Specific and Sensitive Measurement of FK506 ... - Clinical Chemistry

7 downloads 0 Views 1MB Size Report
May 20, 1992 - in a ratio 2:1, the result of cis-trans isomerism of the. C-N amide bonds. FK506 is soluble in alcohols, haloge- nated hydrocarbons, and ether.
CLIN. CHEM. 38/10, 2025-2032 (1992)

Specific and Sensitive Measurement of FK506 and Its Metabolites in Blood and Urine of Liver-Graft Recipients Uwe Christians,’ Felix Braun,’ Michael Schmidt,2 Norbert Kosian,2 Hans-Martin Schiebel,3 Ludger Ernst,4 Michael Winkler,5 Christiane Kruse,’ Annette Linck,’ and Karl-Friedrich Sewing’ A specific and sensitive assay for quantifying the immunosuppressant FK506 and its metabolites in blood and urine was developed. 32-O-Acetyl FK506 was synthesized and used as internal standard. FK506 and its metabolites were purified from the samples by solid-liquid extraction and were injected into a high-performance liquid chromatographic (HPLC) system linked to a mass spectrometer (MS) by particle-beam interface. The FK506 derivatives were separated from interfering material by use of a 100 x 4 mm C8 analytical column and water/ acetonitrile or water/methanol gradient elution; they were detected by negative chemical ionization with methane as reagent gas. The limit of detection was 25 pg in a standard solution, and the limit of quantification in blood was 250 pg (extracted from 1 mL of blood). The CV was

tie

FK506

fraction FK506

3 +OH

11.3% at 5 ng, and no interferences with other drugs were found. fraction AddItIonal Keyphrases: immunosuppressive chromatography, liquid

.

drugs

urine

‘Institute f#{252}r Aligemeine Pharmakologie, Medizimsche Hochschule Hannover, Konstanty-Gutschow-Str. 8,3000 Hannover 61,

FRG. f#{252}r G’eoanalytik,

2 -

2 CH,

mass spectrometsy

FK506 (Fujisawa, Osaka, Japan) is considered a valuable and potent immunosuppressant after organ transplantation (1) and in the therapy of immunological diseases (2, 3). FK506 is isolated from Streptomyces tsukubaensis (4) and has a macrolide structure (CH69NO12) with a molecular mass of 803.5 Da (Figure 1). In solution (CDC13), FK506 forms two rotamers in a ratio 2:1, the result of cis-trans isomerism of the C-N amide bonds. FK506 is soluble in alcohols, halogenated hydrocarbons, and ether. It is very sparingly soluble in aliphatic hydrocarbons and water. The molecule does not contain any chromophores, and its ultraviolet absorption maximum is 192 nm (5-7). FK506 is metabolized by the intestinal and liver cytochromeP-450 system to at least nine metabolites (8). There is strong evidence that cytochrome P-450fflA4 is responsible for the FK506 metabolism (9). The reactions involved in FK506 metabolism are O-demethylation and hydroxylation. The metabolites formed retain biological activity. The immunosuppressive activity of the metabolites is 10% of that of FK506 (8, 10). All of these metabolism studies were performed in vitro with human

2Labor

FK506

Hildesheim,



Institut

f#{252}r Orgarnsche

Chemie, and4 Institut Mr Anorganische mid Analytische Chemie, Technische UmversitAt Braunschweig, 5Klinik f#{252}r Abdominalund Transplantationschirurgie, Medizinische Hochschule Hannover. Received January 31, 1992; accepted May 20, 1992.

fractiont3.45 FK506 CH, -

+

OH

CH3+20H a2CH3+ OH -2CH3 +20H -2CH3 + 30H -3CH#{247}20H -

Fig. 1. Structure and metabolic pathways of FK506 and its metabolites The carbon atomsare numberedaccording to Jones et at. (7)

liver microsomes. No data concerning the in vivo metabolism of FK506 are available, because of the lack of a specific assay sensitive enough to allow quantification in body fluids or tissues of patients. The structure of FK506 and its metabolic pathways are displayed in Figure 1. Thus far, six different methods for quantifying FK506 have been described. The assay that was used to measure plasma concentrations of FK506 for clinical and pharmacokinetics studies (11-13) is an enzyme immunoassay, based on use of a monoclonal antibody (14-16) that probably is cross-reactive with the FK506 metabolites, according to clinical observations (17). An automated whole-blood assay for the IMx analyzer (Abbott Labs, Abbott Park, IL) was developed on the basis of this antibody (18). HPLC/ultraviolet detection (UV)6 was used to isolate and quantify FK506 metabolites in 6Nonstandard

abbreviations:

HPLC/UV,

high-performance

liq-

uid chromatography

with ultraviolet detection; MS, mass spectrometry; DCI, direct chemical ionization; FAB, fast atom bombardment; and NMR, nuclear magnetic resonance. CLINICAL CHEMISTRY, Vol. 38, No. 10, 1992 2025

in-vitro studies (8, 10). The main disadvantage of the method is a detection limit of 10 g/L, which is above the therapeutic range of the plasma and blood concentrations of FK506 usually expected in patients. To enhance the sensitivity of the HPLC assay, Friob et al. (19) collected fractions of FK506 and one metabolite after HPLC separation, which they quantified by enzyme immunoassay. Takada et al. (20) enhanced the sensitivity of their HPLC assay by derivatizing FK506 with dansyl hydrazine and using chemoluminescence detection. Zeevi et al. (21) developed a bioassay to

1

fraction

o

5

I

iO

hme(m

monitor

the immunosuppressive activity of plasma of FK506-treated patients. None of these assays, however, allows the specific and appropriately

sensitive

quantification

of FK506

and all

of its metabolites, a measurement that is required to assess the pharmacokinetics of the drug and to investigate whether

the metabolites

are of clinical

significance. 2#{228}

fime(min)

MaterIals and Methods Apparatus For HPLC/mass spectrometry (MS) analysis, we used a Model 109011L liquid chromatograph connected by a particle-beam interface to an HP5989A mass spectrometer. Data were recorded and analyzed by an HP-TJX ChemStation. HPLC/LJV analysis was performed with a Model 1090A chromatograph equipped with a Model 1040 diode-array detector and an HP85B data processor and integrator unit (all from Hewlett-Packard, Waldbroun, FRG). For direct chemical ionization (DCI) and fast atom bombardment (FAB)-MS, we used a Finnigan MAT 8430MS (Finnigan MAT, Bremen, FRG); for structural analysis of the internal standard, we used a Bruker 400-MHz nuclear magnetic resonance (NMR) spectrometer (Bruker, Karlsruhe, FRG). Chemicals

and Reagents

The analytical HPLC columns were filled with Nudeosil#{174} C8 (3-gm particles, 100-nm porewidth; Macherey Nagel, Duren, FRG). All solvents were of HPLC quality and purchased from Merck, Darmstadt, FRG. The Chromsystems extraction kit for cyclosporine and its metabolites was a gift from Chromsystems (Munich, FRG), and FK506 was from Fujisawa Pharmaceuticals. FK506 and its derivatives were stored in acethnitrile/ water (75/25 by vol), pH 3.0, at 4#{176}C. The solution was stable under these conditions for 4 weeks. and Structure of the Internal Standard We incubated 5 mL of 1 g/L FK506 reagent in acetonitrile/water (70/30 by vol) with 5 mL of acetic anhydride at 75 #{176}C for 120 mm. Acetic anhydride was evaporated at 60#{176}C under a stream of nitrogen. The residue Synthesis

was dissolved vol), pH 3.0,

in 1 mL of acetonitrile/water (75/25 by and injected into a preparative HPLC

system, consisting

of two sequentially

linked 250 x 10

filled with Nucleosil. FK506 and its acetylderivatives were eluted by gradient elution. Water used in the mobile phase was adjusted to pH 3.0 with sulfuric acid. A linear gradient was run from 48% to 51% mm columns

2026 CLINICAL CHEMISTRY, Vol. 38, No. 10, 1992

8L6 3000 200O 00

0500

I

I

-

0

5

m,i

Fig. 2. Isolation of the internal standard, 32-O-acetyl FK506 T: Chromatogramafterincubation of 5 mg of FK506 at 75 ‘C for 2 h and fractions collected. Middle: Chromatogramof the isolated fraction 3. Bottom: DCI-MSof the isolated fraction 3, which contained 32-O.acetyl-FK506

acetonitrile

at a flow rate of 5.5 mL/min; the detection of the monitor was 205 nm. The oven temperature was 75#{176}C, the injection volume 250 ML. Fractions were manually collected. To identify the structures, we mixed the isolated fraction with an equal volume of dichioromethane. The sample was vortex-mixed and centrifuged (2 mm, 2000 x g), and the dichloromethane layer was evaporated. For analysis by DCI-MS, we reconstituted the residue in methanol at -0.5 g/L. For NMR analysis, we dissolved the residue in 500 L of CDC13 (Merck) containing tetramethylsilane, 0.5 g/L, as the internal shift reference. The chromatograms and mass spectra are shown in Figure 2. The FK506 derivative with an HPLC retention time of 40.5 mm (fraction 3, Figure 2) showed a mass spectrum with a molecular ion at 846 amu, which indicated an acetylation in one position. The structure of this acetyl-FK506 was derived from 400 MHz ‘H-NMR experiments and comparison with FK506, including homonuclear decoupling and twodimensional homonuclear shift correlation (H,H-COSY90). Acetyl-FK506 showed a shifted ddd absorption (J = 11.4, 9.2, 4.8 Hz, all ± 0.2 Hz) at 8 = 4.68, which was assigned to the hydrogen geminal to the acetylated hydroxyl group. In the COSY spectrum, this hydrogen displayed spin-spin coupling to a ddd [11.3, 9.4, 45 (± 0.2) Hz] at 8 = 3.22. The derived partial structure wavelength

CH(OAc)-CH(OCH3) is part of the cyclohexyl residue; i.e., the acetylated hydroxy group is situated at C-32. Like FK506, the acetyl derivative occurs in the form of two rotational isomers in a ratio of 2:1, due to cis-trans isomerism of the C-N amide bond. For use as internal standard, 10 L of the isolated fraction was injected into an analytical HPLC/UV system. FK506 and (or) the internal standard were eluted from the analytical columns (250 x 4 mm, C8, 3-gm particles) by a linear acetonitrile/water (pH 3.0) gradient: at time 0 mm, the mobile phase was 45% acetonitrile; at time 40 mm, it was 55% acetonitrile. The flow rate was 0.7 mL/min, the oven temperature 75#{176}C, and the detection wavelength 205 nm. The internal standard was quantified by using the FK506 calibration curve. The concentration of the internal standard was adjusted with acetonitrile/water (75/25 by vol) to 1 j.g/L.

TT’TTIT,

Extraction of FK506 and Its Metabolites We tested the extraction procedure described by Christians et al. (8, 10) and a commercially available extraction kit for cyclosporine and its metabolites (Chromsystems), both of which were based on solidliquid extraction. For the former procedure, we added 10 pL of the internal standard solution (10 ng of 32-0acetyl-FK506) and 2.1 mL of acetomtrile/water (70/30 by vol) to 1 mL of EDTA-anticoagulated blood or 1 mL of urine (8, 10). After vortex-mixing and centrifuging (2 miii, 2000 x g) the samples, we aspirated the supernates through glass extraction columns filled with LiChroprep#{174} (25-40 pm, C8; Merck), which had previously been primed with 3 mL of acetonitrile and 3 mL of water (pH 3.0). The samples were washed with 3 mL of methanol/water (50/50 by vol) and 1 mL of hexane. FK506 and its metabolites were eluted by centrifuging 1.5 mL of dichloromethane through the columns. The dichioromethane was evaporated and the residues were dissolved in 250 pL of acetonitrile/water (75/25 by vol) and washed with 500 L of hexane. Using the Chromsystems kit, we added 10 pL of the internal standard solution to 1 mL of blood or urine. We followed the extraction process described for cyclosporme and its metabolites as described in the kit manual. The Chromsystems kit yielded considerably cleaner extracts, with fewer substances interfering with the HPLC/UV assay of FK506 and the internal standard (Figure 3). The absolute recoveries of 75-90% were not significantly different between the two extraction procedures. Therefore, we used the Chromsystems extraction procedure during the study. Tests with the internal standard (32-O-acetylFK506), FK506, and an isolated metabolite (demethylFK506) showed no significantly different recoveries from one another.

time (nfl)

Fig. 3. Extraction of FK506 and its metabolites by using the Chromsystems extraction kit (A, B,) or according to Christians et al. (8, 10) (C, D) A, C: bloodsamplessupplementedwith 50 ng of FK506and 50 ngof Internal standard;B, 0: blank blood samples. A.’iows mark the retention times of FK506andthe internalstandard.HPLC/UV:250 x 4 mm (C0,3-turnparticles) analyticalcolumn,202 nm detection wavelength, acetonltrile/water (pH 3.0) gradientelution(see text), flowrate 0.7 mL/min,columntemperature 75 ‘C

al), the samples were evaporated by centrifugation der reduced pressure and redissolved in 20 L acetomtrile/water (70/30 by vol); 10 pL of this injected into the HPLC system. The same gradient run as described for normal-bore HPLC, but the rate was 0.25 mL/min.

unof was was flow

HPLC/MS of FK506 and Its Metabolites We injected 150 pL of the extract into the HPLC. FK506 and its metabolites were eluted from a 100 X 4 mm analytical column by using the following acetonitrue/water gradient: 0 mm, 60% acetonitrile; 8 mm, 80% acetonitrile; 8.1 mm, 95% acetonitrile; and 12.5

mm, 95% acetonitrile. lowing methanol/water 8 mm, 80% methanol;

Alternatively, we used the folgradient: 0 mm, 70% methanol; 8.1 mm, 95% methanol;

14 min,

HPLC/UV

95% methanol. The column temperature was 40#{176}C, the flow rate 0.3 mL/min. The source temperature of the MS was 250 #{176}C and the quadrupole temperature was 120 #{176}C. For MS analysis we used chemical ionization with methane or butane (purity >99.5%; Messer, Griesheim,

The same columns and HPLC conditions as described for quantifying the internal standard by HPLC/UV were used, but with detection at 202 nm. For narrowbore HPLC (100 x 2.1 mm, filled with 3-pm C8 materi-

FRG) at 160 Pa and detected negative ions. The autotune option was used and the multiplier current was set to 3000 V. The MS was run in the selected-ion mode focused on the following masses: 776 amu (double deCLINICAL CHEMISTRY, Vol. 38, No. 10, 1992

2021

methylated

FK506 derivatives),

790 amu (demethylat-

ed FK506 derivatives), 792 amu (double demethylated and hydroxylated FK506 derivatives), 804 amu (FK506), 808 amu (demethylated, hydroxylated FK506 derivatives), 836 amu (double hydroxylated FK506 derivatives), and 846 amu (internal standard, 32-0-acetyl-

FK506). To quantify

FK506

and its metabolites,

recovery,

and concentrations using an FK506 calibration curve.

we calculated

were

calculated

by

Identification of the FK506 Metabolites

were incubated with FK506 and an NADPH-regenerating system for 15 mm at 37 #{176}C. The reaction was stopped by adding acetonitrile, and FK506 and its metabolites were purified by solid-liquid extraction. The samples were injected into a preparative HPLC system consisting of two sequentially linked 250 x 10 mm columns packed with C8 (7-pm particles). The flow rate was 5.5 mL/mun, the column temperature was 75#{176}C, and the signal was monitored at 205 mm. FK506 and its metabolites were separated by using a concave acetonitrile/ water gradient: at analysis time 0 mm, 37% acetomitrue; at 20 mm, 51% acetonitrile; at 35 miii, 85% umn

washing

The elutiom was followed by a 5-mn step with

95% acetonitrile

col-

and reequili-

bration to the start conditions within 10 mm. Fractions were manually collected. The metabolites were extracted from the isolated fractions by liquid-liquid extraction with an equal volume of dichloromethane. The dichloromethane layer was separated and evaporated. The residues were dissolved in 10 L of methanol, and the structures of the metabolites were identified by DCI-MS, taking into account molecule ions and characteristic fragments. Butane was used as the reagent gas and negative ions were detected. The isolated metabolites were quantified by HPLC/IJV, as described for the internal standard, with use of an FK506 calibration curve. These metabolites were used for identifying peaks and for preparing calibration curves in the HPLC/MS system. Quality Assessment Quality controls. EDTA-anticoagulated blood was supplemented with FK506, 5 or 25 g/L, and 1-mL portions were transferred imto 10-mL centrifuge tubes.

Samples were incubated at 37#{176}C for 30 mm and frozen at -20 #{176}C until use. Calibration controls. Each calibration curve comprised five data points at concentrations of 0, 1, 5, 10, and 50 g/L with n = 5 per data point. We added the respective FK506 concentrations to blood samples, incubated them for 30 mm at 37 #{176}C, and froze these standards at -20 #{176}C until use. In a set of 10 samples assayed, 2 samples were quality controls, and 2 samples 2028

Ciprofloxacin Cyclosporine metabolites

Assay Interfered with

18.2

Metabolites

33.5, 35.2

FK506

Ketoconazole Lidocaine

29.3, 35.5 17.9 33.4 32.8 15.9, 17.3 18.7

FK506 Metabolites FK506 FK506

Midazolam Nicardipine Omeprazole Vancomycin

Metabolites Metabolites

(pH 3.0) gradient:time 0 mm, 43% acetonltrlle; 20 mm, 50% acetonitrile;35 mm, 59% acetonitrile; 50 mm,25% acetonitrlle; detection wavelength, 205 nm;otherconditions as In Fig. 3. Acetonitille/water

FK506 metabolites were isolated as previously described (8, 10). Human liver microsomes, isolated by standard differential centrifugatiom techniques (22),

acetomtrile.

Retention time, mln

Drug

(AM1A, AM1)

the recovery of the internal standard. The peak areas of FK506 and its metabolites were corrected according to this

Table 1. Interference of Other Drugs wIth HPLC/UV Assay of FK506 and Its Metabolites

CLINICAL CHEMISTRY, Vol.38, No. 10, 1992

were calibration

controls. We measured 10% of the patients’ samples twice. Cross-validation with the FK506 enzyme immunoassay of plasma. For cross-validation with the enzyme immunoassay of FK506 in plasma, we selected without conscious bias 25 blood samples from 17 different livergraft patients (no more than two samples per patient) and measured FK506 by both methods. Interferences of other drugs with the assays. To evaluate interferences of other drugs with the chromatographic analysis of FK506, we dissolved these drugs in acetomtrile/water (50/50 by vol) at a concentration of 1 g/L and injected 25 L into the HPLC system for analysis as described above. We tested the following drugs: acetylsalicylic acid, acyclovir, amikacin, ampicillin, atenolol, azathioprine, azlocillin, cefotaxime sodium, cyclosporime, cilastatin sodium, cimetidine, ciprofloxacin, clomdine hydrochloride, dexamethasone, diazepam, diitiazem, dopamune, epinephrine hydrochloride, erythromycin, flucloxacillun sodium, ganciclovir, gentarnicin, imipenem, ketocomazole, lidocaine, mezlocillin, midazolam, micardipune hydrochloride, nifedipine, omeprazole, piperacillin sodium, ranitidine, rapamycun, rolitetracy-

clime, spironolactome,

sulfamethoxazole,

triamcunolone,

trimethoprim, and vancomycin hydrochloride. For HPLC/MS analysis, we searched the Wiley/NBS (John Wiley and Sons, Chichester, UK) and Pfleger/ Maurer (VCH, Weinheim, FRG) MS spectra libraries for

potential

interferences.

Results The limit of quantification

of the HPLC/UV

assay was

15 mg of FK506, the limit of detection was 5 ng [determined by injection of 10 pL of a 500 zgfL solution of FK506 in acetonitrile/water, pH 3.0 (75/25 by vol)], and the calibration curve in blood was linear to 225 ng (r = 0.998). The interassay coefficient of variance (CV) at 100 was 9.5% (SD 1.3%) (ii = 8); close to the detection limit, it was >50%. Several drugs commonly used after organ transplantation were found to interfere with the HPLC/UV assay (Table 1). We tested various columns, filled with C18, C8, endcapped C8, C4, and cyano-propyl materials. FK506, the internal standard, and the me-

tabolites were eluted from the alkyl-modified materials in a double-peak elution pattern, caused by cis-trans isomerism of the C-N amide bond. Only the cyanopropyl modified silica gel resulted in a single but broad peak. There was no difference between endcapped and nomendcapped material. The use of narrow-bone HPLC had no significantly positive effect on the detection limit. After incubation of FK506 in acetic anhydride, DCI-MS detected single-, double-, and triple-acetylated derivatives. The conditions described gave the highest

yield of 32-O-acetyl FK506. The double- and tripleacetylated derivatives were less stable during ionization than was 32-O-acetyl-FK506. The temperature of the MS source and the use of

A

I.

0

The limit of detection

of the HPLC/MS

8

10

12

S.. 3H

tral analysis.

sensitivity.

6

B

negative chemical ionization led to almost mo fragmentation of FK506 and its metabolites during mass spec-

The use of a higher source temperature enhanced the fragmentation, especially of the metabolites, whereas a lower source temperature led to faster contamination of the MS source, with a resulting loss of

2

0

8

4

12

C

assay I0

was 25 pg after flow injection of 10 L of FK506 standard solution [FK506 at 2.5 pg/L in acetonithle/ water, pH 3.0 (70/30 by vol)], resulting in a signal-tonoise ratio of 8:1. The mobile phase was acetonitrile/ water (90/10 by vol). The calibration curve was linear from 25 pg to 50 ng (r = 0.999). Flow injection of the extracted blood samples showed peaks that interfered with FK506. This interfering material could be separated from FK506 and its metabolites by using a 100 x 4 mm C8 column and gradient elution (Figure 4). In a first step, the gradient eluted the interfering material and then eluted the FK506 metabolites, FK506, and the internal standard. A complete chromatographic separation was not necessary, because the metabolites could be differentiated by their molecular ions. It was much more important to elute the compounds of interest at almost the same time because the sensitivity of the MS detec-

tion was dependent on the acetonitrile/water or methanol/water ratio of the mobile phase. The use of a gradient allowed us to elute FK506 and its metabolites from the columns at a composition of the mobile phase that yielded the greatest sensitivity during particle beam separation and MS analysis (Figure 5). The sensitivity of the assay was best at a methanol/water volume ratio

of 90/10. The limit of quantification in blood was 250 pg, determined by extraction of 1 mL of blood and injection of 150 pL of the extract. The intra-assay CV was 10.5% at 5 ng (n = 10). The imterassay CV was 12.3% at 1 .gfL, 11.3% at 5 p.g/L, and 13.2% at 50 g/L (all with n 10). The calibration curve after extraction of FK506 from blood had a correlation coefficient (r) of 0.998. Analysis of extracts of human liver microsomal preparations after metabolism of 10 nmol of FK506 showed that all metabolites previously identified could be detected by HPLC/MS (Figure 6). No other clinically relevant drugs were found to interfere with the assay. Stability of the HPLC/MS system was tested by 50

40

4

6

8

12 tIa.

(aug

Fig.4. HPLC/UV and HPLCMS analysis for FK506 in blood samples A, UV chromatogram of an extracted FK506-free blood sample; B, the same sample after InjectionInto the HPLC/lvlSsystem;C, HPLC/J4Sanalystsof a blood sample with 2 ng of FK506 and 5 ng of the Internalstandard32-0-

acetyl-FK506 added. HPLC/UV conditions: 100 x 4 mm (Ce, 3-Mmparticle) analytical column.Acetonitiile/water gradient as follows: 0 mm,60% acetonltIle; 8 mm, 80% acetonitrlle; 8.1 mm, 95% acetonftnile;12.5 mm, 95% acetonitn , flow rate, 0.3 mL/mln. Columntemperature: 40 ‘C, detection wavelength: 205 nm.HPLC/MS:chromatographicconditionsas described for HPLC/UV; reagentgas, methane;sourcetemperature, 250 ‘C; quadrupcle temperature,120 ‘C. The detection mass was changed from 804 to 846 amu at 10.5 mm 400 U U

V

a I.)

0 K

200

100

0 0

10

20

30

40

50

60 % water

Fig. 5. AssociatIon of detection signal of the HPLCMS analysis and composition of the mobile phase Repeatedflow Injections of long of FK506: mobile phase, methanol/water; flow rate, 0.3 mL/mln;columntemperature, 40 ‘C; reagent gas, methane; sourcetemperature,250 ‘C; quadrupole temperature, 120 ‘C. Shown are mean ± SD (n = 3)

repeated injections of 10 ng of FK506, with one injection every 1.5 mm, beginning right after starting the HPLC pump. It took 20 mm until a stable signal could be detected. The signal was stable for the other injections, CLINICAL CHEMISTRY, Vol. 38, No. 10, 1992 2029

WI

Abc-jancp

,,mu

804

to,,

,;n)

9

6

FIg. 6. HPLC/MS analysis of an extract after incubating 10 nmol of FK506 with human liver microsomes for 7 mm Isolationof humanlivermicrosomes and incubationconditionsas describedby Christianset ai. (8, 10). HPLC/MS:conditionsas in Fig.4. TiC, totalioncount 90

a slight increase in sensitivity of -20%, which could be attributed to a time-dependent change of the shape of repeller in the MS source. Long-term stability could be evaluated by analysis of quality-control samples during a sequence of 75 extracted blood samples; the change of signal was 9.3% (n = 8), which is almost equal to the intra-assay CV described above.

a

rnu

with

Cross-validation of the HPLC/MS blood assay of FK506 and its metabolites with an enzyme immunoassay of FK506 in plasma in 25 patients showed no significant correlation between both assays. With the system described, as many as 75 samples could be run before the skimmers of the particle beam

time Cm,,,)

Fig. 7. HPLC/MS analysis of a blood samplefrom a liver-graft recipient TIC, total ion count; 804 amu, FK506; 790 amu, demethylFK506. HPLCIMS conditions as in Fig. 4 Ab.ord.n,.

interface had to be cleaned or exchanged. The source of the MS had to be cleaned every 200-250 samples. In blood samples, FK506 and a demethylated metabolite at 790 amu (demethyl-FK506), were found (Figure 7); in urine, FK506 was found, as were metabolites 776 (di-demethyl FK506), 790 (demethyl FK506),

792 amu (di-demethyl

hydroxyl

at and

FK506) (Figure 8).

DiscussIon The HPLC/MS assay had several advantages over the other assays tested. The HPLC/UV assay was not sensitive enough. Most patients measured in this study had trough blood concentrations of FK506 in the range of 5-10 g/L, which was below the detection limit of the HPLCIIJV assay. The major problem of HPLC of FK506, its metabolites, and the internal standard is the double elution pattern caused by the cis-trans isomerism of the C-N amide bond. Especially in low concentrations, FK506 and its derivatives were eluted as broad peaks, with incomplete separation of the rotamers, which caused

eral

high intra-assay

clinically

variability. important drugs

Furthermore, interfere with

sevthe

HPLC/UV assay. Detection of FK506 by MS had the advantage that, besides the different HPLC retention times and the 2030 CLINICAL CHEMISTRY, Vol. 38, No. 10, 1992

I 3

6

9 to,.

ft

-r’#{149}-’-r’-’’ I--3

6

tiM

9 h,n)

Fig. 8. HPLC/MS analysis of a 24-h urine sample from a liver-graft recipient The sample was collectedfrom thesamepatientonthe sameday as the blood sample in Fig. 7. TIC, total ion count; 804 amu, FK506; 776 amu, di-demetbyl FK506; 790 amu, demethyl FK506; 792 amu (not shown, but present in the patientsurine),di-demethythydroxyl FK506.HPLCIMSconditionsas in Fig. 4

different m/z values, a second criterion was available to identify FK506 and its metabolites. With HPLC alone, only an incomplete separation of the metabolites was possible (8,10). The specific detection by MS allowed us to choose HPLC conditions that reduced peak broadening and the double-peak elution of FK506. The best

were obtained with a short (100 x 4 mm) C8 column. Sensitivity of the MS detection depended on the

results

flow rate of the mobile phase: 0.3 mL/min gave the best results. Using an increased column temperature of 75 #{176}C in the HPLC/IJV assay (8, 10) increased the isomerization rate of the drug, which resulted in narrower peaks. However, because the step gradient and the analytical column used for HPLCIMS are not able to separate the FK506 rotamers, use of higher temperatures gave no advantage in the latter assay. The coluxnns were maintained at 40#{176}C, which was enough above room temperature to be maintained by the coluinn oven. Using either acetonitrile or methanol in the mobile phase made no significant difference in the detection limit, but the method involving methanol resulted in sharper peaks. Liquid-liquid and solid-liquid extraction of plasma samples gave different results in the enzyme immunoassay-presumably because of different recoveries of the FK506 metabolites and cross-reaction of the antibody used for enzyme immunoassay with the metabolites. Thus, we checked the recoveries of isolated metabolites in the solid-liquid extraction methods used and found these not to be significantly different from that of FK506 or the internal standard. Solid-liquid extraction procedures were easier to handle for a large number of samples and could also be automated (23). The use of an internal standard was required in the HPLC/MS assay for two reasons: to compensate for losses during extraction, and to compensate for the variable detection sensitivity of the MS. The more samples we ran on the MS, the more the skimmers and the MS source were contaminated, and the more the sensitivity decreased. The question of whether to use blood or plasma for therapeutic drug monitoring of FK506 is still under discussion (24, 25). One major disadvantage of plasma as matrix is that the distribution of FK506 between blood cells and plasma depends on the temperature and hematocrit of the sample (26). Furthermore, it is not known whether the distribution coefficients between blood and plasma at a given temperature are the same for FK506 and its metabolites. For practical reasons, blood is much easier to handle, is reproducible, and is not subject to the erroneously high concentrations of FK506 seen in hemolyzed plasma. The problem of choice of matrix was exhaustively considered for the immunesuppressant cyclosporine as well; today, blood is the generally accepted matrix of choice (27,28) because the cyclosporine concentrations in blood show better correlation than plasma values with such clinical events as rejection and (or) cyclosporine toxicity, and methods based on blood can easier be standardized. Furthermore, the FK506 concentrations determined in plasma by enzyme immunoassay are in the ngfL range, which is at the detection limit of the analytical methods available. Concentrations of FK506 in blood are 20- to 50-fold higher than those in plasma (24,29), and thus are in a concentration range that allows for more reliable and

reproducible quantification. For these reasons we selected blood as matrix for this study. The lack of a significant correlation between enzyme immunoassay in plasma and HPLC/MS in blood is mainly due to the different matrices used. As discussed above, the FK506 concentration in plasma is dependent on concentration, hematocrit, and temperature (24-26). Thus far, only two specific assays for FK506 have been available with a sensitivity suitable for measurement in plasma or blood of patients. One was based on derivatization

of

FK506 with dansyl hydrazine,

column-

switching HPLC, and chemiluminescence detection (20). Derivatization is a time-consuming procedure and is usually accompanied by a high CV, because of the derivatization reaction. Only one assay (19) differentiates between FK506 and its metabolites: in this assay, FK506 and its metabolites are separated by HPLC, and the respective fractions collected are quantified by enzyme immunoassay. This is possible because the antibody used for the enzyme immunoassay obviously is not specific for FK506 but cross-reacts with the metabolites.

However, quantification

of the metabolites

is not possi-

ble with that method because the extent to which the antibodies cross-react with the FK506 metabolites is not yet known.

In conclusion,

HPLC/MS

with a particle-beam

inter-

face allows the direct specific and sensitive quantification of FK506 and all its known metabolites within a 15-mm HPLC run, after simple solid-liquid extraction with a commercially available extraction kit. We thank F. Mandel (Hewlett-Packard, Waldbronn, FRG) for his support and fruitful discussions. This study was supported by DFG grant Pi 48/11-4, project D5, and SFB 265, project A7. References 1. Starzl TE, Todo S, Fung J, Demetris AJ, Venkataramanan R, Ashok J. FK506 for liver, kidney, and pancreas transplantation. Lancet 1989;ii:1000-4. 2. Abu-Elmagd K, Van Thiel D, Jegasothy BY, et al. FK506: a new therapeutic agent for severe recalcitrant psoriasis. Transplant Proc 1991;23:3322-4. 3. Mochizuki M, Masuda K, Sakane T, et al. A multicenter open trial of FK506 in refractory uveitis including Behcet’s disease. Transplant Proc 1991;23:3343-6. 4. Goto T, Kino T, Hatanaka H, et al. Discovery of FK506, a novel immunosuppressant isolated from Streptomyces tsukubaensi8. Transplant Proc 1987;19:4-8. 5. Tanaka H, Kuroda A, Marusawa H, et al. Structure of FK506: a novel immunosuppressant isolated from Streptomyces. J Am

Chem Soc 1987;109:5031-3. 6. Tanaka H, Kuroda A, Marusawa H, et al. Physico-chemical properties of FK506, a novel immunosuppressant isolated from Streptomyces tsukubaensis. Transplant Proc 1987;19:11-6. 7. Jones TK, Reamer RA, Desmond R, Mills SG. Chemistry of tricarbonyl hemiketals and application of Evan’s technology of the total synthesis of the immunosuppressant (-)-FK506. JAm Chem

Soc 1990;112:2998-3017. 8. Christians U, Kruse C, Kownatzki R, et al. Measurement of FK506 by HPLC and isolation and characterization of its metabolites. Transplant Proc 1991;23:940-1. 9. Au Shah I, Whiting PH, Omar G, Thomson AW, Burke MD. Effects of FK506 on human hepatic microsomal cytochromeP-450dependent drug metabolism in vitro. Transplant Proc 1991;23: 2783-5.

10. Christians U, Radeke HH, Kownatzki R, Schiebel HM, SchottCLINICAL CHEMISTRY, Voi. 38, No. 10, 1992

2031

mann R, Sewing KF. Isolation of an immunosuppresaant metabolite of FK506 generated by human microsome preparations. Clin Biochem 1991;24:271-5. 11. Venkataramanan H, Warty VS, Zemaitis MA. Biopharmaceutical aspects of FK506. Transplant Proc 1987;19:30-5. 12. Venkataramanan H, Jam A, Warty VS, et al. Pharmacokinetics ofFK506 following oral administration: a comparison of FK506 and cycloeporine. Transplant Proc 1991;23:931-3. 13. Venkataramanan H, Jam A, Warty VS, et al. Pharmacokinetics of FK506 in transplant patients. Transplant Proc 1991;23: 2736-40. 14. Tamura K, Kobayashi M, Hashimoto K. A high sensitive method to assay FK-506 levels in plasma. Transplant Proc 1987; 19:23-9. 15. CadoffE, Venkataramanan R, Krajack A. Assay of FK-506 in plasma. Transport Proc 1990;22:50-1. 16. Kobayashi M, Tamura K, Katayama N, et al. FK506 assay: past and present-characteristics of FK 506 ELISA. Transplant Proc 199123:2725-9. 17. Winkler M, Jost U, Ringe B, Gubernatis G, Wonigeit K, Pichimayr H. Association of elevated FK506 plasma levels with nepbrotoxicity in liver-grafted patients. Transplant Proc 1991;23: 3153-5. 18. Grenier F, Luczkiw J, Bergmann M, et al. An automated whole blood assay for the IMX analyzer. Transplant Proc 1991;23:

2748-9. 19. Friob MC, Hassoun A, Latinne D, Lhoest G, Otte JB, Wallemacq PE. A combined HPLC-ELISA evaluation of FK506 in transplant patients. Transplant Proc 1991;23:2750-2. 20. Takada K, Oh-Hashi M, Yoshikawa H, Muranishi S. Determination of a novel potent immunosuppressant (FK-506) in rat

2032

CLINICAL CHEMISTRY, Vol.38, No. 10, 1992

serum and lymph by high-performance liquid chromatography with chemiluminescence detection. J Chromagtogr 1990;530: 212-8. 21. Zeevi A, Venkataramanan R, Warty V, et al. In vitro assessment of FK 506 immunosuppressive activity in transplant pa-

tients. Transplant Proc 1991;23:2897-9. 22. Guengerich FP. Microsomal enzymes involved in toxicologyanalysis and separation. In: Hayes AW, ed. Principles and methods of toxicology. New York: Raven Press, 1982:609-34. 23. Huber R, Zech K. Determination of drugs and their metabolites in biological samples by fully automated HPLC with on-line solid-liquid extraction and pre-column switching. In: Zech K, Frei FW, eds. Selective sample handling and detection in high-performance liquid chromatography, part A. J Chromatogr Libr 1989;

39A:81-144. 24. Beyssens AJ, Wijnen RMH, Beuman GH, Van der Heyden J, Kootatra G, Van As H. FK506 monitoring by EIA in plasma or blood. Transplant Proc 1991;23:2745-7. 25. Japanese FK 506 Study Group. Japanese study of FK506 in kidney transplantation: benefit of monitoring of whole blood FK506 concentrations. Transplant Proc 1991;23:3085-8. 26. Machida M, Takahara 5, Ishibashi M, Hayashi M, Sekihara T, Yamanaka H. Effect of temperature and hematocrit on plasma concentration of FK 506. Transplant Proc 1991;23:2753-4. 27. Critical issues of cyclosporine monitoring report of the task force on cyclosporine monitoring. Cliii Chem 1987;33:1269-88. 28. Consensus document of the Hawk’s Cay meeting on the therapeutic drug monitoring of cyclosporine. Transplant Proc 1990;22:1357-61. 29. Jusko W, D’Ambrosio R. Monitoring FK506 concentrations in plasma and whole blood. Transplant Proc 1991;23:2732-5.