Measurement of sirolimus in whole blood using high-performance liquid chromatography with ultraviolet detection

CLINICAL

THERAPEUTICS”/VOL.

22, SUPPL. B, 2000

Measurement of Sirolimus in Whole Blood Using High-Performance Liquid Chromatography with Ultraviolet Detection David W. Halt, DSc,’ Terry Lee,2 and Atholl Johnston, Php ‘Analytical Unit, St. George5 Hospital Medical School, and ‘Clinical Pharmucology, St. Bartholomew :Yand the Royal London School of Medicine and Dent&r?/, London, England ABSTRACT Background: Sirolimus, a potent immunosuppressivedrug, exhibits intrapatient and interpatient variability of absorption and metabolism.Thus, therapeutic drug monitoring is important. Objective: This paper describesa reverse-phasehigh-performance liquid chromatography (HPLC) method, using ultraviolet (UV) absorption for detection, for measuring sirolimus levels in human whole-blood samples. Methods: The stability of sirolimus in whole blood was assessedunder conditions likely to be encountered during transport of study samplesto a central laboratory. The performance of the HPLC-UV assayin measuringsirolimus was comparedwith that of 3 established,validated HPLC assayswith tandem mass-spectrometric(MS/MS) detection. Resultsof the HPLC-UV assayalso were compared with results produced by a prototype microparticle enzyme immunoassay(MEIA). Results: Inaccuracy for 3 in-housecontrol sampleswas 54%, whereaswithin-assay repeatability (coefficient of variation [CV]) was ~5% and between-assayreproducibility was 56.6%. Mean recovery of sirolimus from blood was 8 1.5% + 4.3%. The lower limit of quantification was set at 6.5 ng/mL, and the repeatability CV at this concentration was 4.2% (n = 6). Sirolimus-containing whole-blood sampleswere stable for 3 freeze/thaw cycles when stored at -2O’C and for r2 days when stored at ambient temperature. The sampleextract was shown to be stable for up to 54 hours at ambient temperature (-22OC) after extraction. Resultsof the HPLC-UV assaywere consistentwith those of the HPLCi MS/MS assaysbut lower than those produced by MEIA. Conclusion: This HPLC-UV method is considered suitable for therapeutic drug monitoring of sirolimus. Key words: sirolimus, rapamycin, high-performance liquid chromatography (HPLC), therapeutic drug monitoring. (Clin Thu. 2000;22[SuppI B]:B38-B48)

Accepted for publication November Printed in the USA. Reproduction

B38

17, 1999. in whole or part

is not permitted.

0119-291

8.‘00/$19.00

D.W.

HOLT

ET AL.

INTRODUCTION Sirolimus* (formerly

rapamycin)

is a po-

tent immunosuppressiveagent produced

by Streptomyces hygroscopicus.’ It has a molecular massof 9 13.6 d and an uhraviolet (UV) absorption maximum at 278 nm. In Europe, sirolimus hasbeenusedas primary therapy to prevent organ rejection after kidney transplantation in concentration-controlled studies.2Therapeutic drug monitoring of sirolimus may be important becauseit exhibits both intrapatient and interpatient variability of absorption and metabolism. These differences are due in part to pharmacokinetic interactions between sirolimus and concomitantly administered inducers or inhibitors of the cytochrome P-450 (CUP) enzyme system.3 Whole blood is the matrix of choice for the measurement of sirolimus because -95% of the drug is sequesteredwithin erythrocytes.4 Ethylenediamine-tetraacetic acid (EDTA) is the preferred anticoagulant because it minimizes clotting problems.Sirolimus is metabolized by the CYP-450 enzyme system, principally by CYP3A4, the same isozyme involved in the metabolismof cyclosporine and tacrolimus. A number of sirolimus metabolites have been identified that result from demethylation, hydroxylation, or both.5 The immunosuppressivepotency of these metabolites is still under investigation. The aim of this study was to establish the following parametersfor sirolimus: (1) within- and between-assayreproducibility, linearity, and inaccuracy; (2) recovery of sirolimus from whole blood; and (3) stability of whole-blood samplescontaining *Trademark: Rapamune”’ (Wyeth-Ayerst Philadelphia, Pennsylvania).

Laboratories.

sirolimus at room temperature, after repeated freezing and thawing, and under conditions likely to be encounteredduring transport of study samplesto a central laboratory. In addition, the stability of the sampleextract was assessed over 54 hours. MATERIALS

AND METHODS

Sirolimus, desmethoxyrapamycin (internal standard), and sirolimus-free EDTA whole blood were supplied by WyethAyerst Research, Princeton, New Jersey. High-performance liquid chromatography (HPLC)-grade methanol, acetonitrile, and methyl tert-butyl ether were obtained from Rathbum ChemicalsLimited, Walkerburn, Scotland. 1-Chlorobutane was purchased from Sigma-Aldrich Co. Ltd., Poole, Dorset, England. Hexane was purchased from Merck Ltd., Lutterworth, Leicestershire, England. The reverse-phaseHPLC system consistedof a Shimadzu LC- 1OAS (Dyson Instruments,Houghton-le-Spring, Tyne and Wear, England) liquid chromatography pump set to a flow rate of 1.5 ml/minute, a Shimadzu SIL-IOA autosampler,a Shimadzu SPD-1OA UV spectrophotometric detector set at 278 nm, and a Shimadzu CR4AX chromatopacintegrator. Sirolimus was separatedusing a 25 cm x 4.6 mm Ultrasphere(Hichrome Limited, Reading, Berkshire, England)C 18-bondedsilica column (5-pm average particle size), heated to 50°C in a Shimadzu CTO-IOA column oven. The mobile phaseconsistedof acetonitrileideionized water (65135; ELGA Limited, High Wycombe, Buckinghamshire, England). The assaywas calibrated using 7 nonzero calibrators nominally covering the range 5 to 350 ng/mL (actual concentrations 6.5, 13.0, 25.9, 51.8, 103.7, 207.4, B39

CLINICAL THERAPEUTICS’

and 356.4 ng/mL) and a sirolimus-free calibrator. The highest calibrator and the lowest non-zero calibrator were run in duplicate. Calibration data were fitted using a ‘/x2 curve fit to generate a calibration curve. Three quality control samples were prepared in-house in the sirolimus-free whole blood, using an independent methanolic stock solution of sirolimus. These samples had nominal sirolimus concentrations of 15, 75, and 225 ng/mL. The calibrators, control samples or patient samples (I mL), and internal standard solution (100 p,L) were pipetted into 8.0-mL polypropylene tubes. The extraction solvent (4 mL methyl tert-butyl ether: I-chloroethane:methanol, 45:45: IO) was added and the tubes capped. The contents of the tubes were gently mixed for 20 minutes at 100 rpm. The tubes were then centrifuged for IO minutes at 15OOg, following which all the solvent layer was transferred to a IO-mL conical polypropylene tube and evaporated to dryness in a

Savant SC200 SpeedVac@ concentrator (Thermoquest, Basingstoke, Hampshire, England) maintained at 60°C. The sample extracts, which had been stored at room temperature, were reconstituted with 0.5 mL acetonitrile/water (50:50) and 0.5 mL hexane. The tubes were then capped and the contents gently mixed for 10 minutes. After centrifugation, the top hexane layer was removed by vacuum aspiration and discarded. The sample extracts were left uncapped, in the dark, for -15 minutes to allow any residual hexane to evaporate. The sample extract was then transferred to an autosampler vial and 200 pL injected onto the analytical column for analysis. Typically, the retention time of sirolimus and the internal standard were -22 and 28 minutes, respectively. Figure I illustrates a typical chromatogram of a sample extract. Recovery of sirolimus and the internal standard from whole blood were assessed by comparing the peak areas of chro-

I

I

I

I

I

1

5

10

15

20

25

30

Time

(min)

Figure I. Chromatogram of an extract from a blood sample collected from a patient receiving sirolimus after kidney transplantation. The concentration of sirolimus in the sample was 14.4 ng/mL. B40

D.W.

HOLT

ET AL.

matograms from extracts of the 3 in-house control samples with those from solutions of the 2 analytes in 50% v/v acetonitrile and deionized water, injected directly onto the column. The nominal concentration of the internal standard was 0.125 pg/mL. The stability of the sample extracts in the autosampler tray was tested by extracting 4 aliquots of the 3 in-house control samples and pooling the sample extracts. The combined extracts were stored at ambient temperature (-22OC) and injected onto the analytical column 5 times over 54 hours. The result for each extract was calculated against a single set of calibrators extracted at the same time as the control samples and injected at the start of the experiment. Linearity was tested by assaying a set of calibrators and then assaying each calibrator 5 times and calculating the measured concentration of each calibrator from the calibration curve. A plot of percentage deviation from the expected concentration against the expected concentration was drawn and inspected for trends. Inaccuracy and imprecision were assessed using the 3 in-house control samples. Each sample was assayed as 5 replicates in a single batch for within-batch determination (repeatability) and as 5 replicates on 3 separate days to determine the between-batch accuracy and precision (reproducibility). The criteria for the acceptance of the control samples were that the mean measured values should be within +15% of the expected value for the 2 highest concentrations and *20% for the lowest concentration. Stability of sirolimus in whole blood was assessed by 5 replicate analyses of each of the 3 in-house control samples. The samples were initially analyzed within 2 hours of preparation to obtain reference

sirolimus concentrations. The samples were then used to assess the stability of sirolimus during storage at both ambient temperature and under refrigeration (-4’C) after -24 and 48 hours, respectively. Aliquots of the 3 samples also were stored deep frozen (- -20°C), then thawed at ambient temperature and reassayed. The 3 aliquots were refrozen and thawed twice more to provide data from 3 freeze/thaw cycles. In addition, the stability of the control samples was examined under conditions likely to be encountered during transport of study samples to a central laboratory. For this, aliquots of the control samples were extracted and sirolimus measured on the day of preparation. An aliquot of each control sample was then deep-frozen at --20°C. The remaining control samples were stored at -4OC. After -2 hours the frozen samples were dispatched by courier from London, England, to Paris, France, using packing material later used for the transport of study samples. When the samples arrived in Paris they were cold but not frozen. The samples were then returned to London by courier, where they were stored for 24 hours at 4OC and then analyzed along with the remaining quality control samples that had also been stored at 4OC. The total time from initial dispatch to sample analysis was 4 days. Five aliquots of each sample at each concentration were analyzed. Two method comparisons were performed. First, the performance of the HPLC-UV assay was compared with that of 3 established, validated HPLC assays with tandem mass-spectrometric (HPLCI MS/MS) detection performed in commercial laboratories in the United States. Sirolimus was measured in 55 blinded samples (11 sets of 5) in random order: a sirolimusB41

CLINICAL THERAPEUTICS’

Table 1. Recovery of sirolimus mean f SD.

and the internal standard from whole blood. Values are % Recovery

Sirolimus ConcentraGon (ng/mL)

No. of Samples

IS.1

5 5 5

75.6 226.8

81.9 + 3.6 82.5 t 1.6 80.3 IIT7.6 81.5 k4.3

Mean

free sample, 5 pooled samples from subjects receiving sirolimus after kidney transplantation, and 5 samples spiked with the drug to known concentrations. These samples were treated as patient samples and analyzed over a total of 6 assay runs, during which I77 other samples from patients receiving sirolimus also were analyzed. Only after analysis of the samples were the identities of the 55 samples revealed. Second, the results of the HPLC-UV assay were compared with those produced by a prototype microparticle enzyme immunoassay* (MEIA).6 A total of 135 samples from 35 renal transplant patients treated with sirolimus were reanalyzed by MEIA using zero calibrator mode I calibration. The samples had been stored at --20°C for 56 months before they were reanalyzed by MEIA. RESULTS The actual concentrations at which recovery of sirolimus was assessed were 15.1, 75.6, and 226.8 ng/mL. The mean recovery of sirolimus from whole blood was *Trademark: Park, Illinois).

B42

IMx”

(Abbott

Laboratories,

Sirolimus

Abbott

Internal Standard 60.4 64.3 63.3 62.7

+ 2.8 k 2. I k 5.8 + 3.6

independent of concentration and was 81.5% + 4.3%, whereas that of the internal standard was 62.7% + 3.6% (Table I). The sample extract was stable over 54 hours, as judged by a plot of measured concentration over time (data not shown). The percentage deviation from expected value for the measured concentrations of the calibrators assayed to test linearity ranged from -8.7% to +9.2%. There was no systematic deviation from the expected value (data not shown). Using the criteria noted above, inaccuracy for the measurement of sirolimus in the control samples was well within the range allowed. The mean percentage deviation from the expected value was 4.0%, 2.1%, and 1. I % from the lowest to the highest control sample, respectively. The lower limit of quantification (LLOQ) was set at 6.5 ng/mL, the value of the lowest calibrator. The repeatability (coefficient of variation [CV]) at this concentration was 4.2%. and the percentage deviation from the expected value was 9.2% (n = 6). The upper limit of quantification was set at 356.4 ng/mL, the value of the highest calibrator. The repeatability at this concentration was 2.5%, and the percentage deviation from the expected value was -9.7% (n = 6).

D.W. HOLT ET AL.

Table II. Within- and between-assayvariability. Values are mean +-SD. SirolimusConcentration(ng/mL) No. of Samples Control I Assay 1 2 3

5 5 5

Meanwithin-assay cv (%I)

Meanbetween-assay

3

cv (70)

Control2

14.9 f 0.8 16.1 + 0.5 16.1 c 0.4 15.7 + 0.6 3.7 IS.2 + 1.0 6.6

77.9 78.4 75.4 77.2

f 2.5 + 5.4 t 3.8 + 3.9 5.0 72.5 rt 0.3 0.4

Control 3 231.8 237.1 218.7 229.2

f 11.6 f 4.7 + 4.9 + 7. I 3.1 236.4 k 15.6 6.6

CV = coefficient of variation

Table III. Mean concentrationsof sirolimusin samplestransportedby courier at subambient temperatures(N = 5 at each point). SirolimusConcentration(ng/mL) Control I

Control2

Control 3 207.9 218.6 208.4 226.8

initial value Transitsamples

14.5 15.0

69.5

4 Days at 4°C

IS.3

Targetvalue

IS.1

67.7 75.6

Mean within-assay repeatability (CV) was s5.0%, and between-assayreproducibility was ~6.6% acrossthe range of the 3 in-house control samples(Table 11). There was no trend toward a loss of sirolimus during the proceduresdesigned to test the stability of sirolimus in whole blood (data not shown). Nor was there any trend toward a lossof the drug during sample transit at subambient temperatures. Table III summarizesthe meandata for samplesshipped between London and Paris and sampleskept in a refrigerator. The results of the comparisonbetween the HPLC-UV and HPLCiMSiMS assays

71.2

using 55 blinded samplessuggestedthat a small mean positive inaccuracy existed in the resultsproduced by the HPLC-UV assay (+1.5-+5.7 ng/mL) compared with the resultsproduced by the HPLCIMSIMS assays.The bias varied when the results were compared with individual laboratories using HPLCIMSIMS, becausethe 3 laboratories using HPLCIMSIMS did, themselves,differ in terms of inaccuracy. The positive bias shown by HPLC-UV was consistent across the range of concentrations,asshown in Figure 2, in which the difference between the results using HPLC-UV and the mean data using B43

CLINICAL THERAPEUTICS’

62 E m

5-

54iG .6 3> ?2Y %'

0

0.3

5

40

65

80

Mean Sirolimus Concentration by HPLC/MS/MS (ng/mL) Figure 2. Bias (systematic error) of the high-performance liquid chromatographyultraviolet (HPLC-UV) assay compared with the mean value obtained by the HPLC-tandem mass spectrometry (HPLCIMSIMS) assays for 5 blood samples spiked with cyclosporine and a sirolimus-free blood sample.

1

3

12

21

33

Mean Sirolimus Concentration by HPLC/MS/MS (ng/mL) Figure 3. Bias (systematic error) of the high-performance liquid chromatographyultraviolet (HPLC-UV) assay compared with the mean value obtained by the HPLC-tandem mass spectrometry (HPLCIMSIMS) assays for 5 pooled blood samples from patients receiving sirolimus.

HPLCiMSiMS are plotted for each of the spiked samples. Because the inaccuracy was at or below the limit of detection set for the HPLC-UV assay, it was not always evident when the sirolimus-free samples were assayed. In the 5 sirolimusfree samples, only 2 had measurable peaks, whereas at a spiked concentration B44

of 0.3 ng/mL, 3 of 5 samples gave positive results. However, the inaccuracy was similar for both the spiked samples and the pooled patient samples (Figure 3), from which it can be seen that the concentrations of sirolimus in 2 of the pooled patient samples were also below the LLOQ for the HPLC-UV assay. It was

D.W.

HOLT

ET AL.

concluded that, within the measurable concentration range of the HPLC-UV assay, agreement between the 2 techniques was acceptable, because the sirolimus concentrations for 22 (73.3%) of the 30 samples above the LLOQ for the HPLC-UV assay were within 20% of the mean value obtained by HPLCIMSIMS. The source of the interference was never identified and may have been specific to this batch of samples. It is worth noting that the in-house check on inaccuracy at a nominal concentration of 15 ng/mL, noted above, showed that it was ~5% and that batch-to-batch in-house controls at this concentration were almost always (95%) within 20% of the target value over the entire period the assay was in use. The results for the comparison of sirolimus measurement by HPLC-UV assay and MEIA are shown in Figures 4 and 5. Of the 135 samples, 2 (1.5%) were excluded from the comparison because the sirolimus concentrations measured by HPLC-UV were below the LLOQ of the assay (2.3 and 9.1 ng/mL by ME1 A). The median (interquartile range) sirolimus concentrations for the HPLC-UV assay and MEIA were 18.2 ng/mL (I 3.4-22.9) and 20.1 ng/mL (14.2-28.9), respectively. The mean percentage difference between the 2 assays was 17%. The equation for the regression line was MEIA = 1.27 HPLC - 1.6 (r* = .81) (Figure 4). No trend for a concentration-dependent difference was observed between the 2 assays, as shown by a Bland and Altman plot7 (Figure 5). Assuming that the sample extracts were chromatographed overnight, it was possible to run 20 to 25 samples per day, with appropriate calibrators and controls. Sample extract evaporation in the SpeedVac’“’ concentrator for such an assay took
DISCUSSION Sirolimus was recently licensed in the United States to prevent organ rejection in patients receiving renal transplants, with the recommendation that it be used with cyclosporine and corticosteroids. Although not a requirement of the license, measurement of sirolimus was recommended for pediatric patients and patients with hepatic impairment. Measurements are also recommended when potent inducers or inhibitors of the CYP3A4 isozyme are coadministered, or if cyclosporine dosing is markedly reduced or discontinued. To implement routine therapeutic drug monitoring, an analytical method should be validated over a clinically relevant concentration range; the method should use apparatus that is readily available in a broad range of laboratories, and it should be relatively simple to perform. The HPLC-UV assay described here addresses some of these needs. To date, >3000 measurements of sirolimus have been made by this laboratory using this assay. The samples generally were collected just before dosing (trough samples), but many samples were from full pharmacokinetic profiles used to estimate dosing parameters or to study drug interactions with sirolimus.* These profiles necessitated the use of a broad calibration range to minimize the need for sample dilutions. Using a sample volume of only 1 mL and a simple HPLC-UV assay, sirolimus concentrations were measured in 3 control samples, spanning the concentration range 15.1 to 226.8 ng/mL, to within 5% of the target values. Within- and between-assay imprecision for these samples was acceptable, as was the recovery of sirolimus from EDTA-anticoagulated whole blood. Securing stocks of EDTA-anticoagulated whole B45

CLINICAL

2

80

E Pn a

ijj 2

1 , ,’

60-

,’

THERAPEUTICS’

,

’ Line of Identity

0

20

Sirolimus

40

Concentration

60

by HPLC-UV

80

(ng/mL)

Figure 4. Comparisonof sirolimus concentrations in I33 blood samplesfrom renal transplant patients, as measuredby the high-performance liquid chromatographyultraviolet (HPLC-UV) assay compared with the microparticle enzyme immunoassay(MEIA) and the line of identity. 140 > ? 0 2

120loo-

=

60-

5

40-

2= E

20-

0 0 0

80-

Ok & -20 z= 5 -40 8

-60

-

0

-80 0

I 20

Sirolimus by HPLC-UV

I 40

I 60

I 80

Concentration and MEIA (ng/mL)

Figure 5. Comparisonof sirolimus concentrationsin 133 blood samplesfrom renal transplant patients, as measuredby the high-performance liquid chromatographyultraviolet (HPLC-UV) assay and the microparticle enzyme immunoassay (MEIA). The percentagedifference between the MEIA and the HPLC-UV assay is plotted against the meanvalue for the 2 assays.The horizontal line is the mean percentagedifference from HPLC-UV. B46

D. W.

HOLT

ET AL.

blood for the preparation of sirolimus calibrators and control samples is not always easy. We attempted to validate this assay using titrated human whole blood obtained from a local blood transfusion service, but the results were disappointing, due to a peak coeluting with sirolimus, presumably a plasticizer or buffer constituent (data not shown). The methods for 2 other published HPLC-UV assays were based on a I-mL sample volume. ‘v9 Both achieved superior sensitivity by means of a more extensive and time-consuming sample clean-up procedure. Subsequently, additional improvements in terms of sensitivity were made to one of these methods.‘” Interfering peaks, from endogenous or exogenous sources, are common when UV detection is used. This assay was subject to interference from unidentified peaks running close to sirolimus in some samples; for this reason, the retention time for the drug was adjusted within a range of -1.5 to 22 minutes by manipulating the proportion of acetonitrile in the mobile phase. These adjustments also took into account batch-to-batch differences in column packing material. From the practical standpoint of sample transport to the laboratory and sample storage before analysis, our data demonstrated that sirolimus blood samples can be stored for 22 days at ambient temperature (-22’C, protected from light) and can withstand 3 freeze/thaw cycles without incurring any loss of drug. Others have shown sirolimus in whole blood to be unstable at higher temperatures (>30°C),” so we investigated the transport of samples in subambient packaging. This transport was feasible but costly, because the freezer packs make the package larger and heavier.

As would be expected, comparison of the HPLC-UV assay with HPLCIMSIMS assays showed that HPLCiMSiMS is more sensitive and selective. Despite calibration differences between the assays, the agreement between the 2 techniques was acceptable; clearly, however, values below the LLOQ for the HPLC-UV assay were unusable and should not be reported. It seems likely that the bias noted for the HLPC-UV assay was caused by an inherent defect in the detection system, leading to a small peak being identified erroneously as sirolimus. Whether there were any problems caused by the sample matrix was not established, but it is unlikely that the bias was caused by metabolites of sirolimus, since the same bias was seen in spiked samples and in samples from subjects receiving the drug. At the time this assay was validated, pure samples of sirolimus metabolites were not available for chromatography on this system. However, the agreement between this UV-based assay and the results from tandem MS assays suggests that, within the quantifiable range of the assay, interference from metabolites was not significant. When compared with MEIA, the chromatographic technique gave an overall lower result. This, too, was anticipated because MEIA cross-reacts with some sirolimus metabolites.” The percentage difference in the reported values is similar to those reported by a center that compared MEIA with an HPLCIMSIMS assay for the measurement of sirolimus in samples from kidney transplant patients (D. Hicks, Wyeth-Ayerst Research, personal communication, 1999). CONCLUSIONS HPLC-UV is a feasible assay to use in a laboratory contemplating therapeutic drug B47

CLINICALTHERAPEUTICS”

monitoring of sirolimus for patients receiving the drug after kidney transplantation. A throughput of -20 samples per day, plus calibrators and controls, is possible. As with any chromatographic assay based on the use of UV detection, care must be taken to exclude the possibility of interference from coeluting analytes.

ACKNOWLEDGMENTS This study was performed with the financial support of Wyeth-Ayerst European Clinical Research, Paris, France, to establish a resource for monitoring their multicenter clinical studies of sirolimus. The samples for comparison between the HPLC-UV and HPLCIMSIMS assays were supplied by Wyeth-Ayerst Research, Princeton, New Jersey.

Address correspondence to: David

W. Holt, DSc, Analytical Unit, St. George’s Hospital Medical School, London SW 17 ORE, UK.

REFERENCES I. Sehgal SN. Rapamune (sirolimus, ram pamycin): An overview and mechanism of action. T/w Drrq Monit. 1995; 17:660-665. 2. Groth CC, Backman L, Morales JM, et al. Sirolimus (rapamycin).based therapy in human renal transplantation: Similar efficacy and different toxicity compared with cyclosporine. Sirolimus European Renal Transplant Study Group. Trcrncpltrr7t~/tior,. 1999;67:

1036-I

042.

3. Trepanier DJ, Gallant H, LeGatt DF, Yatscoff RW. Rapamycin: Distribution, pharmacokinetics and therapeutic range

B48

investigations: An update. Cli/l Bioclzrriz. 19983 I :345-X I. 4. Yatscoff RW. Boeckx R. Holt DW, et al. Consensus guidelines for therapeutic drug monitoring of rapamycin: Report of the consensus panel. T/w Drrq Mmit. 1995: 17:676-680.

5. Streit F, Christians II, Schiebel HM, et al. Structural identification of three metabolites and a degradation product of the macrolide immunosuppressant sirolimus (rapamycin) by electrospray-MS/MS after incubation with human liver microsomes. DruR Metub Dispos. I996;24: 1272-l 278. 6. Jones K, Saadat-Lajevardi S, Lee T, et al. An immunoassay for the measurement of sirolimus. Cli/l T/zer. 2000;22(Suppl B): B49-B6 I. 7. Hollis S. Analysis of method comparison studies. A1717 C/i77 BiocV7e777. 1996;33: l-4. 8. Svensson JO, Brattstrom C, Sawe J. Determination of rapamycin in whole blood by HPLC. Thrr Drug Mo77if. l997:19: 1 12-l 16. 9. Napoli KL. Kahan BD. Sample clean-up and high-performance liquid chromatographic techniques for measurement of whole blood rapamycin concentrations. J Clzrwuotogr B Bionwd Appl. I994;654: 111-120. IO. Napoli KL, Kahan BD. Routine clinical monitoring of sirolimus (rapamycin) whole~blood concenlrations by HPLC with ultraviolet detection. C/i77 Cl7r777. 1996:42:1943-1948. II

Fen-on GM, Jusko WJ. Species differences in sirolimus stability in humans, rabbits, and rats. Drq Metrth Dispos. 1998;26:83-84.