Metabolomics As a Tool for the Characterization of Drug-Resistant Epilepsy

Abstract

Purpose: Drug resistance is a critical issue in the treatment of epilepsy, contributing to clinical emergencies and increasing both serious social and economic burdens on the health system. The wide variety of potential drug combinations followed by often failed consecutive attempts to match drugs to an individual patient may mean that this treatment stage may last for years with suboptimal benefit to the patient. Given these challenges, it is valuable to explore the availability of new methodologies able to shorten the period of determining a rationale pharmacologic treatment. Metabolomics could provide such a tool to investigate possible markers of drug resistance in subjects with epilepsy.

Methods: Blood samples were collected from (1) controls (C) (n = 35), (2) patients with epilepsy "responder" (R) (n = 18), and (3) patients with epilepsy "non-responder" (NR) (n = 17) to the drug therapy. The samples were analyzed using nuclear magnetic resonance spectroscopy, followed by multivariate statistical analysis.

Key findings: A different metabolic profile based on metabolomics analysis of the serum was observed between C and patients with epilepsy and also between R and NR patients. It was possible to identify the discriminant metabolites for the three classes under investigation. Serum from patients with epilepsy were characterized by increased levels of 3-OH-butyrate, 2-OH-valerate, 2-OH-butyrate, acetoacetate, acetone, acetate, choline, alanine, glutamate, scyllo-inositol (C < R < NR), and decreased concentration of glucose, lactate, and citrate compared to C (C > R > NR).

Significance: In conclusion, metabolomics may represent an important tool for discovery of differences between subjects affected by epilepsy responding or resistant to therapies and for the study of its pathophysiology, optimizing the therapeutic resources and the quality of life of patients.

Keywords: biomarkers; drug-resistant epilepsy; epilepsy; ketone bodies; metabolomics.

Figure 1

Scores plots obtained from nuclear magnetic resonance spectra of serum samples from controls and patients with epilepsy. (A) Scores plot from the multivariate orthogonal partial least square discriminant analysis model between controls (C): C (●) and patients with epilepsy: P (O): each point represents a single serum spectrum, with the position determined by the contribution of the 159 variables. (B) Validation of the corresponding model by permutation test (n = 500). (C) Scores plot from the multivariate orthogonal partial least square discriminant analysis of a three classes model: healthy subjects (●), responder (R) patients (Δ), and non-responder (NR) patients (). (D) Statistical validation of the corresponding model by permutation test.

Figure 2

Scores plots obtained from nuclear magnetic resonance spectra of serum samples from controls and responder (R) and non-responder (NR) patients. (A) Scores plot from the multivariate orthogonal partial least square discriminant analysis (OPLS-DA) model between controls (●) and NR patients (). (B) Statistical validation of the corresponding model by permutation test (n = 500). (C) OPLS-DA between controls (●) and R patient (Δ) and (D) statistical validation of the corresponding model by permutation test (n = 500). (E) OPLS-DA model between R (Δ) and NR patients (). (F) Statistical validation of the corresponding model by permutation test (n = 500).