tandem mass spectrometry for the

---

Anal. Chem. 1988,58,2453-2460 (10) Schlesinger. Z.; Slevers, A. J., Surf. Sci. Left. 1981, 702,L29-L34. (11) Otto, A. Z.fbys. 1988, 216, 398-410. (12) Ford, G. W.; Weber, W. H. fbys. Rep. 1984, 113, 195-287. (13) Heavens, 0. S. Optical froperties of Tbh Solid Films; Dover: New York, 1965: Chapter 4. (14) Johnson, P. B.; Christy. R . W. f h y s . Rev. B 1972, 6 , 4370-4379. (15) Graf, R. T.; Koenig, J. L.; Ishida, H. Appl. Spectrosc. 1985, 3 9 , 405-408. (16) Born, M.; Wolf, E. frincip/e of Optics, 3rd ad.; Pergamon Press: Oxford. 1965: DD 96. (17) Malltson, 1: H. Appl. Opt. 1963, 2 , 1105.

2453

(18) Kretschmann, E. 2.f b y s . 1971, 241, 313-324. (19) Hjortsberg, A.; Chen. W. P.; Bursteln, E. Appl. Opt. 1978, 17, 430-434. (20) Greenler, R. G. J . Cbem. fbys. 1969, 50, 1963-1968. (21) Mirabella, F. M. Jr. Appl. Spectrosc. Rev. 1985, P I , 45-178.

RECEIVED for review January 9,1986. Accepted June 2,1986. The authors gratefully acknowledge the financial support of the IBM Corporation.

High-speed Liquid Chromatography/Tandem Mass Spectrometry for the Determination of Drugs in Biological Samples Thomas R. Covey, E d g a r D. Lee, a n d J a c k D. Henion* Equine Drug Testing and Toxicology, New York State College of Veterinary Medicine, Cornel1 University, 925 Warren Drive, Ithaca, New York 14850

High-speed Mskpertonnance Uquld chromatography (HPLC) comMned wlth tandem m a s spectrometry (MS/MS) has been utHized to analyze crude equine wine and plasma extracts at a rate of 60 samples per hour. The atmospheric pressure chemical lonlzauOn source produces abundant (M 1)+ ions for drugs and their metabdltes which are then fragmented by ColllslOnaWy activated dissociation. Short, 3 pm partlcle HPLC columns were used to separate the analytes of hierest from high levels of endogeneous matrix materials which facilitated a continuous l-h high-speed LC/MS/MS analysis of 60 equine urine and plasma samples contahlng phenylbutazone and As metabolites. Selected reaction monitoring was used for the high-speed determlnatbn of phenylbutazone and its metabolites. LC/MS/MS determinations of the metabolttes of propiopromazine and promazine in equine urine are shown.

+

Tandem mass spectrometry (MS/MS) has developed into an established analytical technique. The increase in sensitivity, specificity, and speed attainable with MS/MS has led to an explosion of applications for the analysis of complex organic samples ( I ) . Early reports of mixture analysis capability for MS/MS featured the analysis of crude samples by direct insertion probe sample introduction. Samples of sewage sludge (2),urine (3),and biological matrices (4)were analyzed for targeted compounds a t relatively high levels. Brotherton et al. (5)reported procedures for the determination of drugs in the urine and plasma of racing animals using the direct insertion probe with varying degrees of sample cleanup. We also investigated the direct insertion probe (DIP) approach, but found satisfactory results could be obtained only with some preliminary sample cleanup by liquid-liquid extraction followed by preparative thin-layer chromatography (6).This work indicated that direct analysis by direct insertion probe using atmospheric pressure chemical ionization (APCI) MS/MS was not practical for crude extracts in our work primarily due to suppressed ionization of minor constituents in the volatilized sample ( I ) and the presence of several iso-

baric components of similar structures that would frequently be present in a sample. Here we employ rapid HPLC separation on-line with MS/MS in place of the preparative thinlayer chromatography. The determination of drugs and their metabolites in biological samples poses several analytical problems. The most important analytical needs for laboratories engaged in drug testing and toxicology include high sample throughput with concomitant high sensitivity and selectivity. Although mass spectrometry currently provides the best available sensitivity and selectivity, the gas chromatographic (GC) and high-performance liquid chromatographic (HPLC) inlet systems for MS do not typically provide high sample throughput capabilities. Run times of several minutes to hours are typical for base line separation of the components of complex mixtures. The high-speed capabilities of MS/MS, however, provide separations on the order of seconds if full-scan daughter ion spectra are recorded. The slow step in an MS/MS analysis is not the separation step; it is sample preparation and the speed that samples can be introduced into the ion source that is limiting. Sample introduction by direct insertion probe is relatively slow and cumbersome and provides no separation of analyte from the matrix or interfering components that have the same molecular weight and similar daughter spectra. The use of a moving-belt LC/MS interface as a direct insertion probe with repetitive spotting could speed up sample introduction, but matrix and interference effects would still exist. The use of short, efficient chromatography columns offers a rapid means for providing some separation of the components of a mixture, thereby minimizing the matrix effects and potentially separating interfering compounds. Short GC columns have been used as a means of rapidly introducing crude extracts into an MS/MS system with high sample throughput (7). Fast HPLC is an alternative approach for achieving rapid sample introduction and still maintaining the chromatographic separation which is frequently required when isobaric components of similar structure are present in a mixture (8). HPLC offers several advantages over GC as a means of sample introduction into the mass spectrometer. HPLC is

0003-2700/86/0358-2453$01.50/00 1986 American Chemical Society

2454

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12, OCTOBER 1986

amenable to the separation of complex mixtures of polar compounds without the use of elevated temperatures or derivatization. Modern 3-pm packing material in short (3-5 cm) columns can provide rapid sample analysis (9). In addition, mild ionization can be achieved with a variety of LC/MS interfaces. This extends the analytical capability of mass spectrometry to those compounds too fragile to produce ions containing molecular weight information when analyzed by the higher-energy ionization techniques. Most methods for on-line LC/MS produce abundant pseudomolecular ions resulting from proton addition or abstraction mechanisms. Structural information resulting from fragmentation of the molecular ion is generally lacking due to the mild ionization and low internal energy of the ion. LC/MS techniques such as direct liquid introduction (DLI) (IO),thermospray ( I I ) , electrospray (12),ion evaporation (13), and nebulization into an atmospheric pressure chemical ionization source (APCI) (14-16) all characteristically display this property. This can be a disadvantage for those LC/MS analyses where structural information in addition to molecular weight is required. However, this combination of mild ionization and mild sample introduction is ideally suited for MS/MS analysis and represents a major advantage of HPLC over GC and direct insertion probe sample introduction. As stressed by Levsen and Beckey (In,MS/MS analyses of mixtures are facilitated by soft ionization techniques, which give primary spectra with low fragmentation abundances. Structurally significant ions are produced during the collisionally activated dissociation (CAD) of the abundant parent ion produced during the initial ionization process. For complex mixture analysis, LC/MS and GC/MS are a t a disadvantage over LC/MS/MS and GC/MS/MS. Although great strides have been made toward improving chromatographic resolution, it is not an uncommon situation to have coeluting components during the analysis of complex samples. In fact, even the high chromatographic resolution attainable by capillary gas chromatography cannot achieve adequate separation power to unequivocally identify components based strictly on retention time. The principle of MS/MS allows one to separate the components of a mixture on an entirely different basis from that of the chromatographic process, Le., mass rather than chemical affinity for stationary and mobile phases. The combined separation power of chromatography and MS/MS allows isolation and identification of the components of a mixture. These factors suggest that coupling HPLC to MS/MS could be an ideal combination, providing an interface is implemented that is as reliable and easy to use as those employed for GC/MS. The heated nebulizer LC/MS interface is such a device. A wide range of reversed-phase solvent compositions a t flow rates between 0.5 and 1.5 mL/min can be used with this interface at a single temperature and gas flow setting. Nebulization into an atmospheric pressure ion source eliminates the vacuum system problems generally associated with LC/MS interfacing. This paper demonstrates that the features of high-speed HPLC and MS/MS using an APCI ion source and heated nebulizer interface can provide practical rapid sample throughput with high sensitivity and specificity for drugs and their metabolites in biological samples. EXPERIMENTAL SECTION Reagents. The drug standards utilized in this work were obtained from commercially available sources or independent synthesis. The promazine and propiopromazine metabolites were synthesized as reported elsewhere (18). All authentic standards were checked for purity by thin-layer chromatography (TLC) and HPLC before use. Organic solvents used for HPLC were obtained from Fisher Scientific (Rochester, NY) and vacuum filtered through 0.2-pm poly(viny1idene fluoride) filters (type GVWP, Millipore Corp., Bedford, MA). HPLC grade water was obtained

from an in-house water purification system (Barnstead Nanopure 11, Boston, MA) and filtered. The equine plasma and urine samples were obtained from experimental horses subsequent to therapeutic administration of the respective drugs or from racehorses that had received unknown doses of administered drug. The crude extracts of plasma and urine were obtained by liquid-liquid extraction of the biological fluid as reported previously (18). For the isolation of promazine, propiopromazine, and metabolites, 10 mL of urine and 2 mL of 1 M Na2C03were extracted with 4 mL of 2-propanol/methylene chloride 1/3 (v/v). The organic layer was evaporated to dryness at 60 "C under dry N2 and dissolved in 200 pL of HPLC mobile phase (57/43 methanol/water). The injection volume was 20 pL, equivalent to 1 mL of original sample. For the isolation of phenylbutazone and metabolites, 10 mL of urine and/or plasma was acidified to pH 3.5 with HCl and extracted with 5 mL of hexane/methylene chloride (2/1). The organic layer was then evaporated to dryness at 60 "C under dry N2 and dissolved in 200 pL of HPLC mobile phase (57/43 methanol/water). The injection volume was 20 pL representing 1 mL of original sample. Chromatography. The high-performance liquid chromatograph consisted of two Model 590 pumps driven by a Model 680 solvent programmer, a Model 440 UV detector with a fixed wavelength of 254 nm (Waters Associates, Milford, MA), and a Model ISS 100 robotic injector (Perkin-Elmer Corp., Norwalk, CT). The chromatographic column used was a 4.6 mm i.d. X 33 mm Perkin-Elmer Pecosphere 3 x 3 packed with 3-pm '2-18 particles. The exit of the UV detector was connected to a heated nebulizer LC/MS interface (Sciex, Thornhill, Ontario) (14-16). All connections between the components were made with 100 pm i.d. stainless steel tubing. The compositionsof the HPLC eluents were 57/43 methanol/water for promazine,63/37 methanol/water for propionylpromazine, and 53/47 acetonitrilelwater for phenylbutazone and its metabolites. The flow rate was 1.5 mL/min in all experiments. Mass Spectrometry. The heated nebulizer interface was introduced through the standard solids probe inlet of a TAGA 6000E triple quadrupole mass spectrometer equipped with an atmospheric pressure chemical ionization (APCI) source (Sciex, Thornhill, Ontario). HPLC flow rates of 1-1.5 mL/min could be continually introduced into the APCI source under either LC/MS or LC/MS/MS conditions. A heater temperature of 500 "C was used resulting in a measured vapor temperature of 100 "C. Zero air introduced into the interface to effect nebulization was optimized at the following settings: make up flow, 7 L/min; nebulizer pressure, 5 bar. Under LC/MS conditions quadrupole 1was operated in the full scan mode beginning at mlz 200 with quadrupole 3 operating in the rf only mode. When LC/MS/MS experiments were undertaken, the mass spectrometer was operated in either the daughter ion, selected reaction monitoring (SRM), or parent ion scan mode (1). In the promazine CAD work Q1 was operated with a resolution of 300 while a resolution of 600 was maintained in Q1for the phenylbutazone work. Unit resolution (full width at half maximum = 0.6 daltons) was maintained across the mass range scanned by Qa in all CAD work reported here. Under scanning conditions a scan rate of 2 cycles/s was used. For selected reaction monitoring (SRM) a dwell time of 100 ms was used. The TAGA 6000E MS/MS contains an open central quadrupole region which that is cryogenically pumped (19). The CAD region in this quadrupole field is formed by a free-jet expansion of the collision gas, ensuring that the fragmentation region is well-defined and that no collisions occur in quadrupoles 1 and 3. For these experiments argon was used as the collision gas with an effective target thickness of approximately 200 X 10l2atoms cm-2. The collisionally activated dissociation (CAD) experiments were optimized at a laboratory collision energy of 70 eV in order to obtain useful fragmentation information and optimize daughter ion sensitivity. Each andyte was studied using both positive and negative ion APCI to determine the best combination of sensitivity and selectivity. In the present paper the positive ion APCI mode was used. RESULTS A N D DISCUSSION The detection and determination of administered drugs and their metabolites in biological fluids are not easy tasks. The

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12, OCTOBER 1986

(n+lI 359

I

2455

CCnTRa

A

al

c P M/ 2

Figure 1. APCI mass spectrum of 2-( 1-hydroxyprop-1-y1)promazine sulfoxide (HPPS) with sample introduced into the ion source through the heated nebulizer LC/MS interface.

increased sophistication of illegally used drugs in sporting events has supported the need to have improved detection and confirmation schemes that unequivocally identify the presence of foreign substances in the body fluids of the participants. In addition, the trend toward screening human individuals prior to employment or for other purposes demands confirmatory procedures beyond the uncertainty of many simplified screening techniques. Propiopromazine. Propiopromazine is a phenothiazine tranquilizer that itself is rarely detectable since metabolism is extensive (20). Its metabolites are polar, thermally unstable, and difficult to handle in urine samples collected up to 2 days postadministration of the drug. The major metabolic pathways include reduction of the side-chain carbonyl along with several oxidized forms of the nitrogen and sulfur heteroatoms. Metabolites resulting from reduction of the carbonyl to a secondary alcohol require derivatization for GC/MS determination. The N-oxide and sulfoxide moieties are labile a t the elevated temperatures associated with GC conditions and are therefore not amenable to this technique. When the sulfoxide (HPPS) metabolite 2-(l-hydroxyprop-l-yl)promazine is introduced into the APCI ion source with the heated nebulizer LC/MS interface, the mass spectrum is dominated by the (M + 1)+ion (Figure 1). The determination of phenothiazine tranquilizers in racehorse urine by high-speed APCI LC/MS is demonstrated in Figure 2. The upper trace (Figure 2A) is the UV chromatogram acquired simultaneously with the lower (Figure 2B) APCI LC/MS total ion current (TIC) chromatogram and extracted ion current profile for m / r 359 (Figure 2C). The samples were injected a t 1-min intervals beginning with standard HPPS, followed by a control equine urine extract, and the urine extract from a racehorse thought to contain the metabolites of propiopromazine based upon preliminary TLC screening. Comparison of the chromatogram of the control extract with the positive urine extract in both the UV and TIC traces shows no indication for the presence of the parent drug or metabolites. The LC/MS extracted ion current profile for m / z 359 in Figure 2C indicates the presence of the metabolite HPPS. The mass spectrum (Figure 3) taken from the region in the total ion current chromatogram of the positive urine extract where the metabolite is expected to elute suggests an (M + 1)+ion a t m / z 359. The lower mass region of this spectrum is complex due to the coelution of interfering components in the sample. The APCI LC/MS mass spectrum for the synthetic standard of the metabolite (Figure 1)shows a simple spectrum with the majority of the ion current residing in the (M + I ) + ion. As described above, the technique of MS/MS can readily deal with the interference in the full scan spectrum shown in Figure 3. The mixture analysis capability of MS/MS allows

4.0

=

C Minutes)

Time

lael

e l . . e. 0

If

If

Time

,

(

.

2 0

r

.

,

.

4.0

f-IKTECTIOFS (Minutes)

h!

4

Time (Minutes)

Figure 2. (A) UV chromatogram of

standard HPPS (100 ng), control urine extract, and the extract from the urine of a horse administered propiopromazine. Samples were injected at 0, 1, and 2 min, respectively. (B) LC/MS total ion current (TIC)chromatogram (200-400 amu). (C) LC/MS extracted ion current profile of m l z 359 for the Same set of samples described above. us to utilize the LC/MS/MS mode and focus only ions of interest into the collision cell of the tandem mass spectrometer. In addition to separation, increased specificity over the simple mass spectra available from APCI is achieved with the fullscan CAD daughter ion spectrum. For most compounds we

2456

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12, OCTOBER 1986

A I

I

'J

359-b30

- 359

'OSITIVE

i

WIN

cxrm=l

Figure 3. LClMS APCI mass spectrum obtained at 2.6 rnin for the urine extract. This spectrum is a composite of several unknown components including possibly HPPS (mol wt 358).

have studied with this system, optimal fragmentation and daughter ion current signal are obtained with a laboratory collision energy between 50 and 80 eV. After an appropriate collision energy was chosen, the HPLC experiment described previously (Figure 2) was repeated in the MS/MS mode. The LC/MS/MS daughter ion current profiles (Figure 4) result from the CAD of the m / z 359 (M l)+ions. The ion current profile shown in Figure 4B represents that observed for the m / z 359 86 CAD fragmentation process. The signal observed for the positive urine extract (at 2.7 min) in the ion current profile of Figure 4 has the same retention time as that of the authentic material (0.7 min). No signal is observed at the corresponding retention time in the control extract (1.7 min). The full-scan daughter ion mass spectrum for the 2.7-min analyte in the positive urine extract (Figure 5B) compares favorably with that of the authentic material (Figure 5A) and confirms the presence of 2-(l-hydroxyprop-l-yl)promazine sulfoxide in the urine extract of the racehorse. The three injections of the standard material, control urine extract, and extract of the racehorse positive urine (Figure 4AB) were completed within 4 min. Promazine. The determination of phenothiazine tranquilizers in racehorses is complicated by extensive metabolism of the administered drug. Illustrated in Figure 6 are the structures of promazine and three of its major metabolites found in equine urine (20). Although capillary GC/MS determination is possible for parent promazine and 3hydroxypromazine (the latter requires derivatization), the elevated temperatures associated with GC precludes facile GC/MS determination of the more thermally labile promazine N-oxide and promazine N-oxide sulfoxide metabolites. The underivatized metabolites can be chromatographed and successfully ionized by using LC/MS techniques. The LC/MS analysis of equine urine for 3-hydroxypromazine was done by making successive injections (1 per min) of 50 ng of standard 3-hydroxypromazine, a control equine urine extract, and the extract of a positive racehorse urine. This horse urine sample was thought to contain metabolites of promazine based upon TLC screening the required mass spectral confirmation. The LC/MS TIC trace (Figure 7A) does not reveal resolution of the metabolites from interfering endogenous components of the urine extract. The APCI LC/MS mass spectrum in the urine extract (not shown here) is a mixed spectrum similar to that observed for HPPS (Figure 3). The mixture analysis capability of tandem mass spectrometry can provide the final "sample cleanup" necessary for producing a clean mass spectrum. The LC/MS/MS TIC trace (Figure 7B) for a daughter ion scan of the (M + 1)+parent ion at m / z 301 for 3-hydroxypromazine records only the signal from ions characteristic of the targeted analyte in the complex

b 4.0

0.0

3 T 3i m e zt-ImcrIw C Minutes)

3 5 9 - b 86

+

+

(

Time minutes)

Figure 4. (A) LC/MS/MS T I C chromatogram of the same samples described in Figure 2. The ion current trace is the composite of all 30' the daughters resulting from CAD of the parent ion (359' 359'). (B) Extracted ion current profile for m / z 359 86 for the same samples described in Figure 2.

- -

+

urine extract. The full scan daughter ion spectrum (Figure 8) shows no interference (compare to standard in Figure 9A) and thus supports this contention. Full scan LC/MS does not have the specificity necessary to produce any measurable signal above the high chemical background (Figure 7A). The analysis of urine extracts for drugs that are extensively metabolized, as is the case with the phenothiazine tranquillizers, requires techniques that can screen for metabolites of those compounds that contain common structural features. The daughter ion spectrum of promazine and its metabolites is dominated by ions characteristic of the side chain of the parent molecule. Bond cleavages a and y to the aliphatic nitrogen atom formally correspond to daughter ions at m / z 58 and 86, respectively. The daughter ion spectra in Figure 9, parts B and C, lack a shift of 16 mass units to higher mass expected from the N-oxide moiety. This suggests that the acyclic N-oxide oxygen atom is lost during the collision process. Because characteristic daughter ions are produced from these

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12, OCTOBER 1986

c H ,CH ,c H I

ihl,+l

N(CH ,)

2457

POSITIM

CONTROL

359

U

1.0

0.0

4

2.8

4

4.0

3.0

4-

IKTECTIW

1 POSITIM U?IN

U

/

E Q)

L

=-S-f-

a

\coMR#

WIN

0 e

.

O

h

-

1

~

A'

l

l

-

I

-

M/Z

+

Figure 5. (A) CAD daughter ion spectrum of (M 1)' ( m l z 359) of standard HPPS. (B) CAD daughter ion spectrum of HPPS in a positive racehorse urine extract.

i

1

u

CH ,CH ,cH,N(cH,)~ I

D

I

1

htW:lOO 3-HYDROXY PROMAZINE

MW:SI6 I-HYDROXY PROMAZINE-N-OXIM

MW:300

MW-316

PROMAZINE-N OXID€ SULFOXIDf

Figure 8. Structures of promazine and three of its metabolites

phenothiazine compounds, parent ion scanning may be used to screen for unknown metabolites and other drugs belonging to the same chemical class ( 4 ) . An example of parent ion scanning with LC/MS/MS for the metabolites of promazine is illustrated in Figure 10. In this experiment the triple quadrupole mass spectrometer was operated in the CAD mode with a collision energy of 70 eV in order to provide a strong signal from the characteristic daughter ions at mlz 58 and 86. Quadrupole 1 was scanned from m / z 200 to 400 while quadrupole 3 alternately monitored daughter ions at m / z 58 and 86. Thus, those parent ions giving rise to daughter ions at m / z 58 and 86 are flagged as potential metabolites of promazine. The ion current profiles (Figure 10) display the signals observed from a single injection of a crude racehorse urine extract. Inspection of these data reveals five different chromatographic peaks: two peaks for parent ions of m / z 333, two peaks for parent ions of m / t 317, and one peak for a parent ion of m / z 301. The molecular weights for the corresponding compounds are presumed to be 332,316, and 300 since the parent (M + 1)+ions resulted from APCI. Multiple hydroxylations of the original drug could account for m / z 332 and 316 ions. This experimental protocol can be

lj,,I, ,I, ,

%

Ea

m

, . ,

,

,

,

,

,

,

,

, ,,,

,,

,

,

,

,

,

,

,

,

.j

1 8 8 1 2 8 1 4 8 1 6 8 1 m a 3 3 ~ 2 4 b 3 ~ a a s 3 a p s

m/ z

Figure 8. CAD daughter ion spectrum of 3-hydroxypromazine in the positive racehorse urine extract.

very useful when a complex sample must be evaluated in an effort to determine which components are related to the administered drug. Radiolabeled isotope techniques can also be used to trace metabolites, but no indication of structure is available. In addition, different radiolabeled metabolites that co-elute from the HPLC cannot be distinguished with the counting detector. When the tandem mass spectrometer is used as a detector, it can often distinguish co-eluting metabolites that have different molecular ions or differences in their daughter ion spectra provided the daughter ion spectra are known. Phenylbutazone. The examples of fast LC/MS/MS cited above demonstrate the rapidity with which this technique can perform analyses of complex samples. Rapid analysis, however, does not equate with high sample throughput in a laboratory. The analysis of a few samples in a short period of time does not necessarily imply that a large number of samples

2458

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12, OCTOBER 1986

I I Y

c

a

0

c

3

n

‘lL llL 1

u

.I.

-u

I

a

317 -*5 8 , 8 6

0.m

M/ 2

1.m

2.00

301-58,86

Y

P U

a n

0.00

< MW:316

3 158 M/

z

3 - 3

‘1 MWc316

+

Figure 9. LClMSlMS CAD daughter ion spectra of the (M 1)’ ions of (A) 3-hydroxypromazine, (B) promazine N-oxide sulfoxide, and (C) 3-hydroxypromazine N-oxide.

can be repetitively analyzed a t the same speed per sample. In the case of analyses performed with combined chromatography and mass spectrometry, crude extracts can impose limitations on the number of samples that can be analyzed over an extended period of time by causing degradation of the column and/or loss of sensitivity due to ion source contamination. One situation where large numbers of samples need to be analyzed for a few targeted compounds occurs when pharmacokinetic profiles of drugs and their metabolites need to be determined. After administration of a drug, samples of urine and plasma are taken at short time intervals over periods of days. The importance of the analgesic phenylbutazone to the racehorse industry requires a good understanding of the distribution of this drug and its metabolites in the body fluids of the horse for at least 2 days postadministration. We have used the technique of LC/MS/MS to perform such analyses on 30 duplicate crude extracts of plasmas and urines samples taken from a horse over a 48-h time period. The full-scan CAD daughter ion spectra for the (M + 1)+ ions of phenylbutazone (PB, Figure 1lC) and its major metabolites, hydroxyphenylbutazone (PBOH, Figure 11A) and oxyphenbutazone (OPB, Figure 11B), are shown. For sub-

i.m

2.88

Retention Time (Minutes)

Flgure 10. LClMSlMS parent ion scan of a single racehorse urine extract. Quadrupole 1 scanned from m l z 200-400 while quadrupole 3 alternately monitored daughter ions m l z 58 and 66. (A) Sum of extracted ion current of daughter ions mlz 58 and 86 originating from parent ion m l z 333. (B) Sum of extracted ion current of daughters mlz 58 and 86 originating from parent mlz 317. (C) Sum of extracted ion current of daughter ions m l z 58 and 86 originating from parent ion mlz 301.

sequent analyses selected reaction monitoring (SRM) was used where four daughter ions were monitored for each analyte. The daughter ions monitored are those mass-labeled in the corresponding spectra (Figure 11). Chromatographic conditions were adjusted by increasing the acetonitrile content of the mobile phase to elute the components of interest in less than 1 min. Phenylbutazone was the last eluting analyte at 43 s, with PBOH at 15 and OPB a t 23 s, respectively. When the HPLC mobile phase was changed, no reoptimization of the temperatures or gas flows of the heated nebulizer LC/MS interface was required. A 60-sample, 48-h phenylbutazone pharmacokinetic study in the horse is illustrated in Figure 12. Although PB, OPB, and PBOH were all monitored, only the ion current profile for the four daughter ions of P B is displayed in this figure. An autosampler was used to introduce the samples a t 1-min intervals. Five phenylbutazone standards (400,200,100,50, and 25 ng) followed by a zero-hour equine urine extract with 12 postadministration urine extracts and a zero-hour equine plasma extract with 11 postadministration plasma extracts are shown. The detection limit on standards was approximately 1ng. The signal for the 25-ng standard .appears weak due to the normalization factor in the chromatogram. The above analyses required 30 min, after which they were repeated to provide a duplicate set of data. The concentration/time profile of the drug in the urine and plasma samples is graphically portrayed (Figure 12). For accurate quantitation an internal standard would be required, but a t this rapid analysis rate an interspersion of external standards should suffice. As mentioned above, SRM of four daughter ions on each of the major metabolites (OPB and PBOH) was done in addition to monitoring those from PB. A 5-min “window” (Figure 13) of the 42-47-min region for the equine urine pharmacokinetic study (Figure 12) displays the ion current of one parent-daughter ion pair for each analyte. The hour of the postadministration urine sampling is labeled above each 1-min analysis. Despite the fact that some chromatographic

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12,OCTOBER 1986

B

concentrations of the metabolites observed during this time region of the experiment (24-32 h postadministration) is consistent with that reported in the literature (21). Inspection of the ion current profile for the 42- and 43-min injection times of PBOH in Figure 13 shows a small peak on the descending shoulder of the PBOH trace. This peak is clearly resolved from the PBOH and is also resolved from the OPB peak. This peak represents the presence of an unknown component with a presumed (M + 1)+of m / z 325 with a daughter ion of m/z 104, a highly likely candidate for another metabolite. A parent ion scan was performed (Figure 14) on the sample injected at 42 min in Figure 13. Quadrupole 1 was scanned m / z 250-350 with quadrupole 3 monitoring the daughter ion mlz 93. This daughter ion was chosen because it is present in the spectra of the administered drug and its known metabolites. Two unknown components appear as the third and fourth peaks in Figure 14B. These unknown components are likely metabolites of P B since their molecular weights are 16 daltons greater than that of the administered drug (hydroxylation) and they have at least two daughter ions common to the parent drug (mlz 93 and 104, 104 trace not shown). The absence of this peak along with those for the other metabolites in a zero-hour equine urine extract further supports the possibility that these peaks are metabolites of PB. This analysis is an excellent example of the value of having a chromatographic step prior to MS/MS analysis when compounds of the same nominal molecular weight and similar structure are present in the sample. Without chromatography the unknown metabolites discussed above would not have been observed. In fact, parent ion scanning of the sample without chromatographic separation of the sample shown in Figure 14 would show only two components instead of five. In this instance one would observe only one peak whose (M 1)+ion was m/z 325 and another component whose (M + 1)+ion was m / z 309. When there is a need for the capability to analyze complex samples containing polar and heat-sensitive compounds while maintaining high sample throughput such as is the case in large drug testing laboratories or in major drug metabolism studies, the technique described here could be very beneficial.

20

CH ,CH ,CH ,CH I

I

2459

MW:321

+

Figure 11. LCIMSIMS CAD daughter

+

spectra of the (M 1)' ion of (A) hydroxyphenylbutazone (Pf3OI-l). (8) oxyphenbutazone (OPB), and (C) phenylbutazone (PB).

resolution has been sacrificed for the sake of speed, good separation of the drug and its metabolites is achieved with these very short retention times (Figure 13). The relative

CONCLUSIONS The concept of rapid, reliable sample analysis is a desirable goal. Usually, so-called rapid tests are relatively simple, inexpensive, and nonspecific. Simplicity and low cost are important, but without adequate specificity, analytical tests can be in error and result in considerable problems. The repetitive 1-minute LC/MS/MS sample analysis times re-

0 Retantion Time

(Minutes1

FWre 12. One-hour 60-sample LC/MS/MS analysis of samples from a phenylbutazonepharmacokinetic study in the horse. SRM of four daughter ions of each of the following: PB, OPB, and PBOH. Only the ion current from PB daughter ions is shown here. Injections were made with an autoinjector at zero minutes and every integral minute thereafter. Plasma samples were taken from the horse at 0, 0.5,1. 2,4,6 , 8, 10, 12, 24, 32,and 48 h with the analyses done in that order. Urine samples were taken at 0, 1, 2,4,6 , 8, 10,24,26, 28,30,32,and 48 h with the analyses done in that order.

2460

ANALYTICAL CHEMISTRY, VOL. 58, NO. 12, OCTOBER 1986

-

PB 309-1 20 o p ......... ~ 325+108 PBOH- 325-1 04 24 hr

26hr

30 hr

2 8 hr

3 2 hr

100

I

C

e a

0

-

e:

Time t Minutes1

Figure 13. LCIMSIMS SRM ion current profile taken from the 4247min region of Figure 12. The trace Is the ion current from daughter ions m l z 120+, 108+, and 104’ from parent ions 309+, 325+, and 325+, respectively.

117A

d A

250-350+93

metabolism studies, it is imperative that the analytical techniques used combine these important features. The high cost associated with a tandem mass spectrometer can be offset if the instrument provides high sample throughput and solves problems that cannot be solved, or can be solved only with difficulty, by other techniques. This paper has demonstrated that mass spectrometric analysis of polar, labile compounds with high sample throughput is a currently available technology. Unfortunately, the complexity of an LC/MS/MS system requires experienced, qualified individuals for its successful operation. Hopefully this requirement may be lessened somewhat as “benchtop” LC/MS/MS systems become available and automation becomes more common. In the not-too-distant past, GC/MS was considered to be a very difficult and complex instrumental method. Today’s modern benchtop GC/MS systems have made the technique routine, and presumably the same will also be true for LC/ MS/MS. ACKNOWLEDGMENT We acknowledge assistance by E. A. Dewey with the drug administration trials and G. A. Maylin, director of the Drug Testing and Toxicology laboratory, for helpful discussions and continued support of this work. Registry NO.PB, 50-33-9; HPPS, 89453-67-8;PBOH, 568-76-3; OPB, 129-20-4;promazine, 58-40-2;3-hydroxypromazine,316-85-8; promazine N-oxide sulfoxide, 52208-24-9; 3-hydroxypromazine N-oxide, 103499-85-0.

LITERATURE CITED

7

309-93

R e t e n t i o n Time (Minutes) Figure 14. Total parent ion current profile for the urine extract shown in the 42-43-min region of Figure 13. Q , was scanned from m l z 250 to 350 with Q3 monitoring daughter ion 93. (B) Extractea ion current profile of parent daughter ion pair 325 92. (C) Extracted ion current 93. profile of parent daughter ion pair 309

--

ported here lack the simplicity and low cost associated with many immunoassay-related tests (22),but this technique does offer an unrivaled combination of selectivity, sensitivity, and speed. In today’s arena of forensic analysis, and in basic drug

(1) McLafferty, F. W., Ed. Tandem Mass Spectromefry;Wiley: New York, 1983. (2) Hunt, D. F.; Shabanowitz, J.; Giordani, A. B. Anal. Chem. 1980, 52, 386-390. (3) Kondrat, R. W.; McCluskey, G. A,; Cooks, R . G. Anal. Chem. 1978, 5 0 , 2017-2021. (4) Perchalski, R. J.; Yost, R. A.; Wilder, B. J. Anal. Chem. 1982, 54, 1466-147 1. (5) Brotherton, H. 0.; Yost, R. A. Anal. Chem. 1983, 55, 549-553. ( 6 ) Henion, J. D.; Maylin, G. A.; Thomson, B. A. J. Chromatogr. 1983, 277, 107-124. (7) Johnson, J. V.; Yost, R. A. Anal. Chem. 1985, 57,759A-768A. (8) Covey, T. R.; Henion, J. Presented at the 33rd Annual Conference on Mass Spectrometry and Allied Topics, May 26-31, San Diego, CA, 1985: n 813. (9) DiCesak J. L.; Dong, M. W.; Atwood, J. G. J. Chromatogr. 1981, 217. 369-386. (10) Lee,’E. D.; Henion, J. D. J. Chromatogr. Sci. 1985, 23, 253-264. (11) Vestal, M. L. Science 1984, 226, 275-280. (12) Whitehouse, C. M.; Dreyer, R. N.; Yamashita, M.; Fenn, J. B. Anal. Chem. 1985, 57,675-681. (13) Thomson, B. A.; Iribarne, J. 8. J . Chem. f h y s . 1979, 77, 4459-4464. (14) Henion, J. D.; Thomson, 8. A,; Dawson, P. H. Anal. Chem. 1982, 5 4 , 45 1-456. (15) Thompson, 8. A. Presented at the 31st Annual Conference on Mass Spectrometry and Allied Topics, May 8-13, Boston, MA, 1983; p’65. (16) Crowther, J. B.; Covey, T. R.; Sllvestre, D.; Henion, J. D. LC Mag. 1985, 3 , 240-254. (17) Levsen, K.; Beckey, H. D. Org. Mass Spectrom. 1974, 9 , 570-580. (18) Dewey, E. A.; Maylin, G. A. Cornell Vet. 1984, 7 4 , 38-49, (19) Dawson, P. H.; French, J. B.; Buckley, J. A.; Douglas, D. J.; Simmons, D. Org. Mass Spectrom. 1982, 17,212-216. (20) Dewey, E. A.; Maylin, G. A.; Ebel, J. G.; Henion, J. D. Drug Metab. Dispos. 1981, 9 , 30-36. (21) Maylin, G. A. Am. Assoc. €q. fracf. 1974, 243-248. (22) Chait, E. M.; Ebersole, R. C. Anal. Chem. 1981, 53, 682A-689A.

RECEIVED for review December 9, 1985. Accepted May 13, 1986.