Acetone as an Anticonvulsant


  • Sergei Likhodii,

    1. Department of Pharmacology and University of Toronto Epilepsy Research Program (UTERP), University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Kirk Nylen,

    1. Department of Pharmacology and University of Toronto Epilepsy Research Program (UTERP), University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • W. McIntyre Burnham

    1. Department of Pharmacology and University of Toronto Epilepsy Research Program (UTERP), University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author

Address correspondence to W. McIntyre Burnham, Ph.D., Department of Pharmacology, Medical Sciences Building, 1 King's College Circle, Toronto, Ontario M5S 1A8, Canada. E-mail:


Recent interest in the anticonvulsant effects of acetone has stemmed from studies related to the ketogenic diet (KD). The KD, a high-fat diet used to treat drug-resistant seizures, raises blood and brain levels of three ketones: beta-hydroxybutyrate, acetoacetate, and acetone. An obvious question is whether these ketones have anticonvulsant properties. We found that neither beta-hydroxybutyrate nor acetoacetate has proven to be anticonvulsant. Acetone, however, is clearly anticonvulsant at physiological, and near-physiological, nontoxic concentrations. Despite knowledge of acetone's anticonvulsant properties since the 1930's, acetone had never been characterized using the standard animal seizure tests. In our recent experiments, acetone was found to be active in animal models of tonic–clonic seizures, typical absence seizures, complex partial seizures, and atypical absence seizures associated with Lennox–Gastaut syndrome. Therapeutic indices are either comparable or better than that of valproate, a standard broad-spectrum anticonvulsant. A number of acetone-like molecules have also been tested, and these also show good potency up to a “cutoff” point of nine carbons contained in the side chain. Above this number, potency disappears, suggesting the possibility of a receptor for acetone and its analogs.

Current interest in the anticonvulsant effects of acetone has stemmed from studies related to the ketogenic diet (KD). In a series of recent papers, Likhodii and colleagues have argued for the “acetone hypothesis” of KD action (Likhodii et al., 2000; Likhodii & Burnham 2002a, 2002b; Likhodii et al., 2003; Likhodii & Burnham, 2004). This hypothesis suggests that the KD stops seizures because it elevates blood and brain levels of the ketones, and specifically of acetone. Acetone is postulated to have direct anticonvulsant effects. The following discussion will review the KD, the ketone bodies it elevates, and the anticonvulsant properties of acetone.

The Ketogenic Diet and Ketone Bodies

The KD, an anticonvulsant diet, classically consists of four parts of fat to one part of combined carbohydrate and protein by weight (Vining, 1999). It forces the body to use fats as a major source of energy. As fats are metabolized in the liver, three metabolites of interest are produced: acetoacetate, beta-hydroxybutyrate, and acetone. These are termed “ketone bodies” (Fig. 1).

Figure 1.

Dose responses for acetone in animal models of seizures and epilepsy. Dose–response curves for acetone tested in four animal seizure models. Effective dose fifties (ED50's) indicate the dose that suppressed seizure activity in 50% of the subjects. From the data related to the kindling model, the open circles relate to the suppression of the generalized convulsion, and the filled circles relate to the suppression of the amygdala focal after-discharge. This figure is reproduced, with permission, from Likhodii et al., 2003.

Anticonvulsant Effects of the Ketones

We have recently investigated the anticonvulsant effects of the three ketone bodies. In our hands, neither acetoacetate nor beta-hydroxybutyrate had anticonvulsant properties (Likhodii, unpublished data). There may be some question about the status of acetoacetate, however, since one study reported anticonvulsant activity in an audiogenic seizure-susceptible mouse model (Rho et al., 2002). In a larger study involving several animal models of seizures, we confirmed that acetone clearly exhibited anticonvulsant efficacy at therapeutically relevant and nontoxic concentrations (Likhodii et al., 2003).

History of Acetone as an Anticonvulsant

The idea that elevations in blood ketone (especially, acetone) levels might be responsible for the KD's anticonvulsant effects is not new. Wilder (1921) first proposed an anticonvulsant action of ketones when he introduced the diet. The first experimental observation confirming Wilder's hypothesis was made by Keith (Keith, 1931, 1932, 1933). Testing a number of compounds against thujone-induced seizures in rabbits, Keith reported anticonvulsant effects of both acetone and acetoacetate. A subsequent study was done by Driver (1947). Apparently unaware of Keith's report, Driver was looking at the ability of compounds to raise seizure thresholds, using phenytoin as a positive control. Driver found that acetone raised seizure thresholds more effectively than phenytoin. Since the time of Driver, there have been more than fifteen studies reporting the anticonvulsant effects of acetone (for references, see Likhodii & Burnham, 2004). Thus, the “acetone hypothesis” dates back to the 1940s.

Acetone Has a Broad Spectrum of Anticonvulsant Actions

Our recent studies grew out of our interest in the KD. They have involved tests of acetone in the animals models used in drug development and have revealed a number of characteristics of acetone as an anticonvulsant. Like the KD, acetone appears to have a broad spectrum of anticonvulsant activity. In recent experiments, we have found it to be active in animal models of tonic–clonic seizures (the maximal electroshock model), typical absence seizures (the subcutaneous pentylenetetrazol model), complex partial seizures with secondary generalization (the amygdala kindling model), and atypical absence seizures associated with the Lennox–Gastaut syndrome (the AY9944 model) (Likhodii et al., 2003).

Dose–response curves derived from these studies are depicted in Fig. 1. Acetone was most effective against maximal electroshock seizures (MES) and AY9944 seizures. Like most anticonvulsants, it was least effective against focal seizures in the amygdala kindling model. Effective doses fifties (i.e., ED50's) are considerably higher than those of the standard anticonvulsants (Krall et al., 1978) – except perhaps for valproate (Table 1). These doses produced concentrations in plasma and cerebrospinal fluid of 1–30 mM, proportional to the dose (Likhodii et al., 2003). Acetone levels produced by the KD, incidentally, are in the range of 1–8 mM (Musa-Veloso et al., 2006). In our hands, acetone's metabolites have not proven to be anticonvulsant. It appears to be acetone itself that has anticonvulsant potency (Likhodii, unpublished observations). A recent study by Gasior et al. (2007) has reached similar conclusions .

Table 1.  ED50's for acetone in experimental models of epilepsy
ModelED50 (mg/kg)
Kindling (generalized)761
Kindling (focal)1,539  

Acetone Is Anticonvulsant at Nontoxic Doses

While ED50's were high, toxic dose fifties (TD50's) were also high, exceeding 1,500 mg/kg. Therapeutic indices (TD50/ED50's) ranged from 1.2 (kindled amygdala focus) to 6 (AY9944 model) (Likhodii & Burnham, 2004). Valproate, in comparison, had a therapeutic index of about 2 when tested in the same paradigm. To our knowledge, full toxicity testing – including testing of organ systems after acute, subchronic, and chronic dosing – has not been conducted for acetone. Children on the KD, however, maintain high acetone levels (Musa-Veloso et al., 2006) for years without apparent ill effects.

Tolerance to Acetone's Anticonvulsant Effects Has Not Been Tested

The question of whether acetone's anticonvulsant effects are long lasting, or whether tolerance (a loss of anticonvulsant activity) would develop, has not yet been addressed. Children, however, do not become tolerant to the anticonvulsant effects of the KD. If the KD's effects are mediated by acetone, then it is possible that tolerance would not develop to acetone's anticonvulsant effects either.

Acetone Probably Does Not Exhibit Major Drug Interactions

Anticonvulsant drug interactions are often related to hepatic metabolism, at least with respect to pharmacokinetics. Acetone is only partly metabolized. About 30% is excreted unchanged in the breath and urine. At physiological concentrations, the remaining 70% is metabolized by the liver by CYP2E1 (Chen et al., 1997). CYP2E1 is induced by acetone (Chen et al., 1997). CYP2E1, however, is only involved in the metabolism of a small number of drugs. Known substrates are the anesthetics, theophylline, and ethanol (Loi et al., 1989; Kharasch et al., 1994; Spracklin et al., 1997 ). Acetone, therefore, would not likely enter into interactions with the standard anticonvulsant drugs.

Acetone Levels Might Be Affected by Seizure Activity

A possible complication for the use of acetone in therapy is the fact that acetone levels might be affected by seizure activity. In animal models, acetone levels are reduced by 20% after seizure activity (Nylen, unpublished data). This might mean that doses would have to be adjusted upwards in patients whose seizures were not fully controlled.

Acetone-like Molecules Are Also Anticonvulsant

We have tested a number of acetone-like molecules in animal seizure models. The analogs differ from acetone in that they have an extended carbon side chain. Some of these compounds are more potent than acetone itself, and have a better therapeutic index (Likhodii & Burnham, unpublished data).

Acetone May Have an Endogenous Receptor

With acetone analogs, potency tends to increase up to 10 carbons, at which point it disappears. This suggests that there may be an acetone “receptor,” possibly a site located within a binding pocket of a specific size (Likhodii & Burnham, unpublished data).


Acetone has broad-spectrum anticonvulsant effects, which occur at nontoxic doses. This is a property of acetone itself, and not of its metabolites. Superior anticonvulsant effects are found with some acetone analogs that were formed by extending acetone's side chain. Acetone or its analogs might be developed for use in the treatment of epilepsy.


I confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. This research was supported by funds from the Canadian Institutes of Health Science and from the Michael Bahen Chair of Epilepsy Research. The authors would like to thank Ms. Heidi Wong for help in the preparation of the manuscript.

Disclosure: The authors declare no conflicts of interest.