Serum bicarbonate (HC03) levels were measured before, during, and after discontinuing topiramate (TPM) as adjunctive therapy for medically refractory epilepsy in 30 children treated at Children’s Hospital, Boston, MA. TPM dose varied from 50 to 650 mg/day, and 2 to 32 mg/kg/day. Larger doses resulted in higher TPM levels and greater likelihood of low HC03 levels. Two-thirds had a greater than 10% decrease in HC03 levels; the reduction was by 8 and 10 mEq/L in 2 cases. None was symptomatic. TPM was discontinued in 7 because of ineffectiveness, and in 2 with anorexia. The serum HC03 returned to pretreatment levels. Monitoring HC03 levels may be indicated, especially in patients predisposed to acidosis. 
COMMENT. Topiramate is a carbonic anhydrase inhibitor, and a decrease in serum HC03 levels with metabolic acidosis may be expected. This study has documented the generally asymptomatic association of lowered HC03 levels in children treated with adjunctive topiramate therapy for refractory epilepsy. Various mechanisms of action have been proposed for TPM. The carbonic anhydrase inhibitory activity is generally described as weak, but is obviously sufficient to lower HC03 levels in a substantial proportion of patients treated.
The mechanism of the anticonvulsant action of acetazoleamide, a potent carbonic anhydrase inhibitor, was first demonstrated in 1954 in animal studies at the Department of Pharmacology, University of Utah (Proceedings of the American Epilepsy Society meeting, New York, Dec 1954) . The anticonvulsant effect is related to the direct inhibition of carbonic anhydrase in the brain, and is independent of the metabolic acidosis which results from inhibition of the enzyme in the kidney. The anticonvulsant effect is not abolished in nephrectomized animals. Metabolic acidosis secondary to acetazolamide (or topiramate) has a very weak anticonvulsant effect, as demonstrated by studies with ammonium chloride in animals tested with electroshock seizures and in the clinic . A supplement of bicarbonate, as proposed by Takeoka and associates, in some children treated with TPM should not appreciably alter the anticonvulsant effect.