799 (figure) lasting from seconds to minutes; attacks frequently dependent on particular head position (5 patients); hyperacusis (5 patients) or tinnitus (7 patients) permanently or during the attack; and measurable auditory or vestibular deficits by neurophysiological methods (10 patients). Response to carbamazepine seemed essential since earlier studies on disabling positional vertigo emphasised the ineffectetiveness of vestibular suppressant medications1 without evidence that antiepileptic drugs (first choice for trigeminal neuralgia) were systematically taken. In audiovestibular testing, 10 of 11 patients exhibited unilateral dysfunction in the symptom-free interval. Carbamazepine was given to all patients orally at an initial dose of 100-200 mg three times daily. All patients responded promptly even to a low initial dose. 8 became symptom-free; 3 reported infrequent residual attacks. The efficacy of treatment Figure: Anterior (A) and posterior (P), and lateral (right [R], left [L]) postural sway In normal subject (top) and untreated patient with neurovascular cross-compression of the right eighth nerve. Postural sway obtained with force-measuring platform (Kistler, Ostfildern). Patient had slightly increased body sway with eyes closed (bottom left), which increased during the attack for 26-72 s (bottom right), usually in a diagonal AP direction. Direction of preferred body sway changed by 90° if head was turned to R or to L The effects were dependent on head position and combination of auditory symptoms (high-frequency hearing loss, tinnitus at 2000 Hz) with probable involvement of vertical canal and otolith input because of upward spontaneous nystagmus, ocular torsion of both eyes, and tilt of perceived vertical. I. has now been evaluated over 6 years in 1 patient and over 2-3 years in 5 others who seem to require a minimum dose of 200-400 mg per day. One or two wash-out phases were done in 4 patients, all of whom relapsed within 2 days to 2 weeks. The strongest argument for a peripheral nerve origin of the condition in our patients is based on the documented unilateral peripheral audiovestibular deficits in the symptom-free interval in 10 of 11 patients. This is supported by the long duration of monosymptomatic attacks over a mean of 7 years, which makes a central brainstem disorder less likely. None of our patients had an abnormal electroencephalogram, even when an occasional attack occurred during the recording, or multiple sclerosis, which was ruled out by magnetic resonance imaging, analysis of cerebrospinal fluid, and evoked potentials. The dependence of vestibular and auditory symptoms on head position is also indicative of a peripheral disorder. We recommend carbamazepine as first-choice therapy in suspected neurovascular cross-compression of the eighth nerve before an operation is contemplated. Th Brandt, M Dieterich Department of Neurology, University of Munich, Klinikum Grosshadern, Marchioninistrasse 15, 81377 München, Germany 1 Møller AR. The cranial nerve vascular compression syndrome, I: a review of treatment. Acta Neurochir 1991; 113: 18-23. 2 Jannetta PJ, Møller MB, Møller AR. Disabling positional vertigo. N Engl J Med 1984; 310: 1700-05. 3 Møller MB, Møller AR, Jannetta PJ, et al. Diagnosis and surgical treatment of disabling positional vertigo. J Neurosurg 1986; 64: 21-28. 4 Brandt Th, Dieterich M. Vertigo in neurovascular cross-compression, vestibular paroxysmia. Nervenarzt 1960; 61: 376-78. 5 Brandt Th. Vertigo: its multisensory syndromes. London: Springer, 1991. Peripheral lactate and neuronal metabolism SIR-Maran and colleagues (Jan 1, p 16) demonstrate that a moderate hyperlactataemia attenuates the counter-regulatory response to insulin-induced hypoglycaemia in healthy, non- diabetic volunteers. Subjective awareness of hypoglycaemia and the glucose threshold at which neuropsychological function deteriorated also fell. Maran et al suggest that lactate directly substitutes for glucose as a metabolic substrate for neurons. However, other mechanisms may be operating. The belief that brain relies largely on oxidative metabolism of glucose arose out of the classic observation that fuel uptake by brain from arterial blood could only be demonstrated for glucose and for ketone bodies after fasting. Furthermore, insulin-induced hypoglycaemic coma can be reversed by systemic administration of glucose or mannose (but not lactate). Neurons do indeed have the metabolic apparatus to utilise lactate. Schurr et all demonstrated that neuronal function could be maintained for many hours in in-vitro brain slice preparations by lactate alone. The evidence pointed to direct entry of lactate via pyruvate into the tricarboxylic acid cycle. However, the in vivo significance of such observations depends on blood-brain barrier facilitated transport processes for lactate. Kinetic data suggest that the capacity of the lactate transporter is very much lower than that for glucose. However, there may be situations where BBB permeability to lactate rises and brain lactate uptake becomes significant: for example, there is evidence for this in acutely hypoglycaemic diabetic dogs.2 Experiments suggest that individual brain regions enter a brief period of non-oxidative, glycolytic metabolism behind the blood-brain barrier upon activation. Transient increases in extracellular lactate are detectable. The subsequent return to normal lactate levels seems to be due to neuronal reuptake and metabolism rather than loss to the blood.3 Astrocyte glycogen