Altered blood-nerve barrier permeability to small molecules in experimental diabetes mellitus.

Permeability-surface area products (PA) were determined with a quantitative in vivo injection technique at the blood-nerve barrier of tibial nerve, and at the blood-brain barrier, in control and streptozotocin-induced diabetic rats. The PA product for [14C]mannitol at the blood-nerve barrier was increased by 100% in diabetic animals, 3.12 +/- 0.15 X 10(-5) ml X s-1 X g-1, compared with controls, 1.61 +/- 0.10 X 10(-5) ml X s-1 X g-1. In contrast, PA for [14C]mannitol at the blood-brain barrier was unaltered in the diabetic animals. Following intravenous injection, no leakage of microperoxidase across the perineurium or endoneurial vessels of diabetic rats could be demonstrated by morphological techniques. Nerve blood-space, as determined with intravenous [3H]inulin, and blood-nerve barrier surface area as determined by morphometric methods, did not differ in diabetic when compared to control animals. Thus, the calculated permeability coefficient for [14C]mannitol at the blood-nerve barrier was about 100% greater in diabetic nerve compared to control nerves. The increased permeability was accompanied by a 7% increase in nerve-water content and a 32% decrease in motor-nerve conduction velocity. The results demonstrate a specific vulnerability of nerve as compared to brain in an animal model of diabetes mellitus. Chronically altered permeability to small water-soluble molecules reduces the protective effect of salt impermeability at the blood-nerve barrier against nerve edema, and may be an important pathogenic mechanism in diabetic neuropathy.