Determination of DP-VPA and its active metabolite, VPA, in human plasma, urine, and feces by UPLC-MS/MS: A clinical pharmacokinetics and excretion study.

DP-VPA is a phospholipid prodrug of valproic acid (VPA) that is developed as a potential treatment for epilepsy. To characterize the pharmacokinetics and excretion of DP-VPA, four reliable ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) methods were validated for quantitation of DP-VPA and its metabolite, VPA, in human plasma, urine, and feces. Protein precipitation and solid-phase extraction (SPE) were used for extraction of C16, C18 homologs of DP-VPA and VPA, respectively, from plasma. Urine and fecal homogenate involving the three analytes were efficiently prepared by methanol precipitation. The determinations of C16 DP-VPA, C18 DP-VPA, and VPA were performed using the positive multiple reaction monitoring (MRM) mode and the negative single ion monitoring (SIM) mode, respectively. The analytes were separated using gradient elution on C8 or phenyl column. Satisfactory results pertaining to selectivity, linearity, matrix effect, accuracy and precision, recovery, stability, dilution integrity, carryover, and incurred sample analysis (ISR) were obtained. The calibration ranges in human plasma were as follows: 0.00200-1.00 µg/mL for C16 DP-VPA, 0.0100-5.00 µg/mL for C18 DP-VPA, and 0.0500-20.0 µg/mL for VPA. The linear ranges in urine and fecal homogenate were 0.00500-2.00 µg/mL and 0.00200-0.800 µg/mL for C16 DP-VPA, 0.00500-2.00 µg/mL and 0.0100-4.00 µg/mL for C18 DP-VPA, and 0.200-80.0 µg/mL for VPA, respectively. The intra- and inter-batch coefficients of variation in three matrices ranged from 1.7% to 12.4% while the accuracy values ranged from 85.4% to 111.7%. The developed methods were successfully applied to determine pharmacokinetics of DP-VPA tablet after a single oral dose of 1200 mg in 12 healthy Chinese subjects under fed condition.

Risk factors for sodium valproate-induced renal tubular dysfunction.

OBJECTIVE: To explore the risk factors for the development of sodium valproate (VPA)-induced renal tubular dysfunction for early diagnosis and treatment. STUDY DESIGN: The subjects were selected from patients who were diagnosed with epilepsy and administered VPA. Blood and spot urine samples were collected and measured the concentration of VPA, the level of serum phosphorus, serum uric acid, serum free carnitine, serum cystatin-c, and urine ß2-microglobulin (BMG). Patients with urine BMG/creatinine levels above 219.2 were treated as renal proximal tubular dysfunction (RTD), with all others treated as non-RTD. RESULTS: Eighty-seven patients, 4-48 years, 53 men and 34 women, were studied. RTD group is 17 patients and non-RTD group is 70 patients. Univariate analyses revealed that the RTD patients were more likely to be bedridden, receiving enteral tube feeding, taking more anticonvulsants, and demonstrating significantly lower serum levels of free carnitine, uric acid, and phosphorus. Among them, bedridden, free serum carnitine, and phosphorus levels were associated with the development of RTD by multivariate analysis. CONCLUSIONS: Bedridden patients receiving VPA are susceptible to hypocarnitinemia, which can cause RTD and may lead to FS. Therefore, urinary BMG should be measured regularly in all patients receiving VPA to assess renal tubular function. An additional measurement of serum free carnitine level should be considered in patients who developed RTD. Supplementation of carnitine for those patients to prevent such complication deserves for further study.

Challenges for Detecting Valproic Acid in a Nontargeted Urine Drug Screening Method.

BACKGROUND: Valproic acid (VPA) is a widely prescribed medicine, and acute toxicity is possible. As such, it should be included in any nontargeted urine drug screening method. In many published liquid chromatography-electrospray ionization-mass spectrometry (LC-ESI-MS/MS) methods, VPA is usually measured using a pseudo-multiple reaction monitoring (MRM) transition. We investigate a simple ultra-high-performance liquid chromatography-quadrupole time-of-flight (QTof) approach to detect the presence of VPA with more confidence. METHODS: Three commercially sourced VPA metabolites were characterized and added to a nontargeted high-resolution MS urine drug screening method. All analyses were performed on a Waters Xevo G2-XS LC-QTof in negative electrospray ionization mode. The mass detector was operated in MS mode, and data were processed with UNIFI software. Sixty-eight patient urine samples, which were previously identified by a well-established gas chromatography-MS method as containing VPA, were analyzed on the Waters Xevo G2-XS LC-QTof, to validate this approach. RESULTS: VPA metabolite standards were characterized, and their detection data were added to the broad drug screening library. VPA metabolites were readily detectable in the urine of patients taking VPA. CONCLUSIONS: The inclusion of characterized VPA metabolites provides a simple and reliable method enabling the detection of VPA in nontargeted urine drug screening.

Observation of Clinically Relevant Drug Interaction in Chimeric Mice with Humanized Livers: The Case of Valproic Acid and Carbapenem Antibiotics.

BACKGROUND AND OBJECTIVE: Human in vitro and dog in vitro/in vivo researches indicate that the drug-drug interaction (DDI) of decreased plasma valproic acid (VPA) concentration by co-administration of carbapenem antibiotics is caused by inhibition of acylpeptide hydrolase (APEH)-mediated VPA acylglucuronide (VPA-G) hydrolysis by carbapenems. In this study, we investigated VPA disposition and APEH activities in TK-NOG chimeric mice, whose livers were highly replaced with human hepatocytes, to evaluate the utility of this animal model and the clinical relevance of the DDI mechanism. METHODS: VPA and VPA-G concentrations in plasma, urinary excretion of VPA-G and APEH activity in humanized livers were measured after co-administration of VPA with meropenem (MEPM) to chimeric mice. RESULTS: After co-administration with MEPM to the chimeric mice, plasma VPA concentration more rapidly decreased than without the co-administration. An increase in plasma AUC and urinary excretion of VPA-G was also observed. APEH activity in humanized livers was strongly inhibited even at 24 h after co-administration of MEPM to the chimeric mice. CONCLUSION: The DDI of VPA with carbapenems was successfully observed in chimeric mice with humanized livers. The DDI was caused by long-lasting inhibition of hepatic APEH-mediated VPA-G hydrolysis by carbapenems, which strongly supports the APEH-mediated mechanism of the clinical DDI. This is the first example showing the usefulness of chimeric mice with humanized livers for evaluation of a DDI via non-cytochrome P450 enzyme.

Influence of acylpeptide hydrolase polymorphisms on valproic acid level in Chinese epilepsy patients.

BACKGROUND: Concomitant use of meropenem (MEPM) can dramatically decrease valproic acid (VPA) plasma level. It is accepted that inhibition in acylpeptide hydrolase (APEH) activity by MEPM coadministration was the trigger of this drug-drug interaction. AIM: To investigate the influence of APEH genetic polymorphisms on VPA plasma concentration in Chinese epilepsy patients. PATIENTS & METHODS: Urinary VPA-d6 ß-D-glucuronide concentration was determined in 19 patients with VPA treatment alone (n = 10) or concomitant use with MEPM (n = 9). A retrospective study was performed on 149 epilepsy patients to investigate the influence of APEH polymorphisms rs3816877 and rs1131095 on adjusted plasma VPA concentration (CVPA) at steady-state. RESULTS: Urinary VPA-d6 ß-D-glucuronide (VPA-G) concentration was increased significantly in patients with MEPM coadministration. The CVPA of patients carrying the APEH rs3816877 C/C genotype was significantly higher than that of C/T carriers, and the difference was still obvious when stratified by UGT2B7 rs7668258 polymorphism. CONCLUSION: APEH polymorphism has significant influence on VPA pharmacokinetics in Chinese population.

In vivo inhibition of acylpeptide hydrolase by carbapenem antibiotics causes the decrease of plasma concentration of valproic acid in dogs.

1. Our previous in vitro studies suggest that inhibition of the acylpeptide hydrolase (APEH) activity as valproic acid glucuronide (VPA-G) hydrolase by carbapenems in human liver cytosol is a key process for clinical drug-drug interaction (DDI) of valproic acid (VPA) with carbapenems. Here, we investigated whether in vivo DDI of VPA with meropenem (MEPM) was caused via inhibition of APEH in dogs. 2. More rapid decrease of plasma VPA levels and increased urinary excretion of VPA-G were observed after co-administration with MEPM compared with those after without co-administration, whereas the plasma level and bile excretion of VPA-G showed no change. 3. Dog VPA-G hydrolase activity, inhibited by carbapenems, was mainly located in cytosol from both the liver and kidney. APEH-immunodepleted cytosols lacked VPA-G hydrolase activity. Hepatic and renal APEH activity was negligible even at 24 h after dosing of MEPM to a dog. 4. In conclusion, DDI of VPA with carbapenems in dogs is caused by long-lasting inhibition of APEH-mediated VPA-G hydrolysis by carbapenems, which could explain the delayed recovery of plasma VPA levels to the therapeutic window even after discontinuation of carbapenems in humans.

Urinary excretion and metabolism of miglustat and valproate in patients with Niemann-Pick type C1 disease: One- and two-dimensional solution-state (1)H NMR studies.

Niemann-Pick type C1 (NP-C1) disease is a neurodegenerative lysosomal storage disease for which the only approved therapy is miglustat (MGS). In this study we explored the applications and value of both one- and two-dimensional high-resolution NMR analysis strategies to the detection and quantification of MGS and its potential metabolites in urine samples collected from NP-C1 disease patients (n=47), and also applied these techniques to the analysis of the anticonvulsant drug valproate and one of its major metabolites in ca. 30% of these samples (i.e. from those who were also receiving this agent for the control of epileptic seizures). A combination of high-resolution 1D and 2D TOCSY/NOESY techniques confirmed the identity of MGS in the urinary (1)H NMR profiles of NP-C1 patients treated with this agent (n=25), and its quantification was readily achievable via electronic integration of selected 1D resonance intensities. However, this analysis provided little or no evidence for its metabolism in vivo, observations consistent with those acquired in corresponding experiments performed involving an in vitro microsomal system. Contrastingly, the major valproate metabolite 1-O-valproyl-ß-glucuronide was readily detectable and quantifiable in 14/47 of the urine samples investigated, despite some resonance overlap problems (identification of this agent was confirmed by experiments involving equilibration of these samples with ß-glucuronidase, a process liberating free valproate). In order to facilitate and validate the detection of MGS in urine specimens, full assignments of the (1)H NMR spectra of MGS in both buffered aqueous (pH 7.10) and deuterated methanol solvent systems were also made. The pharmacological and bioanalytical significance of data acquired are discussed, with special reference to the advantages offered by high-resolution NMR analysis.

Ultrasound-assisted dispersive liquid-liquid microextraction followed by GC-MS/MS analysis for the determination of valproic acid in urine samples.

BACKGROUND: Valproic acid (VPA) is an anticonvulsant drug used for the treatment of epilepsy and bipolar disorder. A method based on simultaneous derivatization and dispersive liquid-liquid microextraction followed by GC-MS/MS analysis has been developed for the determination of VPA in urine samples. RESULTS: This optimized and validated method shows good linearity with R(2) value of 0.999. LOD and LOQ of VPA was found to be 0.4 ng ml(-1) and 1.4 ng ml(-1), respectively. Recovery of VPA was found to be in the range of 80 to 92%. CONCLUSION: The developed method can find its wide applicability for the routine analysis of VPA in toxicological and clinical laboratories.

Dysregulations of UDP-glucuronosyltransferases in rats with valproic acid and high fat diet induced fatty liver.

Both high fat diet (HFD) and valproic acid (VPA) interfere with mitochondrial ß-oxidation of fatty acids, which subsequently triggers microvesicular fatty liver and hepatic dysfunction. UDP-glucuronosyltransferases, the major phase II drug metabolism enzymes, play a pivotal role in detoxifying various exogenous and endogenous compounds. This study aimed to investigate the dysregulation patterns of major UDP-glucuronosyltransferases (UGTs) induced by VPA and/or HFD. Biochemical and histopathological results showed that chronic treatments of VPA and HFD induced fatty liver and liver dysfunction in a synergistic manner. VPA upregulated the mRNA levels of UGT1A1, 1A6, 1A7, and UGT2B1. Notably, the protein expression and enzymatic activity of UGT1A6 were significantly increased in rats treated with HFD or VPA alone, and were further enhanced by HFD and VPA co-treatment. This dysregulation pattern was largely recapitulated in the in vitro HepG2 cells assay by using VPA and oleic acid treatment. Moreover, the induction of UGTs was accompanied by the increased expression of constitutive androstane receptor (CAR) and peroxisome proliferator-activated receptor α (PPARα). In line with the up-regulation of UGT1A1 and UGT1A6, urine recovery of VPA glucuronide (VPA-G) was sharply increased by VPA treatment, and the co-treatment of HFD further aggravated this change. Since VPA is necessarily prescribed for long-term and the prevalence of HFD life style nowadays, the combined effect of HFD and VPA on disturbing UGTs should take concerns in the clinics.

Study of valproic acid-induced endogenous and exogenous metabolite alterations using LC-MS-based metabolomics.

BACKGROUND: Valproic acid (VPA; an anticonvulsant drug) therapy is associated with hepatotoxicity as well as renal toxicity. An LC-MS-based metabolomics approach was undertaken in order to detect urinary VPA metabolites and to discover early biomarkers of the adverse effects induced by VPA. RESULTS: CD-1 mice were either subcutaneously injected with 600-mg VPA/kg body weight or vehicle only, and urine samples were collected at 6, 12, 24 and 48 h postinjection. A metabolomics approach combined with principal component analysis was utilized to identify VPA-related metabolites and altered endogenous metabolites in urine. Some VPA metabolites indicated potential liver toxicity caused by VPA administration. Additionally, some altered endogenous metabolites suggested that renal function might be perturbed by VPA dosing. CONCLUSION: LC-MS-based metabolomics is capable of rapidly profiling VPA drug metabolites and is a powerful tool for the discovery of potential early biomarkers related to perturbations in liver and kidney function.

The relationship between glucuronide conjugate levels and hepatotoxicity after oral administration of valproic acid.

The aim of this study was to investigate the relationship between hepatotoxicity, levels of glucuronide conjugates of valproic acid (VPA), and the toxic metabolites of VPA (4-ene VPA and 2,4-diene VPA). We also examined whether hepatotoxicity could be predicted by the urinary excretion levels of VPA and its toxic metabolites. VPA was administrated orally in rats in amounts ranging from 20 mg/kg to 500 mg/kg. Free and total (free plus glucuronide conjugated) VPA, 4-ene VPA, and 2,4-diene VPA were quantified in urine and liver using gas chromatography-mass spectrometry. Serum levels of aspartate aminotransferase, alanine aminotransferase, and alpha-glutathione S-transferase (alpha-GST) were also determined to measure the level of hepatotoxicity. The serum alpha-GST level increased slightly at the 20 mg/kg dose, and substantially increased at the 100 and 500 mg/kg dose; aspartate aminotransferase and alanine aminotransferase levels did not change with the administration of increasing doses of VPA. The liver concentration of free 4-ene VPA and the urinary excretion of total 4-ene VPA were the only measures that correlated with the increase in the serum alpha-GST level (p < 0.094 and p < 0.023 respectively). From these results, we conclude that hepatotoxicity of VPA correlates with liver concentration of 4-ene VPA and can be predicted by the urinary excretion of total 4-ene VPA.

Rapid detection of drugs in biofluids using atmospheric pressure chemi/chemical ionization mass spectrometry.

We have demonstrated that, with simple pH adjustment, volatile drugs such as methamphetamine, amphetamine, 3,4-methylenedioxymethamphetamine (MDMA), ketamine, and valproic acid could be analyzed rapidly from raw biofluid samples (e.g. urine and serum) without dilution, or extraction, using atmospheric pressure ionization. The ion source was a variant type of atmospheric pressure chemical ionization (APCI) that used a dielectric barrier discharge (DBD) to generate the metastable helium gas and reagent ions. The sample solution was loaded in a disposable glass pipette, and the volatile compounds were purged by nitrogen gas to be reacted with the metastable helium gas. The electrodes of the DBD were arranged in such a way that the generated glow discharge was confined within the discharge tube and was not exposed to the analytes. A needle held at 100-500 V was placed between the ion-sampling orifice and the discharge tube to guide the analyte ions into the mass spectrometer. After pH adjustment of the biofluid sample, the amphiphilic drugs were in the form of a water-insoluble oil, which could be concentrated on the liquid surface. By gentle heating of the sample to increase the evaporation rate, rapid and sensitive detection of these drugs in raw urine and serum samples could be achieved in less than 2 min for each sample.

Effect of antiepileptic drug polytherapy on urinary pH in children and young adults.

OBJECTS: The relationship between antiepileptic drugs (AEDs) polytherapy and urinary pH was studied to demonstrate the effect and difference of AED polytherapy compared to monotherapy. MATERIALS AND METHODS: A total of 271 urine samples from patients receiving AED polytherapy aged from 7 months to 35 years were enrolled. Two AEDs were co-administered to 215 patients, three AEDs to 45 patients, four AEDs to ten patients, and five AEDs to one patient. RESULTS: The distribution of urinary pH shifted to the alkaline range with increasing numbers of co-administered AEDs (p < 0.0001). The distribution of urinary pH shifted to the alkaline side with AED polytherapy that included valproate (p < 0.05) or acetazolamide (p < 0.03). The distribution of urinary pH did not change with or without zonisamide, carbamazepine, phenobarbital, phenytoin, or clonazepam. CONCLUSIONS: Urinary pH should be monitored in patients receiving AED polytherapy, particularly those receiving valproate, acetazolamide, or various AEDs in combination.

Interaction between valproic acid and carbapenem antibiotics.

The serum concentration of valproic acid (VPA) in epilepsy patients decreased by the administration of carbapenem antibiotics, such as meropenem, panipenem or imipenem, to a sub-therapeutic level. This review summarized several case reports of this interaction between VPA (1-4 g dose) and carbapenem antibiotics to elucidate the possible mechanisms decreasing VPA concentration by carbapenem antibiotics. Studies to explain the decrease were carried out using rats by the following sites: absorption of VPA in the intestine, glucuronidation in the liver, disposition in blood and renal excretion. In the intestinal absorption site, there are two possible mechanisms: inhibition of the intestinal transporter for VPA absorption by carbapenem antibiotics, and the decrease of beta-glucuronidase supplied from enteric bacteria, which were killed by antibiotics. This is consistent with a view that the decrease of VPA originated from VPA-Glu, relating to entero-hepatic circulation. The second key site is in the liver, because of no decreased in VPA level by carbapenem antibiotics in hepatectomized rats. There are three possible mechanisms in the liver to explain the decreased phenomenon: first, decrease of the UDPGA level by carbapenem antibiotics. UDPGA is a co-factor for UDP-glucuronosyltransferase (UGT)-mediated glucuronidation of VPA. Second, the direct activation of UGT by carbapenem antibiotics. This activation was observed after pre-incubation of human liver microsomes with carbapenem antibiotics. Third, the inhibition of beta-glucuronidase in liver by carbapenem antibiotics and the decreased VPA amount liberated from VPA-Glu. The third site is the distribution of VPA in blood (erythrocytes and plasma). Plasma VPA distributed to erythrocytes by the inhibition of transporters (Mrp4), which efflux VPA from erythrocytes to plasma, by carbapenem antibiotics. The increase of renal excretion of VPA as VPA-Glu depends on the increase of VPA-Glu level by UGT. One or a combination of some factors in these mechanisms might relate to the carbapenem-mediated decrease of the plasma VPA level.

Kinetics of valproic acid glucuronidation: evidence for in vivo autoactivation.

Sigmoidal or autoactivation kinetics has been observed in vitro for both cytochrome P450- and UDP-glucuronosyltransferase-catalyzed enzymatic reactions. However, the in vivo relevance of sigmoidal kinetics has never been clearly demonstrated. In the current study we investigate the kinetics of valproic acid glucuronide (VPAG) formation both in vivo in adult sheep and in vitro in sheep liver microsomes (pool of 10). After a 100 mg/kg i.v. bolus dose of valproic acid (VPA) to adult sheep (n = 5), the majority of the dose was recovered in urine as VPAG (approximately 79%). Eadie-Hofstee plots of the VPAG formation rate (calculated from urinary excretion rate data for VPAG) were characteristic of autoactivation kinetics and provided estimates of the apparent maximum velocity of an enzymatic reaction (V(max)(app)), the substrate concentration resulting in 50% of V(max)(app) (S(50)(app)), and Hill coefficient (n) of 2.10 +/- 0.75 micromol/min/kg, 117 +/- 56 microM, and 1.34 +/- 0.14, respectively. Comparable estimates of V(max)(app) (2.63 +/- 0.33 micromol/min/kg), S(50)(app) (118 +/- 53 microM), and n (2.06 +/- 0.47) describing overall VPA elimination from plasma were obtained by fitting VPA unbound plasma concentration-time data to a two-compartment model with elimination described by the Hill equation. Consistent with our in vivo observations, Eadie-Hofstee plots of VPAG formation in sheep liver microsomes were characteristic of autoactivation kinetics. To our knowledge, these data provide the first clear demonstration that autoactivation kinetics observed in vitro in liver preparations can translate to the in vivo situation at least under certain experimental conditions and confirm its relevance.

Direct determination of valproic acid in biological fluids by capillary electrophoresis with contactless conductivity detection.

Capacitively coupled contactless conductivity detection (C(4)D) is a new technique providing high sensitivity in capillary electrophoresis (CE) especially for small ions that can otherwise only be determined with indirect methods. In this work, direct determination and validation of valproic acid (VPA) in biological fluids was achieved using CE with C(4)D. VPA is of pharmacological interest because of its use in epilepsy and bipolar disorder. The running electrolyte solution used consisted of 10mM 2-(N-morpholino)ethane sulfonic acid (MES)/dl-histidine (His) and 50microM hexadecyltrimethylammonium bromide (HTAB) at pH 6.0. Caproic acid (CA) was selected as internal standard (IS). Analyses of VPA in serum, plasma and urine samples were performed in less than 3min. The interference of the sample matrix was reduced by deproteinization of the sample with acetonitrile (ACN). The effect of the solvent type and ratio on interference was investigated. The limits of detection (LOD) and quantitation (LOQ) of VPA in plasma samples were determined as 24 and 80ng/ml, respectively. The method is linear between the 2 and 150microg/ml, covering well the therapeutic range of VPA (50-100microg/ml).

Oxidative stress in children receiving valproic acid.

OBJECTIVE: To determine whether valproic acid (VPA) influences urinary levels of 15-F2t -isoprostane (15-F2t -IsoP), a marker of oxidative stress, in children. STUDY DESIGN: Morning urine samples were collected from children with epilepsy receiving VPA (n = 25), carbamazepine (n = 16), or clobazam (n = 12) for > or = 4 weeks and from age-matched control subjects (n = 39). Urinary 15-F2t -IsoP levels were determined by enzyme-linked immunosorbent assay. RESULTS: The mean (+/- standard deviation) urine 15-F2t -IsoP levels (nmol/mmol Cr) were: valproic acid (0.36 +/- 0.15); carbamazepine (0.24 +/- 0.10); clobazam (0.23 +/- 0.10); control group (0.20 +/- 0.09). Patients treated with VPA had significantly elevated 15-F2t -IsoP levels when compared with the control, carbamazepine, and clobazam groups (P < .05). Multiple linear regression analysis demonstrated that younger patient age and exposure to second-hand smoke were significant predictors of elevated urine 15-F2t -IsoP levels within the control group (r2 = 0.261, P = .05 and P = .01, respectively). Subjects not exposed to second-hand smoke receiving valproic acid therapy had a significantly elevated mean urine 15-F2t -IsoP level compared to subjects not exposed to second-hand smoke in the carbamazepine, clobazam and control groups (P < .05). CONCLUSIONS: These data demonstrate that treatment of children with VPA is associated with higher urinary levels of 15-F2t -IsoP, a marker of oxidative stress.

Direct observation of 3-keto-valproate in urine by 2D-NMR spectroscopy.

BACKGROUND: It is known that valproate and its metabolites cause hepatotoxicity. The drug monitoring of valproate is important to determine an effective dose to keep an appropriate concentration in blood. METHODS: In a 2-dimensional (2D)-NMR spectrum of double quantum filtered correlation spectroscopy (DQF-COSY), clear correlation peaks were ascertained to be due to 3-keto-valproate, which was a beta-oxidation product of valproate. RESULTS: A predominant metabolite of valproate was observed by proton NMR spectroscopy in the crude urine of a particular patient with metabolic disorder. The assignment of the signals was determined by synthesized 3-keto-valproic acid ethyl ester. The concentration of 3-keto-valproate in the urine was calculated to be 631 microg/mg creatinine by the integration of the peak of the isolated triplet methyl protons of C(5) at 1.016 ppm. CONCLUSION: Although the NMR spectra of crude urine of the patients who took valproate were usually complicated with many metabolites, the signals of 3-keto-valproate in a DQF-COSY spectrum of the urine of patients were easily connected according to the present assignment. The NMR analysis of the urine of patients who are prescribed valproate is useful for therapeutic drug monitoring and for checking the compliance of the patients.