determination of sirolimus in rabbit arteries using liquid chromatography separation and tandem mass...

Copyright © 2007 John Wiley & Sons, Ltd. Biomed. Chromatogr. 21: 1036–1044 (2007)DOI: 10.1002/bmc

1036 J. Zhang et al.ORIGINAL RESEARCH ORIGINAL RESEARCH

Copyright © 2007 John Wiley & Sons, Ltd.

BIOMEDICAL CHROMATOGRAPHYBiomed. Chromatogr. 21: 1036–1044 (2007)Published online 22 June 2007 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/bmc.849

Determination of sirolimus in rabbit arteries using liquidchromatography separation and tandem massspectrometric detection

Jun Zhang,* Ramona Rodila, Pamela Watson, Qin Ji and Tawakol A. El-Shourbagy

Department of Drug Analysis, Abbott Laboratories, Abbott Park, IL 60064, USA

Received 14 February 2007; accepted 15 March 2007

ABSTRACT: Sirolimus, an effective immunosuppressive agent, is used for drug eluting stents. During stent development, an ana-lytical method for the determination of sirolimus in tissue needs to be established. Normally, tissue samples are homogenized andthen analyzed against the calibration standards prepared in a tissue homogenate. This approach provides insufficient control of thehomogenization process. In this paper, tissue quality control samples were introduced for the optimization of the homogeniza-tion process during method development, but also allowance for the performance evaluation of the entire analytical process. Inaddition, a new approach using rabbit blood as a homogenization medium was developed to stabilize sirolimus in rabbit tissuehomogenates. Calibration standards and quality controls were prepared by spiking different sirolimus working solutions into rab-bit blood. Homogenization quality control samples were prepared by injecting other sirolimus working solutions into empty testtubes and pre-cut arteries within pre-defined masses. A high-throughput homogenization procedure was optimized based on thespecific chemical properties of sirolimus. The linear dynamic range was between 49.9 pg/mL and 31.9 ng/mL to accommodate theexpected artery homogenate concentrations. Additionally, quality controls in rabbit blood were also used in the extraction to sup-port the calibration standards. The accuracy and precision of the quality controls in rabbit blood reflect the extraction perform-ance and the accuracy and precision of the homogenization tissue quality controls reflect the overall performance of the method.The mean bias was between −4.5 and 0.2% for all levels of quality controls in the blood and between 4.8 and 14.9% for all levelsof the homogenization tissue quality controls. The CVs of all concentration levels were ≤5.3% for the quality controls in bloodand ≤9.2% for the homogenization tissue quality controls. The method was successfully applied to determine the concentration ofsirolimus in the rabbit arteries. Copyright © 2007 John Wiley & Sons, Ltd.

KEYWORDS: sirolimus; rapamycin; determination; tissue; LC-MS/MS

*Correspondence to: Jun Zhang, Department of Drug Analysis,Abbott Laboratories, Abbott Park, IL 60064, USA.E-mail: [email protected]

Abbreviations used: FKBP, FK-binding proteins; QC, quality control;RBC, red blood cells; SRM, selective reaction monitoring.

INTRODUCTION

Sirolimus, also called rapamycin, is a biologically activelipophilic macrolide lactone, as shown in Fig. 1(a).Derived from a fungus Streptomyces hygroscopicus, siro-limus was first isolated in a soil sample from EasterIsland (Sehgal et al., 1975). Sirolimus was found to haveanticancer, antifungal, and immunosuppressive activi-ties (Eng et al., 1984; Calne et al., 1989; Kimball et al.,1991). The activity as an immunosuppressant estab-lished this compound as a drug to prevent rejection oforgan transplants. The binding of sirolimus to specificcytosolic proteins called immunophilins is credited forits immunosuppressive activity. Sirolimus blocks transi-tion from G1 to S phase in the cell cycle, by inhibitingthe activity of the protein mTor. mTOR is a key regu-

latory kinase and its inhibition leads to suppression ofcytokine-driven T-cell proliferation, inhibiting the trans-lation of a family of mRNA and inhibiting IL-2 inducedtranscription of the proliferating cell nuclear antigen,which is critical for DNA replication. Sirolimus alsoblocks CD28-mediated sustained upregulation of IL-2transcription in T cells, and inhibits the kinase activityof complexes related to cell cycle progression (Serruyset al., 2002). Further experiments showed that sirolimustreatment reduced the proliferation response aftercoronary angioplasty significantly in a well-characterizedswine model (Gallo et al., 1999). Owing to its anti-proliferative and immunosuppressive properties, sirolimuswas selected as a drug candidate to prevent restenosis.

To be effective as an anti-restenotic agent, the clini-cal drug must reach sufficient concentrations in injuredarteries. In contrast to systemic treatment, local drugadministration provides advantages because the drugcan be applied to the artery at the precise location andat the time of artery injury. Local drug delivery andslow drug release can yield higher drug concentrationsin the tissue and a minimal systemic exposure that may


Determination of sirolimus in rabbit arteries 1037ORIGINAL RESEARCH

the drug in the tissue. Moreover, because sirolimus is ahydrophobic agent, sirolimus may be accumulated inthe lipids next to the artery. Converting a sample fromsolid tissue to liquid format prior to analysis is a chal-lenging task.

A high throughput method for the determinationof sirolimus in rabbit arteries has been developed.An optimized homogenization procedure with a multi-channel homogenizer was used to convert the samplesfrom solid tissue into liquid format. Critical steps wereimplemented to avoid drug degradation, to assure ad-equate recovery and to eliminate cross-contaminationand carryover. The sequential liquid–liquid extractionwas performed using a multi-channel liquid handler.The extracts were transferred, dried down, reconsti-tuted and injected into a liquid chromatograph tandemmass spectrometry system (LC-MS/MS). The samplevolume was 250 µL of tissue homogenate generatedfrom the homogenization of up to 130 mg of arterytissue in 2.0 mL of rabbit blood. The lower limit ofquantitation (LLOQ) was 49.9 pg/mL. The lineardynamic range was between 49.9 pg/mL and 31.9 ng/mL. Three consecutive development batches wereperformed, and the achieved r2 was between 0.9994 and0.9998. The CVs for all the concentration levels ofhomogenization quality controls were ≤9.2%. A totalof 106 of 108 homogenization quality control sampleswere within a bias of 25%. This number reflects all thevariations of the entire analytical process. The methodwas accurate, precise, sensitive, selective and robust.This method was successfully used for sirolimus ana-lysis in rabbit arteries.

EXPERIMENTAL

Chemicals and reagents. HPLC-grade methanol, acetonitrile,ethyl acetate and hexanes were obtained from EMD Science(Gibbstown, NJ, USA). ACS-grade ammonium acetate,formic acid (88%) and 20% potassium hydroxide (KOH)were purchased from Sigma Aldrich (St Louis, MO, USA).Sirolimus was purchased from Sigma-Aldrich. Zotarolimus,as an internal standard as shown in Fig. 1(b), was synthesizedby Abbott Laboratories (North Chicago, IL, USA). Distilledwater was further purified using a Milli-Q water purificationsystem from Millipore (Billerica, MA, USA). Rabbit bloodwith potassium–EDTA as an anticoagulant and rabbit arterytissues were purchased from Lampire Biological Laboratories(Pipersville, PA, USA).

Instruments. Positive displacement pipettes were pur-chased from Gilson (Middleton, WI, USA). A multi-channelhomogenizer (also called autogizer) was purchased fromTomtec (Hamden, CT, USA). A semi-automatic multi-channel pipette MicroLab AT 2 Plus was purchased fromHamilton (Reno, NV, USA). Multi-channel hand-held elec-tronic pipettes were purchased from BioHit (Helsinki,Finland) and 96-well liquid–liquid extraction plates were

Figure 1. (a) Structure of sirolimus and (b) structure of inter-nal standard.

reduce systemic toxicity. Approved sirolimus-coatedstents contain up to 180 µg for a seven-cell design.The coating materials contain 30% sirolimus in a 1:1mixture of polymers polyethylenevinylacetate and poly-butylmethacrylate by weight. Based on the design ofsirolimus-coated stents, the drug release into the circu-latory system and the surrounding tissue should bemonitored in the stent development. The concentrationof sirolimus in the circulation varies with time afterplacement. Many means can be used to determine thesirolimus concentration in whole blood (Volosov et al.,2001; Napoli and Taylor, 2001; Streit et al., 2002;Vogeser et al., 2002; Pieri et al., 2005; Morris et al.,2006). However, sirolimus concentrations in the tissuevary significantly with the distance from the stents andwith time. For example, the arterial tissue immediatelyadjacent to the stents is expected to carry much moresirolimus in unit mass of tissue than the artery fartherremoved from the stents. The concentration determina-tion from the surrounding artery or even myocardiumcan offer critical information about the toxicokinetics of


1038 J. Zhang et al.ORIGINAL RESEARCH

purchased from Marsh Bio Products (Rochester, NY, USA).An HPLC system included an LC-10AD HPLC pump, anSIL-10A XL autosampler and an SCL-10A system controllerpurchased from Shimadzu (Kyoto, Japan). A column-heatingpocket was purchased from Keystone Scientific (Bellefonte,PA, USA). Two flow-switching valves were purchased fromVici Valco Instruments (Houston, TX, USA). An 1100 seriesHPLC pump and a degasser system were purchased fromAgilent Technologies (Palo Alto, CA, USA). A plate sealerand heat seal films were purchased from Abgene (Rochester,NY, USA). An API-4000 mass spectrometer and computercontrol system Analyst Version 1.3.2 were from AppliedBiosystems (Foster City, CA, USA).

Preparation of standards and quality controls. Two separ-ate stock solutions were prepared in 50:50 (v/v) acetonitrile–water. One solution was used to prepare calibration standardsin rabbit blood. The other solution was used to prepare qual-ity controls (QC) in both rabbit blood and rabbit artery tissueQC samples. The calibration standards and QCs were pre-pared by spiking diluted working solutions into rabbit blood.A total of nine calibration standards and four QC levels wereprepared. The calibration standard concentration range wasbetween 49.9 pg/mL and 31.9 ng/mL. The QC concentrationswere 79.9 pg/mL, 99.9 pg/mL, 27.9 ng/mL and 29.9 ng/mL. Allstandards and QCs were stored at approximately −70°C. Thetissue QCs were prepared by injecting the diluted stocksolution into pre-cut rabbit artery tissue of a certain mass.Two tissue QC concentration levels (low tissue QC and hightissue QC) were prepared. The low tissue QC concentra-tion was approximately 125 pg/mL after homogenization in2.0 mL of rabbit blood. The high tissue QC concentrationwas approximately 24.7 ng/mL after homogenization in2.0 mL of rabbit blood. Three tissue mass levels were usedfor both low and high tissue QCs. The low tissue mass wasbetween 5 and 15 mg. The high tissue mass was between 70and 150 mg. After sirolimus solution was injected with a sharpneedle syringe into the tissue in 17 × 100 mm polypropylenetubes by Becton Dickson (BD; Franklin Lakes, NJ, USA),the test tubes were capped and stored in a freezer at approxi-mately −70°C.

High-throughput sample homogenization. Tissue sampleswere removed from the freezer at approximately −70°C,thawed at room temperature in the Becton Dickson test tubeswith light protection and 2.0 mL of rabbit blood was addedinto each tube. All test tubes were then loaded onto a pre-cooled (approximately 10°C) test tube rack of the Tomtecautogizer. The test tube rack temperature was also main-tained at approximately 10°C during the homogenizationprocess. Total homogenization time was set to 1.0 min. Inorder to release the heat generated due to high speed spin-ning of the homogenization probes, a pause of 0.2 min wasallowed after the first 0.5 min of homogenization at 18,000–24,000 rpm. The homogenization was then resumed foranother 0.5 min at the same speed. A critical cleaning/washwas performed to eliminate carryover. After one row ofsamples was homogenized, the probes were raised out of thesolution, but held in the tubes to allow the liquid to drip for0.1 min. The probes were then cleared four times beforebeing washed to release the liquid the homogenization probes

carried. The homogenizer probes were first washed in anopposite flowing water bath for 0.2 min to remove most bloodand aqueous components, and then moved into an oppositeflowing bath of 20% potassium hydroxide aqueous solutionfor 0.2 min to break the sirolimus and to clean the lipidscarried on the homogenization probes. Then, the homogeniza-tion probes were returned to the opposite flowing water bathfor 0.5 min to remove KOH solution remains on the probes,and finally the homogenization probes were washed in aslowly flowing acetonitrile bath for 0.2 min followed by a driptime of 1.0 min to remove all organic solvent on the probes.The cutters were cleared four times and were then ready forthe next sample row. Following homogenization, the sampletubes were capped and wrapped with aluminum foil andstored in a freezer at approximately −70°C for extraction.

High-throughput liquid–liquid extraction. Sirolimus andzotarolimus were extracted from blood or tissue homogenatesin blood using a semi-automated 96-well liquid–liquid tech-nique. The procedure can be summarized as follows: thesamples were taken out of the storage and completely thawedat room temperature with light protection. A 100 µL aliquotof zotarolimus internal standard solution was added, 250 µLof the samples was transferred from a load rack; 100 µL of4:1 (v/v) methanol:100 mM ammonium acetate solution wasadded for blood cell lysate. Solutions and samples were mixedby aspirating and dispensing 300 µL in five times. A 1.0 mLaliquot of 1:1 (v/v) ethyl acetate:hexanes mixture was added.The 96-well plate was sealed with a polypropylene–aluminumseal film and allowed to sit at room temperature for approxi-mately 5 min with light protection. The 96-well plate wasshaken using a multi-tube vortexer for approximately 5 minand then allowed to sit for approximately 30 min at roomtemperature with light protection. The 96-well plate wasshaken using the multi-tube vortexer for another 5 min.The 96-well plate was then centrifuged at approximately4000 rpm for approximately 10 min at approximately 4°C.The polypropylene–aluminum film was punctured to performtransfer of 700 µL organic layer to a new 96-well plate. Theorganic extract was dried down with a stream of nitrogen atroom temperature. A 50 µL aliquot of mobile phase and50 µL of water were used to reconstitute the plate. The 96-well plate was capped and shaken for approximately 5 min toensure mixing of the solutions. A 40 µL aliquot of the re-constituted solution was consecutively injected into theLC-MS/MS for sample analysis.

Chromatography. A mobile phase including 25 mM ammo-nium acetate and 0.03% (v/v) formic acid in 80% (v/v)methanol–water was used to elute sirolimus and zotarolimusthrough a Pharma C18-B, 120A, 5 µm, 2 × 100 mm columnfrom BHK (Naperville, IL, USA). An ODS-P, 120A, 5 µm, 2× 10 mm guard column also from BHK was used. The columntemperature was set to 50°C. A 2.0 mL aliquot of mobilephase was used as a solvent for the injection needle rinse.The HPLC run time was approximately 6.1 min at a flow rateof 0.4 mL/min. A column-switching technique was used toreduce hydrophilic matrix effect components entering intoanalytical column and to divert hydrophilic components towaste before entering into the analytical column. The setuphas been reported in our previous publication (Zhang et al.,



2006a). In this experiment, the liquid chromatograph effluentwas initially diverted to waste in default. At 1.2 min, the filterand guard column was switched off-line and a backwashsolvent that was composed of 95% (v/v) acetonitrile–waterwas delivered through the guard column and filter at a rateof 2.0 mL/min opposite to the direction of the mobile phase.At 2.25 min, the LC effluent was diverted to the massspectrometer to start the data acquisition, the data acquisitioncontinued for 3.0 min. At 2.7 min, the mobile phase waspumped through the guard column and filter for recondition-ing. At 5.25 min, the LC effluent was diverted to waste afterdata acquisition was finished. The guard column and filter wasswitched back inline before the run was completed.

Tandem mass spectrometric detection. The liquid chromato-graph flow was monitored by an API-4000 mass spectrometerwith a turbo ion spray interface operated at positive ioniza-tion mode. The spray needle voltage was 5500 V and thesource temperature was 400°C. The curtain gas setting was10, the gas 1 setting was 60, the gas 2 setting was 55, thedeclustering potential was 60 V and the entrance potentialwas 10 V. The collision energy was 16.8 eV for sirolimus and18.0 eV for zotarolimus respectively. The selective reactionmonitoring (SRM) channels were m/z 931 → 883 forsirolimus and m/z 983 → 935 for zotarolimus. A chromato-gram of reference solution was shown in Figure 2.

Data processing, concentration calculation, and statisticsanalysis. All acquired data was processed by Analyst version1.3.2. A single set of parameters was used for the peak inte-gration. The results data file including information such assample identification, peak area of analyte and peak areaof the internal standard was stored into a LaboratoryInformation Management System (LIMS) Watson version6.2 from Thermo Finnigan (Waypne, PA, USA). The calibra-tion curve was generated by linear least squares regressionwith a weighting factor of 1/(x · x). The regression was per-formed based on the peak area ratio of sirolimus/zotarolimusagainst the nominal concentration of sirolimus. The measuredconcentration of sirolimus was calculated based on the peakarea ratio of sirolimus/zotarolimus. All concentration calcula-tion and statistics analysis were performed using Watson version6.2.

Figure 2. A chromatogram of reference solution containing50 ng/mL of both sirolimus and zotarolimus.

RESULTS AND DISCUSSION

High-throughput homogenization

Because sirolimus is present in the tissue, the sampleformat must be converted to a liquid to allow forsample analysis. Homogenization is an efficient meansfor that conversion. As reported previously, sirolimusis highly distributed into red blood cells (RBC)(Yatscoff et al., 1993) in relation to its high bindingaffinity to ubiquitous membrane-bound or intracellularproteins, the FK-binding proteins (FKBP). Owing tothe cyclic 31-membered macrolide structure, Sirolimusis a hydrophobic, temperature-, light-, oxidant- and pH-sensitive compound with a narrow therapeutic index. Itsbreakdown or degradation produces a ring-open isomer(34-hydroxy sirolimus) retaining less than 10% of theimmunosuppressive activity of sirolimus (Yatscoff et al.,1995). All these properties were integrated into thetissue homogenization. Rabbit blood was used as thehomogenization medium instead of the more widelyused phosphate-buffered saline or saline. At the sametime, the homogenization process generated heat thatcould cause sirolimus degradation; therefore, a pause of0.2 min was integrated into the middle of homogeniza-tion. The whole process could release the heat intothe cooling system. Potassium hydroxide solution wasused to break down sirolimus between two consecutivehomogenization rows and thus avoid carryover. Thebasic solution also helped remove the lipids on thehomogenization probes. A flowing water bath was usedto remove the blood first and later the potassiumhydroxide solution. A flowing acetonitrile bath wasused, finally, to remove lipid remains and water. Thepotassium hydroxide was also used to clean tools forcarryover or cross-contamination elimination.

Calibration curves

In order to evaluate the calibration curves, three runsdesignated for the evaluation of accuracy and precisionwere performed consecutively, including nine concen-tration levels of STDs, the lower limit of quantitation(LLOQ), 49.9 pg/mL, and the upper limit of quantita-tion (ULOQ), 31.9 ng/mL. Table 1 shows the statisticalsummary of calibration standards. Three significantfigures were used for the calculation. The nominal con-centrations are in bold. The coefficient of the deter-mination, r2, was between 0.9994 and 0.9998. The meanbias was between −5.8 and 1.6%. The CV was within3.8% for all concentration levels.

Accuracy of LLOQ and ULOQ

Six replicates of LLOQ and ULOQ samples were in-cluded in each of the three consecutive runs designated



for the evaluation of accuracy and precision. The indi-vidual bias and mean bias were calculated by comparingthe calculated concentrations with the nominal concen-trations. Table 2 shows the statistical summary of LLOQand ULOQ samples. The bias from 17 of 18 LLOQsamples was within an acceptance criterion of 20%.The bias from all 18 ULOQ samples was between −5.0and 1.6%. The mean bias was 1.3% for LLOQ samplesand 1.1% for ULOQ samples. The CV was 10.8% forLLOQ samples and 1.1% for ULOQ samples.

Accuracy and precision of blood quality controls

Four concentration levels of blood quality control sampleswith six replicates of each level were used in the threeconsecutive runs designated for the evaluation of theaccuracy and precision post tissue homogenization. Thecalculated concentrations for all quality control samples

are shown in Table 3. The bias was calculated by com-paring the calculated concentrations and the nominalconcentrations. All quality control samples were withinthe acceptance criterion of 15%. The mean bias wasbetween −4.5 and 0.2%. The CVs, calculated using theconcentrations obtained from all three runs designatedfor accuracy and precision, were within 5.3%.

Accuracy and precision of tissue quality controls

Tissue quality controls were prepared to simulate realsamples. Two mass ranges were used for the tissuequality controls. One mass range was between 5 and15 mg, and was marked as ~10 mg tissue in Table 4.The other mass range was between 70 and 130 mg,marked as ~100 mg tissue in Table 4. In the prepara-tion of the homogenization quality controls, 20 µL ofworking solution of sirolimus was injected into the

Table 1. Statistic summary of calibration standards (nominal concentrations are in bold, unit pg/mL)

Calculated concentration (pg/mL)

Run ID 49.9 79.9 199 499 998 3990 9980 29,900 31,900 r2

1 49.5 81.4 180 474 990 3980 9980 30,200 33,200 0.99962 49.9 80.6 193 482 998 3950 9880 29,700 31,900 0.99983 51.3 79.0 192 455 977 994a 10,200 31,100 32,200 0.9994

Mean (pg/mL) 50.2 80.3 188 470 988 3960 10,000 30,300 32,400%CV 1.9 1.5 3.8 2.9 1.1 0.5 1.6 2.3 2.1%Bias 0.6 0.5 −5.5 −5.8 −1.0 −0.8 0.2 1.3 1.6

a Not used for the statistical calculation because the wrong concentration level was inadvertently chosen for the pipette loading.

Table 2. Statistical summary of LLOQ and ULOQ samples

LLOQ ULOQ

Concentration Concentration(pg/mL) (pg/mL)

Run ID 49.9 %Bias 31,900 %Bias

1 61.4 23.0 32,800 −2.049.8 −0.2 32,400 1.348.8 −2.2 32,400 −0.651.3 2.8 31,500 1.041.8 −16.2 32,200 −0.353.0 6.2 32,000 1.6

2 56.0 12.2 32,200 0.756.8 13.8 31,900 −1.748.7 −2.4 32,300 1.447.3 −5.2 32,300 −5.050.6 1.4 32,400 0.646.7 −6.4 33,100 0.6

3 50.7 1.6 32,300 −0.247.9 −4.0 31,900 0.649.9 0.0 32,300 −2.049.5 −0.8 32,600 −2.840.1 −19.6 32,100 −1.259.7 19.6 32,100 1.1

Mean concentration (pg/mL) 50.5 32,200%CV 10.8 1.1%Bias 1.3 1.1



tissue to prepare the tissue quality control samples. Atthe same time, 20 µL of the working solution was alsospiked into a BD test tube and stored at ~−70°C. Priorto homogenization, these tissue-free samples wereremoved from the freezer, 2.0 mL of rabbit blood wasadded into each tube, and was then homogenized alongwith the tissue quality controls. Table 4 shows thestatistical results from all homogenization qualitycontrol samples. A total of 108 samples went throughthe homogenization process. Among them, 36 weretissue-free samples and 72 were tissue-containedsamples. All 108 homogenization quality controlsamples were equally divided into three homogenizationbatches. Only one batch was homogenized each day.There was no significant variation between the tissue-freesamples and the tissue-containing samples. A total of106 out of 108 homogenization samples came outwithin a bias of 25%; 98 of 108 homogenization samplescame out within a bias of 20%; 86 of 108 homo-genization samples came out within a bias of 15%. Allthese numbers reflect the overall bias of the methodfrom working solution spiked to sample injection simi-lar to the previous report on the chemical analogzotarolimus (Zhang et al., 2006b, 2007). Most homo-genization samples showed positive bias. This mightbe caused by the splashing of the rabbit blood at thebeginning of the homogenization. An increase in thehomogenization media would reduce the variation

Table 3. Statistical summary of quality control (QC) samples (nominal concentrations are in bold)

QC 1 QC 2 QC 3 QC 4

Concentration Concentration Concentration Concentration(pg/mL) (pg/mL) (pg/mL) (pg/mL)

Run ID 79.9 %Bias 99.9 %Bias 27,900 %Bias 29,900 %Bias

1 74.1 −0.7 108 8.1 27,500 −1.7 29,600 −1.274.1 −0.7 103 3.1 27,000 −3.5 29,400 −1.976.7 −0.4 91.1 −8.8 27,000 −3.5 29,500 −1.581.5 0.2 105 5.1 27,200 −2.7 30,700 2.574.9 −0.6 104 4.1 26,100 −6.7 28,700 −4.270.8 −1.1 97.3 −2.6 26,400 −5.6 29,000 −3.2

2 77.3 −0.3 91.8 −8.1 27,300 −2.4 29,000 −3.274.4 −0.7 99.5 −0.4 26,400 −5.6 29,300 −2.274.8 −0.6 105 5.1 26,300 −6.0 29,100 −2.973.7 −0.8 95.8 −4.1 26,600 −4.9 29,600 −1.274.8 −0.6 100 0.1 26,400 −5.6 28,600 −4.675.9 −0.5 99.7 −0.2 27,200 −2.7 29,600 −1.2

3 73.5 −0.8 103 3.1 27,900 −0.2 29,600 −1.274.8 −0.6 109 9.1 27,800 −0.6 30,300 1.189.7 1.2 95.7 −4.2 28,000 0.1 29,800 −0.577.9 −0.3 100 0.1 27,600 −1.3 29,800 −0.576.5 −0.4 93.2 −6.7 27,600 −1.3 30,100 0.578.6 −0.2 101 1.1 27,500 −1.7 30,100 0.5

Mean 76.3 100 27,100 29,500concentration(pg/mL)%CV 5.3 5.2 2.2 1.9%Bias −4.5 0.2 −3.1 −1.4

because the relative volume loss could be reduced.However, it was not practical to use more rabbit bloodfor large studies. The mean bias, expressed as the accu-racy of the method, was within 14.9% for all levels ofhomogenization quality controls. The CVs, expressed asthe precision of the method, were within 9.2% for alllevels of homogenization quality controls.

Selectivity

Six individual lots of rabbit tissue were used to screenpossible interference from endogenous matrix com-ponents. These blank tissues were homogenized andextracted as a normal tissue. The peak area response atthe retention time of sirolimus for the selectivity sampleswas to be compared with the peak area response of thesirolimus at the LLOQ. There was no peak observedfrom any one of the selectivity samples. Figure 3(a)shows a chromatogram of sirolimus in a selectivitysample with internal standard; Fig. 3(b) shows achromatogram of an LLOQ sample and Fig. 3(c) showsa chromatogram of the internal standard, zotarolimus.

Applications

The method was used to analyze the sirolimus tissuesamples for a pharmacokinetics study. Thirty-two rab-bits were implanted with 13 × 3.0 mm Cypher stents,



Tab

le 4

. Sta

tistic

al s

umm

ary

of h

omog

eniz

atio

n qu

ality

con

trol

sam

ples

(no

min

al c

once

ntra

tions

are

in b

old,

uni

tpg/

mL

)

Mea

sure

d co

ncen

trat

ions

(pg

/mL

)

125

pg/m

L24

,700

pg/m

L

Run

ID

No

tissu

e%

Bia

s~

10m

g tis

sue

%B

ias

~100

mg

tissu

e%

Bia

sN

o tis

sue

%B

ias

~10

mg

tissu

e%

Bia

s~1

00m

g tis

sue

%B

ias

113

14.

915

624

.915

524

.127

,100

9.9

25,8

004.

624

,400

−1.1

139

11.3

154

23.3

139

11.3

28,1

0013

.925

,500

3.4

27,9

0013

.113

911

.315

020

.114

516

.127

,000

9.5

23,9

00−3

.124

,600

−0.2

141

12.9

131

4.9

154

23.3

27,9

0013

.125

,300

2.6

26,7

008.

314

314

.513

68.

913

04.

127

,600

11.9

25,1

001.

825

,400

3.0

130

4.1

123

−1.5

139

11.3

28,1

0013

.925

,700

4.2

25,9

005.

02

135

8.1

133

6.5

165

32.2

28,2

0014

.425

,300

2.6

28,5

0015

.613

14.

914

012

.115

120

.927

,800

12.7

26,2

006.

228

,300

14.8

131

4.9

152

21.7

155

24.1

27,4

0011

.125

,900

5.0

28,6

0016

.012

2−2

.314

919

.314

616

.927

,900

13.1

26,4

007.

128

,700

16.4

142

13.7

137

9.7

173

38.6

27,7

0012

.325

,900

5.0

25,6

003.

813

04.

113

68.

913

04.

126

,300

6.6

26,1

005.

826

,500

7.5

314

314

.514

415

.312

71.

727

,300

10.7

28,7

0016

.427

,200

10.3

143

14.5

143

14.5

134

7.3

27,0

009.

526

,100

5.8

27,2

0010

.313

58.

112

71.

714

012

.128

,000

13.5

26,6

007.

926

,200

6.2

127

1.7

126

0.9

125

0.1

27,9

0013

.124

,900

1.0

27,3

0010

.714

012

.114

717

.713

911

.328

,800

16.8

25,2

002.

228

,500

15.6

138

10.5

143

14.5

136

8.9

28,4

0015

.226

,800

8.7

25,2

002.

2

Mea

n13

514

014

327

,600

25,8

0026

,800

(pg/

mL

)%

CV

4.6

7.1

9.2

2.2

3.8

5.3

%B

ias

8.6

12.4

14.9

12.3

4.8

8.7



Figure 3. (a) A chromatogram of sirolimus in a selectiv-ity sample. (b) A chromatogram of sirolimus in an LLOQsample. (c) A chromatogram of zotarolimus in a selectivitysample.

and a group of four animals were necropsied at eachtime point of 0.25, 1, 2, 3, 5, 7, 14 and 28 days afterimplantation. The arteries were removed and analyzedusing the method described above. The tissues includestented arteries, proximal and distal arteries segments,and aortas. Figure 4 shows the results of sirolimus con-centration in different arteries.

CONCLUSIONS

A high-throughput analytical method for the determi-nation of sirolimus in rabbit arteries was developed andused for a pharmacokinetics study. The method intro-duced whole blood as a homogenization medium to

Figure 4. Application (13 × 3.0 mm Cypher stents).

stabilize the homogenate of sirolimus. Homogenizationquality control samples were introduced into themethod to monitor the method variation of the entireprocess from the preparation of the homogenizationcontrol samples to the sample injection. Based on theexperimental results, 106 of 108 homogenization qualitycontrol samples were within a bias of 25%. This isappropriate for the overall bias acceptance. As long asthe homogenization protocol is developed, tissue-freesamples can be used to monitor the overall accuracyand precision of the method, including homogenization.The introduction of homogenization quality controlsamples can also be used for the stability evaluation inboth solid tissue and tissue homogenate formats. Withminor modifications, this method can be used forsirolimus analysis in other tissues like myocardium,liver, kidney etc.

Acknowledgments

The authors appreciate the consistent encouragementand support from Drs Richard Krasula, Klaus Krauser,Fritz Richter, Sandra Burke and Lewis Schwartz.

REFERENCES

Calne RY, Collier DS, Lim S, Pollard SG, Samaan A, White DJ andThiru S. Rapamycin for immunosuppression in organ allografting.Lancet 1989; 2: 443–444.

Eng CP, Sehgal SN and Vezina C. Activity of rapamycin (AY-22,989)against transplanted tumors. Journal of Antibiotics 1984; 37: 1231–1237.

Gallo R, Padurean A, Jayaraman T, Marx S, Roque M, Adelman S,Chesebro J, Fallon J, Fuster V, Marks A and Badimon JJ. Inhibi-tion of intimal thickening after balloon angioplasty in porcine coro-nary arteries by targeting regulators of the cell cycle. Circulation1999; 99: 2164–2170.

Kimball PM, Kerman RH and Kahan BD. Production of synergisticbut nonidentical mechanisms of immunosuppression by rapamycinand cyclosporine. Transplantation 1991; 51: 486–490.

Morris RG, Salm P, Taylor PJ, Wicks FA and Theodossi A. Com-parison of the reintroduced MEIA assay with HPLC-MS/MS forthe determination of whole-blood sirolimus from transplant recipi-ents. Therapeutic Drug Monitoring 2006; 28: 164–168.



Napoli KL and Taylor PJ. From beach to bedside: history of thedevelopment of sirolimus. Therapeutic Drug Monitoring 2001; 23:559–586.

Pieri M, Miraglia N, Castiglia L, Genovese G, Basilicata P, SimonelliA and Acampora A. Determination of rapamycin: quantification ofthe sodiated species by an ion trap mass spectrometer as an alter-native to the ammoniated complex analysis by triple quadrupole.Rapid Communications in Mass Spectrometry 2005; 19: 3042–3050.

Sehgal SN, Baker H and Vezina C. Rapamycin (AY-22,989), a newantifungal antibiotic. II. Fermentation, isolation and characteriza-tion. Journal of Antibiotics 1975; 28: 727–732.

Serruys PW, Regar E and Carter AJ. Rapamycin eluting stent: theonset of a new era in interventional cardiology. Heart 2002; 87:305–307.

Streit F, Armstrong VW and Oellerich M. Rapid liquid chromatography–tandem mass spectrometry routine method for simultaneous deter-mination of sirolimus, everolimus, tacrolimus, and cyclosporin A inwhole blood. Clinical Chemistry 2002; 48: 955–958.

Vogeser M, Fleischer C, Meiser B, Groetzner J, Spohrer U andSeidel D. Quantification of sirolimus by liquid chromatography–tandem mass spectrometry using on-line solid-phase extraction.Clinical Chemistry and Laboratory Medicine 2002; 40: 40–45.

Volosov A, Napoli KL and Soldin SJ. Simultaneous simple andfast quantification of three major immunosuppressants by liquidchromatography–tandem mass-spectrometry. Clinical Biochemistry2001; 34: 285–290.

Yatscoff R, LeGatt D, Keenan R and Chackowsky P. Blood distribu-tion of rapamycin. Transplantation 1993; 56: 1202–1206.

Yatscoff RW, Wang P, Chan K, Hicks D and Zimmerman J.Rapamycin: distribution, pharmacokinetics, and therapeutic rangeinvestigations. Therapeutic Drug Monitoring 1995; 17: 666–671.

Zhang J, Kim EJ, Ji QC and El-Shourbagy TA. Developing multiplehigh-throughput GLP methods for an investigational drug candi-date in various matrices within a 96-well plate. Rapid Communica-tions in Mass Spectrometry 2006a; 20: 3755–3760.

Zhang J, Reimer MT, Alexander NE, Ji QC and El-Shourbagy TA.Method development and validation for zotarolimus concentrationdetermination in stented swine arteries by liquid chromatography/tandem mass spectrometry detection. Rapid Communications inMass Spectrometry 2006b; 20: 3427–3434.

Zhang J, Reimer MT, Ji QC, Chang MS, El-Shourbagy TA, Burke Sand Schwartz L. Accurate determination of an immunosuppressantin stented swine tissues with LC-MS/MS. Analytical and Bioanaly-tical Chemistry 2007; 387: 2745–2756.

determination of sirolimus in rabbit arteries using liquid chromatography separation and tandem mass...

Documents