Plasma Quantitation of Warfarin and Warfarin Alcohol by Gas Chromatography Chemical Ionization Mass Spectrometry in Patients on Warfarin Maintenance Therapy A. M. Duffield Biomedical Mass Spectrometry Unit, The University of New South Wales, P.O. Box 1, Kensington, 2033, Australia

P. H. Duffield, D. J. Birkett, M. Kennedy and D. N. Wade Department of Clinical Pharmacology, S t Vincent's Hospital, Darlinghurst, 2010, NSW, Australia

A quantitative method has been developed to measure plasma concentrations of warfarin and warfarin alcohol. The analytical procedure uses deuterated analogues as internal standards, and the technique of selected ion monitoring following gas chromatography methane chemical ionization mass spectrometry of the 4'-methyl ethers of warfarin and warfarin alcohol. Concentrations of warfarin and warfarin alcohol have been measured in plasma samples from 4 3 patients maintained on chronic warfarin therapy and compared with the 'apparent warfarin' concentration as measured by a fluorometric procedure. The study demonstrated a high degree of correlation between the gas chromatographic mass spectrometric derived sum of the individual concentrations of warfarin and warfarin alcohol, and the 'apparent warfarin' concentration determined from a spectrofluorometric assay.

CR,

INTRODUCTION

I

CO I

Recently we described' a quantitative assay using gas chromatography methane chemical ionization mass spectrometry (GC CH4 CIMS) to measure plasma concentrations of warfarin in healthy male volunteers. When using this technique to monitor the plasma concentrations of warfarin in patients with haematological disorders the measured plasma warfarin concentrations were consistently found to be lower than values obtained with the Corn and Berberich2 spectrofluorometric assay. Similarly, other investigator^,^ using high performance liquid chromatography (HPLC), demonstrated that the spectrofluorometric procedure2 always generated higher plasma warfarin concentrations. However, these authors3 did not measure the warfarin metabolite(s) responsible for the difference between these assays. We now report a sensitive, specific assay for warfarin (1)and its major plasma metabolite warfarin alcohol (2)in patients on maintenance warfarin therapy. A high degree of correlation ( r = 0 . 9 5 ; p < 0.001) was found between the total of the individual plasma concentrations of warfarin and warfarin alcohol as determined by our procedure and the spectrofluorometrically determined concentration of warfarin'.

1:R = R ' = H 3:R='H;R'=H 5: R=H;R'=CH3 7:R = 2H; R ' = CH3

((R

3

R( OH I

2:R=R=H 4: R = 'H; R' = H

6:R=H:R'=CH3 8: R = 'H: R' = CH3

was supplied by Applied Science, State College, Pennsylvania USA. Diazomethane was prepared from Diazald (Aldrich Chemical Co., Milwaukee, Wisconsin, USA) by standard procedures. Lithium aluminium deuteride (99'/0 2H)was supplied by Stohler Isotope Chemicals (Azusa, California, USA). Warfarin and warfarin alcohol were gifts from Endo, Sydney, Australia.

Equipment CI (methane) mass spectra were recorded on a Finnigan 3200 GCMS system interfaced to the same manufacturer's 6110 data system. The GCMS system was fitted with a Finnigan diverter valve assembly.

GC conditions

EXPERIMENTAL Reagents All chemicals were of analytical reagent quality. Diethyl ether was distilled prior to its use. G C column packing

G C separations were achieved using a glass column (5 ft, 2 m m ID) packed with 3% OV-1 on Gas Chrom Q (100/120 mesh). The gas chromatograph injector port was maintained at 260 "C, the column oven at 210 "C (programmed 1 min after sample injection to 240 "C at

CCC-0306-042X/79/0006-0208$02.00 208

BIOMEDICAL MASS SPECTROMETRY, VOL. 6, NO. 5, 1979

@ Heyden & Son Ltd, 1979

QUANTITATION OF WARFARIN AND WARFARIN ALCOHOL

6 "C/min-'), the GCMS interface oven at 250 "C and the transfer line to the mass spectrometer at 240 "C. In order to minimize ion source temperature fluctuations the mass spectrometer filament (0.8 ma emission) was kept on at all times. Solvent from the G C injection was removed by the Finnigan diverter valve assembly. Methane carrier gas (flow 20mlmin-l) served as the reactant gas (0.8-0.9 Torr) for CIMS.

Preparation of deuterated internal standards ['H5]Warfarin (1-(4'-hydroxy-3'-coumariny1)-1-3-one (3) was prepared as described previously. 1*4 A standard solution containing 0.4 mg 1-' of 3 in diethyl ether was preyared and stored in the dark at 4 "C. [ H6]Warfarin alcoh21 (1-(4'-hydroxy-3'-coumariny1)- 1-phenyl-[ 2,2,2,4,4; HJbutan-3 -01) (4) was prepared by reduction of [ H4lwarfarin (100 rng) in anhydrous ether (20 ml) with excess lithium aluminium deuteride. The reaction mixture was refluxed for 30 min, excess reagent destroyed (H'O), the ether filtered, dried (NazS04),and the solvent removed in oucuo to yield a mixture of diastereoisomers of ['H6]warfarin alcoh$(4) (80 rng) of isotopic composition 86% ['H6], 13% [ H4] (solid sample probe CH4 CIMS). A standard solution containing 0.6 mg lPi of 4 in diethyl ether was kept in the dark at 4 "C.

Methylation with diazomethane Samples were methylated at room temperature with an excess of an ether solution of diazomethane. After lOmin, reagent and solvent were removed under a stream of dry nitrogen at 25-30 "C.

GC CIMS of warfarin-4'-methyl ether (5) and warfarin alcohol-4'-methyl ether (6) Methylation of a mixture of warfarin (1)and warfarin alcohol (2) with diazomethane and methane G C CIMS of the products yielded two peaks with retention times of 5.0 and 5.7 min respectively (conditions described above). The resulting methane CI mass spectra of warfarin-4'-methyl ether (5), and warfarin aIcohol-4'methyl ether (6) are reproduced in Fig. 1. When the pentadeuterated internal standard of warfarin-4'-methyl ether (7)was subjected to these identical GCMS conditions a considerable quantity of deute;ium was exchanged by protium. As described previously the deuterium content of 7 on GCMS analysis remained constant provided the same G C conditions were used during the assay (standard curves and plasma analyses).

50

&GH5

325 IMHI'

I

265

I

I00

I

I

,

260

I50

250

I il

353 365 ,

I,

I1

3 h

300

400

m/z

Figure 1. Methane CIMS of (a) warfarin-4-methyl ether (5) and (b) warfarin alcohol 4'-methyl ether (6).

was detected using m / z 293 [MH-CH30HIf for warfarin alcohol 4'-methyl ether, (6) and m / z 299 for its [*H6] labelled internal standard (8). Peak height ratios were calculated by dividing the respective peak height responses of warfarin and warfarin alcohol 4'-methyl ethers by the peak height of the signal generated by their respective internal standards (7 and 8).

Standard curve Warfarin and warfarin alcohol standard solutions were each made by dissolving 5 mg of each in water (25 ml) containing NaOH (0.1 g) and stored in the dark at 4 "C.

1

'OOr

m/z 327.1

0

'OOr

'OOr

L -

m/z 293. I

Selected ion monitoring For the quantitative detection of warfarin (1)the GCMS data system was operated in the selected ion monitoring (SIM) mode using m / z 323 ([MH]+ for warfarin methyl ether) and m / z 327 [(MH]: for its ['H4] labelled internal standard) (see Fig. 2). Similarly, warfarin alcohol @ Heyden & Son Ltd, 1979

O

8

I

3

4

I

I

5 6 Time (min)

I

7

Figure 2. Plasma extract of selected ion detection trace of warfarin4'methyl ether (m/z 323); [*H41warfarin-4-methyl ether ( m l z 327); warfarin alcohol-4-methyl ether ( m / z 293) and 12HGlwarfarinalcohol-4'-methyl ether ( m l z 299) from patient D.N.

BIOMEDICAL MASS SPECTROMETRY, VOL. 6, NO. 5, 1979 209

A. M. DUFFIELD, P. H. DUFFIELD, D. J. BIRKE'IT, M. KENNEDY AND D. N. WADE

1

2 0,

to each of 16 x 1 ml samples of blank plasma. Each was processed as described to yield experimental plasma concentrations of 1.86 pg ml-' f 0.07 pg ml-' (mean f SD) and 1.85 pg ml-' ~t0.08pgml-' for warfarin and warfarin alcohol respectively.

Fluorometric determinations

p g warfarin p g warfarinalcohol

Figure 3. Standard curves for warfarin ( 0 , O )before (a) and after (b)16 plasma analyses and warfarin alcohol (m, A)before (a) and after (b) 16 plasma analyses.

Aliquots of this solution containing warfarin ( 1 , 2 and 3 pg) and warfarin alcohol (1,2 and 3 pg) were added to normal drug-free plasma which was then extracted and derivatized as described under plasma extraction (see below). A linear response was generated by both 5 and 6 in this concentration range as shown in Fig. 3. The standard curves [Fig. 3(a)] for both warfarin and warfarin alcohol 4'-methyl ethers increased slightly in slope [Fig. 3(6)] with the number of samples assayed. To accommodate this gradual increase in slope standard curves were repeated after each group of 16 samples injected (eight patients in duplicate). The slight increases in slope of the standard curves were apportioned between the successive samples analysed.

Collection of blood samples Blood samples (15 ml) were collected from the median cubital vein of patients on long-term warfarin therapy. The samples were transferred immediately into tubes containing heparin (286 U.S.P. Units), centrifuged and the plasma stored at - 20 "C until analysed. Individual doses of warfarin ranged from 1-20 mg per day and all subjects had received the same dose for at least the previous month.

The fluorescence of warfarin and warfarin alcohol in acetone' was examined individually. Both compounds displayed the same excitation and emission maxima (340-400 nm). However, the emission of warfarin alcohol was 2.20f0.12 times that of warfarin. In order to compare the GCMS derived plasma concentrations of warfarin and warfarin alcohol with the results obtained by the non-specific fluorescence technique2 the following procedure was adopted. The plasma concentration (as determined by GCMS) of warfarin plus 2.2 times the concentration of warfarin alcohol was compared with the 'apparent warfarin' concentration as measured by fluorescence.

RESULTS AND DISCUSSION Under the GCMS conditions employed the 4'-methyl ethers of 1 and 2 yielded separate, well-shaped peaks with retention times of 5.0 and 5.7min respectively. This is illustrated in Fig. 2 were the ion current variation with time for masses 323 (5), 327 (7)(internal standard), 293 (6) and 299 (8) (internal standard) are reproduced for a typical plasma analysis. After reference to the standard curve the plasma concentrations of warfarin and warfarin alcohol in Fig. 2 were calculated to be 2.1 and 2.1 pg ml-' respectively. Individual warfarin plasma concentrations ranged from 0.6-3.4 r g ml-' and for warfarin alcohol from 0.2-2.1 pgml- in the 43 patients examined on warfarin maintenance therapy.

Plasma extraction Plasma (250 pl) was made basic with 0.1 ml of 1 M NaOH. Ether (4 ml) was added and the mixture agitated with a Vortex Mixer for 30s, centrifuged, and the organic phase discarded. The aqueous phase was acidified (0.5 m12 M HC1) and again agitated for 30 s on a Vortex Mixer with ether (4 ml) containing warfarin['H5] (3) (0.4 ml 1-') and warfarin alcohol-['H6] (4) (0.6 mg 1-'). After centrifugation the ether extract was evaporated to dryness at 25-30 "Cunder a stream of dry nitrogen. The residues were then derivatized (diazomethane) as described above.

:E

1% 0

.

0

I

I

I

I

I

I

I

I

I

0

1

2

3

4

5

6

7

8

Calculated fluorescence

Reproducibility

To assess the reproducibility of the assay, warfarin (2.0 pg) and warfarin alcohol (2.0 pg) were both added 210 BIOMEDICAL MASS SPECTROMETRY, VOL. 6, NO. 5, 1979

Figure 4. Correlation between estimated 'apparent warfarin' concentration (warfarin +2.2 x warfarin alcohol) and 'apparent warfarin' concentration as determined by the spectrofluorometric assay for 43 patients on long-term warfarin therapy.

@ Heyden & Son Ltd, 1979

QUANTITATION OF WARFARIN AND WARFARIN ALCOHOL

A comparison between the individual warfarin plus warfarin alcohol plasma concentrations as determined by the GCMS technique with those determined for ‘apparent warfarin’ by the fluorometric procedure* for each of 43 patients is shown in Fig. 4. There is a high degree of correlation ( r = 0.95, p < 0.001) between the two estimates. The slope (1.15) of the line of best fit suggests that the fluorometric procedure measures mainly warfarin and warfarin alcohol and that the warfarin alcohol concentration is given a 2.2-fold weighting

using the fluorometric procedure. The clinical implications of measuring warfarin and warfarin alcohol levels in patients on long-term warfarin therapy is discussed e l ~ e w h e r e . ~

Acknowledgements Financial assistance from G. D. Searle & Co. towards the purchase of the GCMS computer system and from the Australian Tobacco Research Foundation is gratefully acknowledged.

REFERENCES 1. P. H. Duffield, 0. J. Birkett, D. N. Wade and A. M. Duffield,

Biomed. Mass Spectrom. in press. 2. M. Corn and R. Berberich, Clin. Med. 13, 126 (1967). 3. E. S. Vessel1 and C. A. Shively, Science 184, 466 (1974). 4. W. F. Trager, R. J. Lewis and W. A. Garland, J. Med. Chem. 13, 1196 (1970).

@ Heyden & Son Ltd, 1979

5. P. H. Duffield,A. M. Duffield. M. Kennedy,D. J. Birkettand D. N. Wade, Aust. N.Z. J. Med. submitted for publication.

Received 29 November 1978

@ Heyden & Son Ltd, 1979

BIOMEDICAL MASS SPECTROMETRY, VOL. 6, NO. 5, 1979 211