blood glucose level of 10.1 +/- 0.5 mM, with
27.0% +/- 2.4% total damage (p = 0.96, compared with control). Total i
schemic damage showed an independent correlation with blood glucose le
vels (r = 0.67, p = 0.0018). The findings indicate that insulin benefi
ts transient focal ischemia and that reducing the blood glucose from 8
to 9 mM to the low-normal range of 3 to 4 mM with insulin dramaticall
y reduces subsequent infarction. The data suggest that the neuroprotec
tive mechanism of insulin action in focal middle cerebral artery occlu
sion is mediated predominantly via alterations in blood glucose levels
. In comparison to global ischemia, focal ischemia appears to show onl
y a minor direct central nervous system effect of insulin. In clinical
situations in which transient focal ischemia to the hemisphere can be
anticipated, insulin-induced hypoglycemia of a mild degree may be ben
eficial.