Epilepsia. 19591 -602, 1978.

Raven Press, New yo&

Identification of Metabolites of Valproic Acid in Serum of Humans, Dog, Rat, and Mouse Cornelis Jakobs and Wolfgang Loscher Department of Pediatrics, Free University of Berlin, Kaiserin-AugusteVictoria-Haus, 0-1000 Berlin 19; and Department of Pharmacology and Toxicology, Free University of Berlin, School of Veterinary Medicine, 0-1000 Berlin 33, West Germany

et al. (1976) identified, in addition to the glucuronide acid conjugate of VPA, five In recent years valproic acid (written metabolic products of the drug in rat urine, variously as di-n-propylacetic acid, 2which indicated two main metabolic pathpropyl-valeric acid, and 2-propyl pentanoic ways. 4-Hydroxy-2-propyl pentanoic acid acid, or VPA) has found increasing use in (4-OH-VPA), 5-hydroxy-2-propyl pentathe treatment of generalized epilepsy, noic acid (5-OH-VPA), and 2-n-propylgluespecially of the petit ma1 type in children and-in combination with other anticonvul- taric acid were probably w (wl, w,)-oxidasants-of more difficult cases (for review tion products, whereas 3-hydroxy-2-propyl see Pinder et al., 1977). Despite an exten- pentanoic acid (3-OH-VPA), as well as 3sive literature on the experimental and keto-2-propyl pentanoic acid (3-keto-VPA) therapeutical aspects, comparatively little were metabolites of VPA via p-oxidation. Involvement of o-and P-oxidation in the attention has been paid to metabolism of of VPA was confirmed recently metabolism VPA, although the drug is entirely elimiin humans (Gompertz et a]., 1977; Kochen nated by metabolism. Renal excretion of et al., 1977). Since in epileptic patients the unchanged VPA in humans was found to be onset of the therapeutic effect of VPA is only 3-7% of the administered dose, observed only after several days or even whereas the main part of VPA was elimimore than a week of intermittent treatment, nated in the form of conjugates and unidentified metabolites (Schobben et al., it seems possible that a slowly accumulat1975; Gugler et al., 1977). Eymard et al. ing metabolite of the drug could be respon(1971) detected five metabolites in rat bile sible for its pharmacological effects. Howfollowing administration of I4C-VPA, but ever, to our knowledge, no study exists on did not determine their structures. Kuhara metabolites of VPA in the serum of human and Matsumoto (1974) and Matsumoto subjects or of experimentally used animals. This article describes the identification of several metabolites of VPA in serum of humans, dog, rat, and mouse. The metaboReceived November 1 1 , 1977. lites were identified by low- and highKey words: Valproic acid-Metabolites in serumresolution mass spectrometry. Studies in humans, dog, rat, and mouse.

INTRODUCTION

591

592

CORNELIS JAKOBS AND WOLFGANG LOSCHER

MATERIALS AND METHODS Patients

Blood samples were taken from 12 ambulatory children (5- 14 years of age) of either sex with generalized or partial epileptic seizures. VPA was the only medication (eight patients) or was added to the current treatment (phenytoin, primidone, ethosuximide) and had been given for at least 4 months. The prescribed dose of VPA (sodium salt) varied between 150 and 600 mg b i d . or t.i.d. and was administered in the form of commercially available 300mg tablets (ErgenyP). The mean serum concentration of VPA in these patients was 84 puglml (19-160 puglml). Animal Studies Three female mongrel dogs weighing 1222 kg received a rapid intravenous injection of 20 mg/kg VPA+. Blood samples were taken after 140 min. The mean serum concentration of VPA at this time was 13.5 puglml (12.0-14.3 pglml). Five male Sprague-Dawley rats, weighing 380-470 g, received 600 mdkg VPA+ orally, and blood was collected by decapitation after 2 hr. The mean serum concentration of VPA at this time was 50 puglml(42-58 puglml). One hundred male NMRI mice, weighing 25-32 g, received 400 mg/kg VPA' orally. The animals were bled by decapitation after 45 min and the blood of 20 mice was pooled for one determination. The mean serum concentration of VPA was 220 p d m l (180260 puglml). Chemicals

VPA was used in the commercial 30% solution form of the sodium salt (ErgenyP', Labaz GmbM, Dusseldorf). Chloroform and sulfuric acid were reagent grade and obtained from Merck. 4-Hydroxy-valproic acid-y-lactone and 5-hydroxy-valproate sodium were kindly provided by Dr. H . Schafer (Desitin Werk Carl Klinke, Hamburg).

' The sodium salt of VPA. Epilepsia, Vol. 79,December 1978

Preparation of Serum Samples

A modified method based on the procedure described by Loscher (1977) was used: The serum sample was acidified with 0.25 vol 12 N H,SO, and extracted with 0.5 vol chloroform. The mixture was shaken and centrifuged. The upper layer and the protein layer were removed. The organic phases of several samples were mixed and concentrated by evaporating in a stream of nitrogen. Ten p l of the sample was injected in the gas chromatography-mass spectrometry (GC-MS) combination. Gas Chromatography-Mass Spectrometry For combined gas-liquid chromatography and mass spectrometry a dual channel Varian 2740 gas chromatograph was used. One channel was fitted to a normal FID detector. The other channel was interfaced via a Biemann- Watson molecular separator to a double-focusing high-resolution mass spectrometer, Type 311 A (Varian MAT) normally operated with an ionization energy of 70 eV. Resolution was of the order of 1,000 in the low-resolution mode and 15,000 in the high-resolution mode. In the last mode the mass spectrometer was also connected to an on-line dedicated computer system (Spectrosystem 11 1 MS, Varian-MAT), which determined exact masses and calculated elemental compositions for the main masses detected. The gas chromatograph was equipped with a glass column 1.83 m x 0.64 cm packed with 10% Carbowax 6000 on Chromosorb WAW, 80-100 mesh (WGA, Dusseldorf). The temperature of the injector was 180"C, the temperature of the molecular separator was 200"C, and the column temperature was 140°C isothermal. Helium gas was used as a carrier with a flow of 45 ml/min. RESULTS In kinetic studies on VPA in humans, dogs, mice, and rats ( L o s c h e r , 1978, Loscher and Esenwein, 1978), as well as in

METABOLITES OF VALPROIC ACID IN SERUM clinical routine analyses of serum concentrations of VPA in epileptic patients, we regularly found several smaller peaks in the gas chromatograms of serum extracts in addition to those of VPA and the internal standard. Comparison with extracts of control serum to which VPA was added, that did not show these peaks, indicated that these compounds resulted from VPA metabolism. Because to our knowledge no study exists on metabolites of VPA in serum we tried to identify these compounds in our GC-MS combination. Figure 1 plots the

593

total ion current (TIC) in the different species we examined. The TIC runs of the VPA-treated cases were compared to those of the untreated group, and mass spectra were made of all outstanding compounds. In the mouse group up to four different compounds were found (numbered 1-4). For humans and mice, three compounds with identical retention times and identical mass spectra were seen, whereas in dogs and rats only two were found. Figures 2 and 3 show the mass spectra from authentic VPA and the compounds numbered 1-4 in Fig. 1. Table 1 gives the

Zmin

Rat

2rnin

L

2 rnin

FIG. 1. GC-MS analyses of free VPA and its metabolites on Carbowax 6000. Plots of the total ion current (TIC) for the specimens examined.

Epilepsia, Vol. 79,December 7978

594

CORNELIS JAKOBS AND WOLFGANG LOSCHER

I

57

I

mlc Mctabdite I

1001

I00

%

L

0

u

-

c

,

50. * .

-*

U

,

41

55

73 113

0-

I. . :. d IL,:

high-resolution data and computer-generated proposals for the elemental composition of the most important mass spectra fragments. The striking similarities between the mass spectra fragments of compounds 1-3 are immediately apparent. The ions at mle 41, 55, 95, 97, 100, and 113 keep reappearing, although with rather different intensities. The mass spectra of authentic VPA Epilepsia, Vol. 79, December 7978

Ltl.

4

..I,

1.11

.I

,

.

and of compounds 1-3 indicated that these compounds were indeed structurally related. Furthermore, the decrease in mass of several peaks in the spectra of compounds 1-3 by 2 atomic mass units (amu) suggested that it differed from VPA by two hydrogen atoms. This gave rise t o the hypothesis that we had found three different dehydroproducts of VPA. This assumption is supported by the fact

595

METABOLITES OF VALPROIC ACID IN SERUM Mrlabolilr 3

100.

-' --* 21

95

5

C

? a

U

a

50.

M' 142

55

' L1

.

I,

FIG. 3. Mass spectra of metabolites 3-4 in Fig. 1.

that metabolite 3 shows a molecular ion (M+) of 142 corresponding to the proposed elemental composition of CXHI4O2, which is identical to C,H,,O, (VPA)-2H. Similarly, the mle difference of 4 amu between mass 111 of compound 4 and mass 115 of VPA, and t h e difference of 2 amu between masses 1 1 1 and 125 of compound 4 and

masses 113 and 127 of metabolite 3 suggested a metabolite that differed from VPA by four hydrogen atoms. Supporting this conclusion is the fact that the computerproposed elemental composition of M+ = 140, C,H,,O, is identical t o C,H,,O, (VPA)-4H. Presumably the observed metabolites

TABLE 1. Comprrtrr-proporud elemc~ritcric o m p o ~ i t i o no f the niriin f'agtnrnts ob.reri,t~liri the high-resolution muss spectra o j VPA und metubolites 1-4

Compound VPA

MIE

Observed

Error

I02

102.0662 11S.0783

-1.8 +2.4

M+-C,,H,, M+-C,H,

+0.3

I27

I00.0527 113.0609 127.0763

M+-C,H,, M+-C,H, M+-CH,,

I00 I I3

1OO.OS 14 1 13.0609

-1.0

+0.6

M+-C,,H, M+-C,H,

97

97.1027 100.0551 113.0603 1 42.0999

t0.8 i2.7 0.0 +0.6

M+-CHO, M+-C,H,, M+-CzH, M+

1 I I.0442 125.057I 140.0838

-0.4 -3.1

M+-C,H, Mi-CH,, M+

I15

Metabolite I

I00 1 I3

Metabolite 2 Metabolite 3

100

I13 142

Metabolite 4

Ill

I25 140

Composition

Assignment

-1.1

+0.4

+o. 1 ~~

fpilepsia, Vol. 19, December 1978

596

CORNELIS JAKOBS A N D WOLFGANG LOSCHER

C&, result from dehydration of the correspondI ing hydroxylated VPA metabolites (such as CH3- C HZ-CHZ- CH-COOH + those already found in urine). This reaction I OH is likely to be caused by the strong acidic 4-OH-VPA condition at our first step of the extraction procedure and/or on the polar GC column C3H7 at high temperature. Such dehydration CH3- C HZ-CHZ- C H- C O I I gives rise to unsaturated compounds. -Od On the other hand, especially in the case 4-OH-VPA-y-lactone of 4-OH and 5-OH-VPA, the possibility is present for formation of the corresponding Metabolite 1 y- and 6-lactones, as this reaction proceeds 5-Hydroxy-Valproic Acid (5-OH-VPA) with great readiness and predominates over To 1 ml of a 10% aqueous solution of other processes in a mineral acid medium 5-OH-VPA-sodium 9 ml water was added. (Fieser and Fieser, 1964). Such a ring Fifty pl(500 pg) from this standard solution closure-also coupled with the loss of an again were prepared in the previously H 2 0 group-although not giving unsatudescribed way. The gas chromatogram of rated compounds, nevertheless produces this sample showed a peak whose mass compounds with the VPA-2H composition. spectrum was identical to that of metaboWe therefore verified our results with pure lite 2. reference compounds (5-OH-VPA and In analogy to the former lactone forma4-OH-VPA-y-lactone.) tion, the following origin, identity, and 4-Hydroxy-Valproic Acid (4-OH-VPA) structure of metabolite 2 seem assured: This acid was provided in form of the C3H7 4-hydroxy-valproic acid-y-lactone (4-OHCH, -CH2-CHZ- CH-COOH + VPA-y-lactone). A standard solution was I prepared by adding 100 pl of the pure OH 5-OH-VPA 4-OH-VPA-y-lactone in 10 ml water. From this standard solution 50 pl (500 pg) was C,H, I prepared in the same way as described for VHz-CHZ-CH2- C H - C O our serum samples. An aliquot of this sample was injected under exactly the same S-OH-VPA-8-lactone GC conditions as for the serum analyses. The GC of this sample showed a peak the Metabolite 2 mass spectrum of which was identical to that of metabolite 1. In order to show that Figure 4 shows the origin of the characterunder the above-mentioned conditions this istic ion in the mass spectra of metabolites metabolite resulted from ring closure of the 1 and 2 . 4-OH-VPA, we made the sodium salt of this acid by mixing 100 pl of the 4-OH- Metabolite 3 The existence of metabolite 3 cannot be VPA-y-lactone standard solution with an equimolar amount of 0.1 i s NaOH. This explained by lactone formation. A hydroxyl sample again was prepared as described group in the p(C-3) position (the only previously. Also in this case a peak with a position left) undergoes dehydration to a,p mass spectrum identical t o that of the unsaturated acid. Combination of an actilactone and metabolite 1 appeared. F o r vated hydrogen atom in the a-position and metabolite 1 the following origin, identity, a hydroxyl group in the adjacent p-position affords a favorable opportunity for the and structure seems to be proved: Epilepsia, Vol. 19, December 7978

METABOLITES OF VALPROIC ACID IN SERUM VPA foulhentic I

Compound

597

Metabolite 3

y 2 CH-

Fragment

o

b

-

11s 102

c

d

$

a

-

-

-

Metabolite 2

Compound

b c d M ' 113 100 97 lL2

Metabolite 1

CH

CH

...&.2 ___.. b CH .--I.? .__.. Cl Structure

Fragment

H

CH-CH~-CHTCH-CO

I 2

a

b

-

113 100

c

M

'

-

.1.*.

+'.

...b

CH

..

?

cl

....

CH3CH-CHrCH-CO

a

127

b

c

M

H

'

113 100 -

FIG. 4. Origin of characteristic ions in the mass spectra of VPA and its metabolites 1-3

elimination of water with establishment of a double bond. In the case of unsaturated compounds, the mass spectra are generally not very sensitive to molecular structure. Nevertheless there are some general rules predicting which fragments can be expected to break off and these, together with the expected McLafferty rearrangements, can be used to explain the structure of these compounds. Thus, it is observed that methyl groups break off only at points of chain branching, whereas the loss of an ethyl group is always observed, at least to some extent. It is also more likely that a chain will break in a p-position to a double bond @-fission) rather than in the a-position. In contrast, an isopropyl group is readily removable, yielding an abundant ion at rnle = 43 and parent M' - 43. The ion at mle = 41 is also abundant and corresponds to a further loss of two hydrogen-atoms. The same relationship exists between the ions mle = 57 (CJ and 55.

For metabolite 3 the following origin, identity, and structure seem to be likely: C3H7 CH,-CH,-

CH-CH-COOH

+

I

OH 3-OH-VPA

I

CH,-CHz-CH= C -COOH 2-propyl-2-pentenoic acid (A-2-VPA)

Metabolite 3

Assuming that nearly the same fragmentation will occur in metabolite 3 as in VPA, the main fragments of VPA and metabolite 3 are discussed as follows. (For origin of the characteristic fragments, see Fig. 4.) Concerning the mass spectrum of VPA: The base peak at rnle = 102 results from cleavage of the isopropyl group on the tertiary C-atom followed by a McLafferty rearrangement involving a H-atom giving CH,-CH2-CH2-CH=C

-OH

+AH Epilepsia, Vol. 19, December 1978

CORNELIS JAKOBS AND WOLFGANG LOSCHER

598

A further cleavage of C,H, from this fragment can explain mle = 73. The mass at m/e = 115 indicates a break-off of a C,H, group (b in Fig. 4). That fragment d and the parent ion M+ are missing is understandable because of the branching and resulting likely cleavage a t the tertiary C-atom. Concerning the mass spectrum oj'metabolite 3. The double bond in the C-2 position together with the C=O from the carboxylic acid group provides a conjugating double bond system and a stabilizing effect on the molecular ion at mle = 142. In this case, M+ - 45 is reasonable, whereas 95 indicates a further dehydrogenation of mass 97. The base peak at mle = 113 is explainable by the preferred cleavage of a C,H, group at the B-position to the double bond. The small mass at mle = 100 is likely in this case because of t h e instability of the McLafferty rearrangement ion: CH,-CH,-CH=C=

CJH,

I

CH,- C H - C H -CH - C H -COOH

I

OH

/

OH 3,4-di-OH-VPA C,H7

CH,- C H -C = C - C O

Lo.-

4- hydroxy-2-propyl-2-pentenoic-acid-ylactone (A-2-4-OH-VPA-y-lactone) Metabolite 4

Although this assumption is not unlikely, we did not succeed in explaining all the fragments at the moment. For this reason, metabolite 4 is missing in Fig. 4. Another possibility t o explain the loss of four hydrogen atoms might be the following pathway: C9H7

C -OH

I

+OH

The proposed elemental composition of the fragments from the high-resolution data supports this suggestion. Metabolite 4

The loss of four hydrogen atoms indicates a double dehydration of a dihydroxyVPA metabolite. It is remarkable that this compound has both phenomena of metabolite 1 and 3 . On one side, the fragment mle = 125 (M+ - CH,) indicates a splitting off of a methyl group, which is only observed at points of chain branching. The formation of a 4-OH-VPA-y-lactone brings a CH,-group in such a position (see also metabolite 1). On the other side, the rather high abundance of the molecular ion mle = 140 indicates again a double bond at C-2, providing a conjugating double bond system with the C = O from the carboxylic acid group (see metabolite 3). Both phenomena would lead to the hypothesis that Epilepsia, Vol. 19, December 1978

metabolite 4 is identical to 3,4-dihydroxyVPA:

CHa- C -CH,-CH-COOH

+

I1 0

(a) C3H7

CH3- C =CH- CH-COOH

--*

I OH (b) C,H,

I

CH,- C = CH - C H - C 0

L-J ( C)

Starting from the 4-keto-VPA (a) followed by enolistation (b) and y-lactone formation (c), the end product of this pathway (A3,4-OH-VPA-y-lactone) also has the composition of VPA-4H. The fragment mle = 125 (M+-CH,) would also be likely in this case, but the high abundance of the molecular ion mle = 140 is not really a sufficient explanation. 5-keto-VPA has a theoretical possibility of appearing. This substance would lead by the reaction sequence a-c to:

METABOLITES OF VALPROIC ACID IN SERUM

599

its origin and identity. @-Oxidation of VPA would lead to 3-OH-VPA, and this acid C H=CH-CH,- C H- C 0 showed a favorable dehydration reaction, I 0 2 giving A-2-VPA, which may explain the A4,S-OH-VPA-8-lactone main fragments of the mass spectrum of which cannot explain the fragments mle = metabolite 3. Two possible pathways are proposed for 125 nor the high abundance of M+ = 140. metabolite 4 to explain the origin of this As shown recently, 3-keto-VPA is compound. The first-and most likelythe most excreted metabolite in urine would give rise t o 3,4-dihydroxy-VPA, (Gompertz et al., 1977, Kochen et al., 1977). Although one may suggest that this which would mean a sequence of a p- and metabolite is found in serum, metabolite 4 a w-oxidation of VPA. The second would cannot be identical to this compound. As lead to the existence of 4-keto-VPA, which mentioned earlier, the lactone formation can be discussed by a further oxidation of (in this case after enolisation) from a p 4-OH-VPA (metabolite 1). Both compounds had been never menhydroxylated carboxylic acid don’t take tioned previously as metabolites of VPA. place and a further dehydration is not posAlthough metabolites 1 and 2-the prodsible because of the resulting adjacent ucts of w-oxidation-are missing in the double-bonds. In the case of 3-keto-VPA, serum of dog and rat, metabolite 4, which the decarboxylation, leading to heptanon-3 is highly favored. The occurrence of the is a proposed product of at least one wlatter artifact in rat urine was described by oxidation, occurred in serum of all species examined. Matsumoto et al. (1976). w-Oxidation seems t o be the preferred DISCUSSION metabolic pathway for mice, whereas in By GC-MS analysis of serum samples humans, dogs, and rats p-oxidation seems of VPA-treated humans, mice, dogs, and to be favored, as can be seen in Fig. I . rats, we found, in addition to unmetabo- The fact that in human serum metabolite 3 lized VPA, two to four other substances (3-OH-VPA) has the highest concentration (metabolites 1 -4), depending on the (semi-quantitative value 5- 10 pg/ml) species examined. With authentic refer- agrees with the finding that 3-keto-VPA ence substances, two of these compounds was the major metabolite in urine being could directly be identified. Metabolite -1, excreted (Gompertz et al. 1977; Kochen identical t o 4-OH-VPA-y-lactone; and et al., 1977). This implies, as we mentioned metabolite 2, identical to 5-OH-VPA-6- earlier, an expected preferred p-oxidation lactone, respectively, resulted from lactone in humans. The first intermediate expectformation of their corresponding hydroxyl ed in this conversion is 2,3-unsaturated acids, COH-VPA and SOH-VPA, favored valproyl-CoA, which would be formed by the strong mineral acid medium before from valproyl-CoA by an acyl-CoA dehyextraction takes place. drogenase. This compound would be hy4-OH-VPA and 5-OH-VPA are metabolic drated by the action of enoylhydratase to products, respectively, of the w2 and wlgive 3-OH-valproyl-CoA, which correoxidation of VPA and have also been found sponds to our metabolite 3. Conversion to in urine (Kuhara and Matsumoto, 1974; 3-keto-valproyl-CoA would be catalyzed Matsumoto et al., 1976; Kochen et al., by a dehydrogenase. This pathway of VPA 1977). Reference compounds for metabois identical to the p-oxidation of fatty acids lites 3 and 4 are currently not available, but and may explain why in the past anticertainly in the case of metabolite 3 the convulsive therapy consisted of either fastmass spectra gave a likely explanation for ing or of a ketogenic diet, either of which C,H,

I

Epilepsia, Vol. 19, December 1978

600

CORNELIS JAKOBS AND WOLFGANG LOSCHER

could also produce similar P-hydroxylated compounds in the second stage. However, in this context, it cannot be ruled out that the therapeutic effect was caused by the resultant ketoacidoses rather than by the intermediate metabolites. Matsumoto et al. (1976) noted that the structure of valproyl-CoA closely resembles that of a-methylbutyryl-CoA, a metabolic intermediate of the branched chain amino acid isoleucine. Since the oxidation steps for isoleucine were suggestive of those for VPA, isoleucine may compete with VPA in the oxidation steps. On the basis of the assumption, Matsumoto et al. administered VPA and isoleucine to rats simultaneously, and the results showed that isoleucine did not affect the w-oxidation of VPA, but did interfere competitively with the p-oxidation of the drug. In this context, we would like to mention that 3-OH-VPA (our metabolite 3), found in serum of all the species examined, closely resembles 2-methyl-3-hydroxy-butyric acid, the corresponding p-hydroxylated branched chain carboxylic acid intermediate in the isoleucine metabolic pathway.

study the metabolites identified to determine whether or not they exert any anticonvulsive activity. These studies will involve the use of stable-isotope-labeled VPA as a starting material in order to reach a more definite conclusion. SUMMARY

In kinetic studies of VPA in humans, dogs, rats, and mice, as well as in clinical routine analysis of serum concentrations of VPA in epileptic patients, 2 - 4 peaks (depending on the species examined) were regularly found in the gas chromatograms in addition to VPA. Comparison with control serum indicated that these peaks resulted from metabolism of VPA. By GC-MS, two of these metabolites could be identified as 5-hydroxy-2-propylpentanoic acid (5-OH-VPA) and 4-hydroxy2-propyl-pentanoic acid (4-OH-VPA), using synthesized reference substances. Both metabolites result from o (ol, w2) oxidation of VPA, 5-OH-VPA only occurring in serum of mice, and 4-OH-VPA in serum of mice and humans. With the aid of low- and high-resolution mass spectra, likely structures of the two remaining metabolites, CONCLUSION both of which were found in serum of all In conclusion, the identification of the the species examined, were proposed. One o-oxidation products of VPA, 4-OH-VPA of these, 3-hydroxy-2-propyl-pentanoic in serum of humans and mice and 5-OH- acid (3-OH-VPA) confirms the involvement VPA in serum of mice alone, seems as- of p-oxidation in the metabolism of VPA. sured. The suggested structure of our third The fourth metabolite, whose identity is metabolite, 3-OH-VPA, which occurs in all uncertain, indicates a substance not despecies and results from p-oxidation, is scribed previously as a metabolite of VPA. very probable. However, we are aware of the fact that only comparison of the spectra ACKNOWLEDGMENT found with those of authentic compounds The authors are much indebted to Dr. can provide absolute identification. The H . Schafer, Desitin-Werk Carl Klinke, same can be said for our metabolite 4. Hamburg, for providing the reference subHere, a new metabolite, not even observed stances and for many helpful discussions. before in urine, is indicated. To establish the structures of the latter REFERENCES two metabolites unambiguously, we are Eymard P , Simiand J , Teoule R , Polverelli M , Werbenec JP, and Broll M . Etude de la repartition undertaking preparation of these comet la resorption du dipropylacetate de sodium pounds to obtain the reference spectra. marque au 'Tchez le rat. J Phurrnucol (Paris) 2:359-368. 1971. Further experiments will quantitate and

Epilepsia, Vol. 19, December 1978

METABOLITES OF VALPROIC ACID IN SERUM Fieser L F and Fieser M. Orgunic Chemistry, Reinhold, New York, 1964. Gompertz D, Tippett P, Bartlett K, and Baille T. Identification of urinary metabolites of sodium dipropylacetate in man; potential sources of interference in organic acid screening procedures. Clin Chim A c t a 74: 153- 160, 1977. Gugler R, Schell A, Froscher W, and Schulz HU. Disposition of valproic acid. Eur J Clin Pharmacul 12:125-132, 1977. Kochen W , imbeck H , and Jakobs C. Untersuchungen uber die Ausscheidung von Metaboliten der Valproinsaure im Urin der Ratte und des Menschen. A r z n e i m . - F o r s c h . ( D r u g R e s . ) 27: 1090- 1099, 1977. Kuhara T, and Matsumoto I. Metabolism of branched medium chain length fatty acid. I. o-Oxidation of sodium dipropylacetate in rats. Biomed Mass Spectrom 1:291-294, 1974. Loscher W. Rapid determination of valproate sodium in serum by gas liquid chromatography. Epilepsia 18~225 -227, 1977. Loscher W. Protein binding and pharmacokinetics of sodium valproate in man, dog, rat and mouse. J Pharmucol Exp Ther 204:255-261, 1978. Loscher W and Esenwein H . Pharmacokinetics of sodium valproate in dog and mouse. A r z n e i m . Fursc,h. (Drug R e s . ) 28:782-787, 1978. Matsumoto I, Kuhara T, and Yoshino M. Metabolism of branched medium chain length fatty acid. 11. &Oxidation of sodium dipropylacetate in rats. Biomed Mass Spectrum 3:235-240, 1976. Pinder RM, Brodgen R N , Speight T M , and Avery GS. Sodium valproate: A review of its pharmacological properties and therapeutic effiacy in epilepsy. Drugs l 3 : 8 l - 123, 1977. Schobben F, van der Kleijn E, and Gabreels FJM. Pharmacokinetics of di-n-propyl-acetate in epileptic patients. Eur J Clin Phurmucol 8:97- 105, 1975.

RESUME Des etudes cinetiques de I’acide valproique (VPA) (ou acide dipropylacetique, ou encore acide 2-propylpentanoique) chez I’homme, le chien, le rat, et la souris ainsi que I’analyse routiniere des concentrations seriques du VPA en pratique clinique chez des patients epileptiques, ont regulierement demontre I’existence sur les chromatographies en phase gazeuse de 2 a 4 pics (selon l’espece examinee) en plus de celui du VPA. L’etude comparative avec du serum de contrBle montre que ces pics sont lies au metabolisme du VPA. Par la chromatographie en phase gazeuse et la spectrometrie de masse deux de ces metabolites ont pu etre identifies comme etant I’acide 5-hydroxy-2propyl-pantanoique ( 5 - O H - V P A ) et I’acide 4hydroxy-2-propyl-pentanoique (4-OH-VPA) grace a I’utilisation de substances synthetiques de reference. Ces deux metabolites resultent de I’oxydation CU (CU 1 , CU 2 ) du VPA, le S-OH-VPA n’apparaissant que dans le serum des souris et le 4-OH-VPA dans le serum des souris et de I’homme. A I’aide des spectres de masse de faible et de haute resolution,

601

les structures probables des deux autres metabolites, qui sont trouvks dans le serum de toutes les especes examinees, ont ete proposees. L’un d’eux, I’acide 3-hydroxy-2-propyl-pentanoique (3-OH - VPA) confirme I’implication d‘une P-oxydation dans le metabolisme du VPA. Le quatrieme metabolite, dont I’identite est encore incertaine, correspond a une substance non encore decrite en tant que metabolite du VPA. (J .-L. Gastaut, Murh eilks)

RESUMEN Realizando estudios cineticos de acido valproico (VPA) (Acid0 dipropilacetico, Bcido 2-propilpentanoico) en humanos, perros, ratas y ratones, ademas de las determinaciones rutinarias de concentraciones sericas de VPA en enfermos epilepticos, se han hallado de modo persistente en la cromatografia d e gases, 2-4 picos (dependiendo de las especies examinadas) ademas del VPA. Las comparaciones con 10s sueros control indicaron que estos picos resultan del metabolismo del VPA. Mediante estudios con cromatografia de gases-espectrometria de masa, se pudieron identificar dos de 10s metabolitos citados como acido 5-hidroxi-2-propil-pentanoico(5-OHVPA) y acido 4-hidroxi-2-propil-pentanoico (4-OHVPA), utilizando sustancias sinteticas de referencia. Ambos metabolitos resultan de la oxidaci6n w ( w , , u p ) del VPA y el 5-OH-VPA solo se encuentra en el suer0 del raton mientras que el 4-OH-VPA solo se detecta en el suero del raton y del hombre. Con la ayuda del espectro de masa de baja y alta resolucion se proponen dos estructuras semejantes a 10s metabolitos restantes que han sido encontradas en el suero de todas las especies estudiadas. Una de estas, el acido 3-hidroxi-2-propil-pentanoico (3-OH-VPA) confirma la participacion de la p-oxidacion en el metabolismo del VPA. El cuarto metabolito cuya identidad es todavia incierta apunta a una sustancia que no ha sido previamente descrita como metabolito del VPA. ( A . Portera Sanchez, Madrid)

ZUSAMMENFASSUNG Sowohl bei Untersuchungen der Pharmakokinetik von Valproinsaure (VPA) (Dipropylessigsaure, 2F’ropylvaleriansaure) bei Mensch, Hund, Ratte und Maus als auch bei klinischen Routineanalysen des Serumspiegels von VPA bei Patienten mit Krampfleiden wurden im Gaschromatogramm zusatzlich zu VPA regelmaisig Peaks von 2-4 weiteren Substanzen (abhangig von der untersuchten Species) beobachtet. Diese Peaks traten nur nach Applikation von VPA auf, so dall angenommen wurde, dal3 e s sich um Metaboliten dieses Antiepileptikums handelte. Mittels Gaschromatographie-Massenspektrometrie wurden zwei dieser Verbindungen als S-Hydroxy-2propylvaleriansaure (5-OH-VPA) und 4-Hydroxy-2propylvaleriansaure (4-OH-VPA) unter Verwendung synthetisierter Referenzsubstanzen identifiziert. Die beiden Metaboliten, die durch eine o (o,, 02b Oxidation aus VPA e n t s t e h e n . wurden n u r in

Epilepsia, Vol. 19, December 1978

602

CORNELIS JAKOBS AND WOLFGANG LOSCHER

Mauseserum (5-OH-VPA)bzw. Serum von Maus und Mensch (4-OH-VPA) nachgewiesen. Fur die zwei verbleibenden Metaboliten, die im Serum aller Species auftraten, wurden mit Hilfe von Niederund Hochauflosungs-Massenspektrometrie wahrscheinliche Strukturen abgeleitet. Der so identifizierte Metabolit 3-Hydroxy-2-propylvaleriansaure (3-OH-

€pilepsia, Vol. 19, December 1978

VPA) zeigt, wie von Urinuntersuchungen bekannt. daJ3 VPA auch uber eine ,&Oxidation abgebaut wird. Der 4. Metabolit, dessen Struktur erst unsicher geklart ist, weist auf eine Verbindung hin, die bisher noch nicht als Metabolit von VPA beschrieben wurde. (Author’s Summary)