Researchers at the Johns Hopkins Hospital, Baltimore, studied retrospectively the comparative efficacy of six most frequently used anticonvulsants when employed in combination with the ketogenic diet (KD) for treatment of 115 children with epilepsy. Mean age at initiation of the KD was 4.7 years. Patients had tried unsuccessfully a median of 4 anticonvulsants, and at KD initiation were receiving a median of 2 anticonvulsants (range 1- 5). At KD onset, the most common anticonvulsants included valproic acid (n=38), topiramate (31), levetiracetam (27), lamotrigine (25), zonisamide (21), and phenobarbital (14). Only 4 children received vigabatrin. Most common seizure types treated with drug/KD combination included Lennox-Gastaut syndrome or mixed/multiple seizures (n=56), infantile spasms (18), and complex partial seizures (19). After 3 months on the diet and no change in the anticonvulsant dose, 72% had a >50% seizure reduction. Patients receiving zonisamide and KD were more likely to have a >50% reduction in seizures than the other children combined who were receiving the other 5 anticonvulsants (P=0.04). Nineteen of the 21 children (90%) receiving zonisamide had a >50% seizure reduction. Children receiving phenobarbital and KD were less likely to have a >50% seizure reduction (P=0.003). The difference in the interaction between KD and zonisamide or phenobarbital was not explained by seizure type or age. Patients responding with a >90% seizure reduction or seizure freedom showed no significant correlation with a specific anticonvulsant/KD combination. 
COMMENT. Zonisamide is not approved for use in children, and its mechanism of action is not definitely known. It is thought to increase seizure threshold by effects on sodium and calcium channels. As a carbonic anhydrase inhibitor, zonisamide is less active than acetazolamide, but this mechanism may have a contributory anticonvulsant effect.
Controlled clinical balance studies of the effects of anticonvulsant drugs and the ketogenic diet (KD) on acid-base, electrolyte, and amino acid metabolism in children with absence seizures [2, 3] found that the KD and acetazolamide, studied as monotherapies, had similar metabolic effects. They both caused a metabolic acidosis, with decreased pH, pCO2, and standard bicarbonate, and a negative balance of electrolytes. In contrast, trimethadione, mephobarbital, and methsuximide as monotherapies (anti-‘petit mal’ medications available in the 1960s), caused a metabolic alkalosis, with an elevation of pH and standard bicarbonate, and compensatory rise in pCO2; the urinary excretion of sodium and potassium and fecal excretion of calcium, magnesium and phosphorus were reduced, and the balance of electrolytes was positive. The effects of the ketogenic diet and acetazolamide on acid-base and electrolyte balance were the reverse of those obtained during treatment with conventional antiepileptic medications and corticotropin. Apart from an increase in serum leucine during treatment with the KD, levels of serum amino acids showed no significant changes. In this study, while all therapies were beneficial, the KD was most effective in the control of absence seizures and reduction of epileptiform discharges in the EEG. Concurrent use of a carbonic anhydrase inhibitor (acetazolamide or zonisamide) and KD would result in accentuated effects on acid-base and electrolyte metabolism and potential improvement in seizure control, but not without an anticipated increase in adverse side effects.