Anal Bioanal Chem (2012) 404:2043–2055 DOI 10.1007/s00216-012-6272-4

TECHNICAL NOTE

Anidulafungin—challenges in development and validation of an LC-MS/MS bioanalytical method validated for regulated clinical studies Tanja Alebic-Kolbah & Michael S. Modesitt

Received: 23 April 2012 / Revised: 5 July 2012 / Accepted: 13 July 2012 / Published online: 28 July 2012 # Springer-Verlag 2012

Abstract Anidulafungin is a semi-synthetic echinocandin with antifungal activity, usually administered as an intravenous infusion. In order to determine the pharmacokinetics (PK) of anidulafungin in pediatric patients, a sensitive high performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) bioanalytical method (M1) was developed and validated for quantification of anidulafungin in plasma. During analysis of incurred samples (samples collected from patients enrolled in a clinical study) an isobaric chromatographic interference was observed. The source of interference was identified as an anidulafungin open-ring form (D1) and its impact on the quantification of anidulafungin was investigated. It was found that accurately quantifying anidulafungin in incurred samples required chromatographic separation of the open-ring form from anidulafungin. The method was redeveloped to achieve the appropriate baseline separation and to avoid experimental conditions that favored opening the anidulafungin ring. The extraction of anidulafungin from plasma by protein precipitation remained unchanged, but the changes in chromatography warranted validation of a new method, M2, 2 years after M1 was validated. Incurred samples from three studies that were previously analyzed by M1 and were within confirmed long-term frozen stability were then reanalyzed by M2. Although the incurred sample reproducibility tests on those samples passed for each of the two methods, comparison of concentrations from the same samples obtained by M1 and M2 revealed that an overestimation T. Alebic-Kolbah (*) Pfizer Inc, 445 Eastern Point Rd, Groton, CT 06340, USA e-mail: [email protected] M. S. Modesitt PPD, 2244 Dabney Rd, Richmond, VA 23230, USA

of anidulafungin following the M1 method exceeded acceptance criteria. The new HPLC-MS/MS method (M2) is applicable for quantification of anidulafungin within a nominal range 50–20,000 ng/mL and requires a 50 μL human plasma aliquot. A linear, 1/concentration squared weighted, leastsquares regression algorithm was used to generate the calibration curve and its parameters were used to quantitate the incurred samples. The inter-assay accuracy in heparin human plasma validation ranged from −4.33 to 0.0386 % and precision was ≤7.32 %. The method M2 was validated for use in regulated bioanalysis and is presently used to quantitate anidulafungin in plasma samples from clinical studies. Keywords Anidulafungin . HPLC-MS/MS . Anidulafungin open-ring form . Pediatric blood collection . Regulated bioanalysis . Validation

Introduction Anidulafungin (Fig. 1a) is a semi-synthetic lipopeptide from the echinocandin group synthesized from a fermentation product of Aspergillus nidulans. The drug substance is a molecular mixture of anidulafungin and fructose. Anidulafungin is a non-competitive inhibitor of (1,3)-β-D-glucan synthase. This enzyme is present in fungal cells but absent in mammalian cells. (1,3)-β-D-glucan synthase is required for synthesis of β-linked glucan, which comprises a major portion of the cell wall in many pathogenic fungi. Suppression of cell wall glucan synthesis leads to osmotic instability and eventual cell death [1]. Following a request for a bioanalytical method that minimizes the total volume of blood collected in a typical pediatric PK study, we developed and validated in 2007 a sensitive HPLC-MS/MS method for quantification of anidulafungin in plasma (method M1). The earlier publications

2044

T. Alebic-Kolbah, M.S. Modesitt

a HO

OH

HO OH OH HO

O

O

-

NH

O

N

NH

- 2(CHOHCH3)

O CH 3

O

N

O

OH NH

HO

HO

NH

OH

NH

NH H3C

O

O

O N

O N

NH

NH H3C

O

OH

NH

H3C

R

NH

O

OH

OH

OH

OH

M-H =m/z 1138

R OH

M-H =m/z 898

O O

R=

CH3

b H3C O

HO

N

OH O

NH

O O

O

N

OH NH

HO

OH HO

CH3

HO O NH

NH2 N

HO

O

O

OH

CH3

CH3

Fig. 1 a Structure of anidulafungin and of the monitored fragment. b Structure of anidulafungin open-ring form D1

describing bioanalytical methods for measuring anidulafungin in plasma [2, 3] did not meet the requirements for reduced blood sample volume (0.2–0.5 mL per draw for 50 μL of plasma), or the required lower limit of quantitation (LLOQ of 50 ng/mL). The M1 method was previously only marginally described in a manuscript publishing clinical data from a combined voriconazole–anidulafungin study [4]. A recently published LC-UV method for anidulafungin in human plasma [5], and two LC-MS/MS methods for analysis of anidulafungin together with a number of other

azoles and echinocandins in peripheral blood [6], or together with several other antifungal drugs in human plasma [7], were developed for therapeutic drug monitoring (TDM) and are not suitable for analysis of PK samples from regulated studies. Another recently published method [8] for determination of micafungin and anidulafungin in human plasma was developed and validated to “facilitate dosage adjustments”. The authors found that “because of the unique properties of the analyte molecules, no acceptable validation results could be achieved” using the mass spectrometer,

Anidulafungin

therefore, they used the “mass spectrometric chromatograms only for identity confirmation of the observed UV-peaks” [8]. Both of our LC-MS/MS anidulafungin plasma methods, M1 and M2, were successfully validated with a calibration range from 50 to 20,000 ng/mL. Our success in validating the anidulafungin LC-MS/MS methods may have been due to our choice of the ionization mode and the anidulafungin product ion. When choosing the MS/MS transition to be monitored for quantification of an analyte in multiple reaction monitoring (MRM), a stable product ion generated by breaking the molecule is generally preferred to the nonspecific loss of water or ammonia. In MRM negative ionization mode we monitored the transitions m/z 1,138→898 for anidulafungin and m/z 1,149 → 909 for the stable isotope-labeled internal standard ([2H]11-anidulafungin). Both compounds shared the same fragmentation pattern, as expected. The loss of two –C2H4O fragments (2× −44 amu) and a loss of –HOC6H4(CHOH)2 (−152 amu), all three from the peptide ring of the deprotonated anidulafungin precursor m/z 1,138, yielded a stable product ion with m/z 898. The molecular structure of anidulafungin is shown in Fig. 1a and the MS/MS spectrum of anidulafungin in negative ionization mode in Fig. 2a. Martens-Lobenhoffer et al. [8] chose a less favorable and often unreliable product ion (m/z 1,123) formed most likely by the loss of water (−18 amu) from the protonated anidulafungin molecular ion; they monitored the m/z 1,141→1,123 transition in the positive ionization mode. Generally, validations should not be considered complete until incurred samples (samples collected from patients enrolled in a clinical study) have been analyzed using that particular method and the validity of the method thus confirmed. During analysis of incurred samples from an anidulafungin clinical study using the validated method M1, an isobaric chromatographic interference was observed. A similar, more pronounced chromatographic interference in the shape of a shoulder preceding the analyte peak was observed while developing the LC-MS/MS method for anidulafungin in rat plasma. The source of interference was identified as an anidulafungin open-ring form (D1). The structure of D1 and its negative ionization mode MS/MS spectrum are presented in Figs. 1b and 2b, respectively. Historically, the ring opening of echinocandins was well known among the medicinal and process chemists. Balkovec et al. published in 1991 and 1992 that the echinocandins with a hemiaminal hydroxyl group at C5 of the ornithine were “unstable at pH>7 and underwent a facile base-catalyzed ring opening and rearrangement to the five-membered ring hemiaminal isomer” [9]. A more recent publication by Norris et al. [10] from 2008 discusses in detail the development and behavior of anidulafungin from the pharmaceutical science point of view. The N-carboxyhemiaminaldiol functional group, being the most reactive group of the echinocandin peptide ring was

2045

easily hydrolyzed, resulting in ring opening and loss of biological activity. The final product of the ring opening was thought to be a pyrrolidino ring on the end of the peptide chain [10]. The anidulafungin peptide ring and the open-ring isomer were described to have the same molecular formula, but different retention times in HPLC [10]. In addition to the degradation curves of the cyclic peptide at various pH values, from pH 3 to 10, the kinetics of the formation of the open-ring at the pH 7.2 was also studied with the purpose to optimize the manufacturing of the anidulafungin API [10] (note that degradation studies were carried out with the echinocandin B cyclic peptide and not with anidulafungin as the final product of the synthesis). During bioanalytical work with anidulafungin, we never observed more than traces of D1 in anidulafungin stock solutions or calibrators and quality control samples in plasma. An unknown interference at such a low level is well within the allowed experimental error or even instrument noise and will not generally trigger the attention of the bioanalytical chemist. However, when the presence of the D1 peak in some of the incurred samples became visible, the impact of the openring isomer on the quantification of anidulafungin in incurred clinical samples was investigated. It was obvious that accurately quantifying anidulafungin in our studies required chromatographic separation of the D1 from anidulafungin because in MRM both anidulafungin and its open form shared the same precursor→product transition (see Fig. 2a and b). The method had to be redeveloped to achieve the appropriate baseline separation and to avoid experimental conditions that favored opening the anidulafungin ring, such as alkaline conditions during sample analysis that are present in M1. The extraction of anidulafungin from plasma by protein precipitation remained unchanged, but the changes in chromatography warranted validation of a new method, M2, 2 years after the successful validation of M1. Incurred samples from three studies that were previously analyzed by M1 were then reanalyzed by M2. Although the incurred sample reproducibility (ISR) tests on those samples passed for each of the two methods, comparison of anidulafungin concentrations from the same samples obtained by M1 and M2 revealed an overestimation of anidulafungin concentration obtained by M1 method. This is the first detailed publication of an anidulafungin LC-MS/MS method validated for use in regulated bioanalysis. The method, M2, addresses the potential interference of the anidulafungin open-ring form D1 with quantitation of anidulafungin and minimizes the potential for generation of D1 from anidulafungin during analysis. Furthermore, while other published bioanalytical methods for analysis of anidulafungin in incurred samples from dose adjustment studies have demonstrated the analyte stability in matrix for only a

2046

T. Alebic-Kolbah, M.S. Modesitt

a

1094.7

2.0e6 1.9e6 1.8e6 1.7e6 1.6e6 1.5e6 1.4e6 566.6

Intensity, cps

1.3e6 1.2e6 1.1e6 1.0e6 9.0e5 8.0e5 7.0e5

1050.8

6.0e5 5.0e5 4.0e5

1139.0

3.0e5

898.7 522.7

2.0e5

942.9

548.4

1.0e5

530.1 461.3 496.1

400

450

500

1000.6 1033.0

550

600

650

700

750

800

850

900

950

1000

1077.0

1050

1100

1150

1200

m/z, amu

b

1094.5

7.0e5 6.5e5 6.0e5 5.5e5 5.0e5

Intensity, cps

566.5

4.5e5 4.0e5 3.5e5 3.0e5 2.5e5 1138.7

2.0e5

1025.0

1051.5

1.5e5 690.3

943.1 1102.3

1.0e5 898.8 732.1

5.0e4 400

450

500

550

600

650

700

750

800

850

900

950

1000

1050

1100

1150

1200

m/z, amu

Fig. 2 a MS/MS spectrum of anidulafungin in negative ionization mode. b MS/MS spectrum of anidulafungin open-ring form D1 in negative ionization mode

Anidulafungin

2047

few days, a period suitable for TDM, the anidulafungin stability in frozen plasma was demonstrated for a period exceeding 2 years when using M2. Our method makes it possible to quantify anidulafungin in plasma samples from long duration worldwide Phase III/IV clinical studies.

Experimental section Chemicals and materials Anidulafungin fructose and its internal standard (PF-04571511, IS, a [H2]11-anidulafungin) were received from Pfizer Inc (Groton/New London, CT). The IS had a stable label on the C5 aliphatic chain, chemical purity of 94.8 % by HPLC, and no unlabeled mass present by quantifying the extracted ion chromatograms from flow injections using negative mode electrospray ionization MS. Human plasma with sodium heparin as an anticoagulant was purchased from Bioreclamation (Hicksville, NY) and Biological Specialty (Colmar, PA). Milli-Q reagent water (Millipore, Billerica, MA) was used throughout the work. Ammonium acetate was purchased from Sigma-Aldrich (St. Louis, MO). Acetonitrile, methanol, glacial acetic acid (99.7 %min.), 0.1 % formic acid in acetonitrile and 0.1 % formic acid in water were purchased from VWR Scientific (Radnor, PA). The 96-position, 2.0-mL, square-well, conical-bottom, polypropylene plate and the adhesive sealing film for 96well plate were purchased from VWR Scientific (Radnor, PA). Ammonium acetate (1.0 M) was prepared by dissolving 7.71 g of ammonium acetate (CH3CO2NH4; FW077.08) and diluting to volume with water in a 100 mL volumetric Table 1 Gradient program for the analytical and guard columns in method M2

Step

Total time (min)

flask. Ammonium acetate (50 mM, pH 4.0) was prepared by diluting 50 mL of 1.0 M ammonium acetate in ~500 mL of water and then adjusting the pH to 4.0 using acetic acid. The solution is then brought to volume in a 1-L volumetric flask with water. Stock diluent, also used as Autoinjector wash 1, acetonitrile–ammonium acetate (50 mM, pH 4.0; 50:50, v/v), was prepared by combining 500 mL of acetonitrile with 500 mL of ammonium acetate (50 mM, pH 4.0). Autoinjector wash 2 was water–acetonitrile–methanol (40:30:30, v/v/v). Instrumentation—LC-MS/MS The HPLC system used for the quantification of anidulafungin consisted of two solvent-delivery systems built from four LC-10AD VP pumps (Shimadzu, Columbia, MD) and an LCPAL autoinjector (CTC Analytics, Zwingen, Switzerland). The injector loop volume was 10–15 μL and the injection volume was set at 10 μL. Chromatographic separations were performed with a BDS Hypersil C8 column, 2.1 mm×100 mm, 5 μm (ThermoFisher, Waltham, MA) at ambient temperature, connected through a Valco valve to the guard column, BDS Hypersyl C8, 2.1× 20 mm, 5 μm (ThermoFisher, Waltham, MA). Two mobile phases were used in the chromatography. Mobile phase A was 0.1 % formic acid in water and mobile phase B was 0.1 % formic acid in acetonitrile. The mobile phase gradients for the analytical column and the guard column are described in Table 1. The Valco valve was set up to switch the flow from the initial position A to position B at 4.30 min, then back to the initial position A at 10.00 min. Analyses were performed on a Sciex API 4,000 (Sciex, Concord, ON, Canada) triple quadrupole mass spectrometer, operating in negative ion mode,

Flow rate (μL/min)

Mobile phase composition A (%)

B (%)

Analytical pump program 0 0.00 1 0.30 2 9.00 3 9.01 4 10.00 5 10.01 Guard wash pump program 0 0.00 1 4.20 2 4.30 3 6.50 4 10.00 5 10.10

500 500 500 500 500 500

70 70 37 10 10 70

30 30 63 90 90 30

100 100 500 500 500 100

70 70 70 10 10 70

30 30 30 90 90 30

2048

using electrospray interface and MRM. Settings were: ion spray voltage at −4,500 V and ion source temperature at 450 °C. Nitrogen was used for collision gas flow with a setting of 8, curtain gas flow at 20 psi, nebulizer gas flow at 45 psi and turbo ion gas flow at 45 psi. A declustering potential of −100 V, exit potential of −10 V, collision energy of −36 V, collision exit potential of −13 V, and a dwell time of 100 ms was set for both anidulafungin and its IS. MRM transitions monitored were: anidulafungin, m/z 1,138→898, and for the IS, m/z 1,149→909. The mass resolution on Q1 and Q3 was set to unit (0.6 to 0.8 amu at peak half height). Preparation of standards Anidulafungin stock solutions were prepared at a nominal concentration of 1.00 mg/mL in acetonitrile–ammonium acetate (50 mM, pH 4.0; 50:50, v/v), and stored at −20 °C. A PF-04571511 (Internal Standard) stock was prepared at a nominal concentration of 1.00 mg/mL in acetonitrile–ammonium acetate (50 mM, pH 4.0; 50:50, v/v), and also stored at −20 °C. An internal standard working solution (ISWS) was prepared on the day of use at a nominal PF04571511 concentration of 100 ng/mL in methanol. Anidulafungin open-ring form (D1) solution An external solution of the anidulafungin open-ring form (D1) was prepared and injected at the beginning and end of each validation run to evaluate the chromatographic separation between anidulafungin and D1. The D1 external solution was prepared according to the procedure outlined below and then diluted to a final concentration of ~2,250 ng/mL in methanol–ammonium acetate (50 mM, pH 4.0; 25:75, v/v); it was stored at −20 °C (the D1 concentration estimation was based on the assumption that 100 % of anidulafungin had converted into D1 and remained stable in the solution). To prepare the anidulafungin open-ring form (D1) solution, at approx. 1 mg/mL, 1 mL of methanol was added to a pre-weighed aliquot of anidulafungin stock (~5 mg) and mixed well. Ten microliters of 1 N NaOH was added to the stock solution, mixed well, and allowed to equilibrate at room temperature for 10 min. The reaction mixture was then diluted with 4 mL of water and mixed well. Using a vacuum manifold, an Oasis HLB 5 cc 200-mg cartridge column was conditioned with 5 mL of methanol, followed by 5 mL of water. The diluted reaction mixture was passed through the column. The column was washed with 5 mL of methanol– water (25:75, v/v), followed by 5 mL of methanol–water (50:50, v/v). D1 was eluted from the column with 5 mL of methanol–water (75:25, v/v). Calibration standards (CAL) were prepared in human plasma, containing sodium heparin, at nominal anidulafungin concentrations of 50.0 ng/mL (CAL 1), 100 ng/mL

T. Alebic-Kolbah, M.S. Modesitt

(CAL 2), 200 ng/mL (CAL 3), 500 ng/mL (CAL 4), 1,500 ng/mL (CAL 5), 4,500 ng/mL (CAL 6), 10,000 ng/mL (CAL 7), 16,000 ng/mL (CAL 8), and 20,000 ng/mL (CAL 9). Quality control (QC) pools were prepared in human plasma, containing sodium heparin, at nominal anidulafungin concentrations of 50.0 ng/mL (QCLLOQ), 150 ng/mL (QCL), 750 ng/mL (QCML), 2,000 ng/mL (QCM), 6,000 ng/mL (QCMH), 15,000 ng/mL (QCH), and 30,000 ng/mL (QCDil). Pools were prepared using glass volumetric flasks. After thorough mixing, each calibration standard and quality control pool was frozen in daily portions stored in polypropylene tubes at −70 °C. The total matrix (plasma) content in calibration standards and QC samples was kept above 95 %. Analytical procedure for human plasma samples Samples were kept frozen at −70 °C until analysis when they were thawed on wet ice. A reagent blank, matrix blank, and matrix blank with internal standard were prepared and extracted with each analytical run. Extraction procedure steps The sample was vortex-mixed before transferring a 50 μL matrix aliquot to a position in a 2.0 mL, square-well, conical-bottom, 96-well polypropylene plate. ISWS (300 μL of 100 ng/mL methanol solution) was added to all samples except blanks without internal standard, to which 300 μL of methanol was added. The plate was covered with sealing film, vortex-mixed for 3 min, and then placed at 2 to 8 °C for 30 min. After additional vortex-mixing, the plate was centrifuged at 5,000 rpm for 5 min at 2 to 8 °C. The supernatant (200 μL) was transferred to a 96-position, 2.0-mL, square-well, conical-bottom, 96-well polypropylene plate. The sample extracts were diluted with 600 μL of ammonium acetate (50 mM, pH 4.0). The plate was covered with a seal and vortex-mixed for 3 min, followed by centrifugation at 5,000 rpm for 5 min at 2 to 8 °C. Data reduction The data system was configured to calculate and annotate the areas of anidulafungin and the IS peaks automatically. A calibration curve was constructed using peak area ratios (PARs) of the calibration standards by applying a linear, 1/ concentration squared weighted, least-squares regression algorithm. All concentrations were then calculated from their PARs against the calibration curve. The peak area ratios of anidulafungin and internal standard were determined using Analyst® Version 1.4.2 (PE Sciex) and concentrations of anidulafungin were calculated by Assist LIMS Version 5 software (PPD).

Anidulafungin

Validation of the method M2 A System Suitability check was performed prior to each validation batch. An extracted QCL (quality control sample at low level at 3× LLOQ, 150 ng/mL) was used to evaluate system suitability for response ratio (analyte peak area/IS peak area), overall peak shape, signal and retention time. The response ratio had to be within a set window. The peak shape, signal and retention time as well as separation from D1 were qualitative evaluations. The Linearity of the anidulafungin assay in human plasma was assessed with calibration lines consisting of nine calibration standards ranging from 50 ng/mL to 20,000 ng/mL. The Selectivity of the method was evaluated in ten different lots of human plasma. Each lot was analyzed with and without addition of IS at N01 for the blank plasma and N01 for blank plasma with IS. Any response with similar retention time to the analyte had to be ≤20.0 % of the individual response for the lower of the two duplicate LLOQ calibration standards. Any response with a similar retention time to the internal standard had to be ≤5.0 % of the response for the internal standard peak in the blank matrix with internal standard sample for each respective lot evaluated. Ninety percent of the matrix lots tested had to pass these criteria. The Ionization Effect Evaluation was performed with ten different lots of human plasma, all fortified with anidulafungin at the QCL level of 150 ng/mL. Results were considered acceptable if 90 % of the matrix lots tested had at least 67 % (2/3) of QCL samples within 15.0 % of the nominal concentration. A Cross Matrix Evaluation was performed by analyzing QCL (150 ng/mL), QCM (2,000 ng/mL), and QCH samples (15,000 ng/mL), in six replicates each, prepared in lithium heparin plasma, in one of the core validation batch runs. Each of the required minimum of three core validation batch runs to assess Precision and Accuracy contained a blank matrix (plasma) sample with and without added IS, freshly prepared duplicate calibration curves, validation QC samples in six replicates at the QCLLOQ (50 ng/mL), QCL (150 ng/mL), QCML (750 ng/mL), QCM (2,000 ng/mL), QCMH (6,000 ng/mL), and QCH (15,000 ng/mL) levels, and two blank plasma samples placed in the sequence in such a position to best assess potential Carryover. Dilution Integrity of anidulafungin in human plasma was assessed for tenfold dilution of QCDil (Dilution QC at 30,000 ng/mL) with blank matrix, performed in six replicates. Extraction Recovery of anidulafungin from human plasma was assessed in six replicates at QCL (150 ng/mL), QCM (2,000 ng/mL), and QCH (15,000 ng/mL) levels. The extraction recovery for IS was performed in 6 replicates at the 100 ng/mL concentration.

2049

Stability experiments were performed for primary anidulafungin stock solutions and for anidulafungin in matrix. The following stability experiments were done for anidulafungin in human plasma: thawed storage matrix stability, frozen storage matrix stability, freeze and thaw matrix stability, extract stability and re-injection reproducibility stability. The matrix storage stability samples were prepared in human plasma at the QCL (150 ng/mL) and QCH (15,000 ng/mL) anidulafungin level and analyzed in six replicates each. The matrix storage stability was assessed by comparing the mean concentration of the stability samples evaluated at each time point to the nominal concentration of the validation samples. For the extract stability (auto injector stability) the comparison was done at QCL (150 ng/mL) and QCH (15,000 ng/mL) level in six replicates each, at 15 °C in 1:3 supernatant/50 mM ammonium acetate, pH 4.0 (v:v). The re-injection reproducibility stability was performed at six QC levels (QCLLOQ at 50 ng/mL, QCL at 150 ng/mL, QCML at 750 ng/mL, QCM at 2,000 ng/mL, QCMH at 6,000 ng/mL, and QCH at 15,000 ng/mL), in six replicates each, at 15 °C in 1:3 supernatant/50 mM ammonium acetate, pH 4.0 (v/v). Acceptance Criteria: (1) an interference must not exceed 20.0 % (5.0 % for IS) of the individual response of the lower of the two duplicate LLOQ calibration standards; (2) a minimum of 75 % of the total number of calibration standards must be included in the final curve; (3) for quality controls, accuracy (% difference from theoretical) must be within ±15 % (20 % at QCLLOQ level) and precision (%CV) must be below 15 % (20 % at QCLLOQ level). For ISR, at least 67 % of reassayed samples must have their concentrations within ±20.0 % of the mean of the original and the reassayed concentration values to pass. The validation experiments set up and the acceptance criteria described herein are an example of the best practice currently used throughout the industry. With the exception of the three Precision and Accuracy core validation runs performed with six levels of Validation QCs, most of the other validation experiments were done with multiple replicates of QCL (which is generally set at 3× LLOQ), and QCH which is set between 75 and 80 % of the upper limit of quantitation. The extraction recovery may not be a critical parameter in an assay using protein precipitation and stable label internal standard; however, the extracted LLOQ must meet the criteria for back calculated concentration (±20.0 % of the nominal value) as well as the signal to noise (S/N) ratio of ≥5. Collection procedure for clinical study plasma samples Blood, 0.2–0.5 mL for children aged 2 days to less than 2 years, or 1–2 mL for older patients, was collected via a heparin/saline lock or indwelling venous cannula from the

2050

arm contralateral to the infusion site. Blood was collected in either sodium or lithium heparinized tubes. Plasma was separated by centrifugation at approximately 1,400×g for 10 min and stored at −20 °C or −70 °C within 1 h of collection (stability of anidulagungin in plasma was confirmed for more than 2 years at both temperature levels). After shipping from the clinical site (on dry ice), the samples were kept frozen at −70 °C until analysis.

Results and discussion During the attempt to adapt the human plasma HPLC-MS/ MS method M1 for rat plasma samples, an interference peak was observed in incurred rat plasma samples. A similar, much smaller interference peak was found in some incurred human plasma samples. About the same time in 2009, a study of clearance mechanism of anidulafungin [11] was published. In addition to in vitro CYP inhibition experiments and incubations with rat and human hepatocytes, the authors monitored in vitro degradation of anidulafungin in buffer and in human plasma. There was no evidence of Phase I or Phase II hepatic metabolism; however, clearance of anidulafungin seemed to be mainly due to a slow chemical degradation producing a ring-open peptide (D1) that lacked antifungal activity. D1 was subsequently being converted to peptidic degradants and eliminated in feces [11]. The rate of anidulafungin degradation in human plasma at 37 °C was similar to the rate observed in pH 7.4 phosphate buffer-saline. The initial concentration of anidulafungin decreased more than 95 % in 96 h. However, the open-ring degradant D1 was not stable in plasma although the initial rate of disappearance of anidulafungin was similar to the rate of D1 formation [11]. When it was confirmed that the interfering peak sharing the same precursor→product ion pair with anidulafungin was the open-ring form D1, it was necessary to identify whether the potential presence of D1 was interfering with quantification of anidulafungin in samples that were analyzed using the M1 method. To answer this question, the chromatographic conditions in M1 were modified to enable baseline separation of anidulafungin from D1. The separation in time was necessary because both anidulafungin and D1 fragment in the negative ionization MS/MS generating product ions of the same m/z ratio (e.g., m/z 898) from the same precursor (m/z 1,138) and with similar intensity (Fig. 2a and b). If D1 is present in the anidulafungin sample, the signal from the monitored anidulafungin MS/MS transition will be increased by the contribution from the same transition belonging to D1. This situation could be compared to the MS/MS of two enantiomers in achiral chromatography. Using the new chromatographic conditions set to separate anidulagungin from D1, we reassayed selected samples

T. Alebic-Kolbah, M.S. Modesitt

from three clinical studies previously analyzed by M1. Results from one study showed that there was no bias resulting from the interference from D1. However, results for the other two studies showed that greater than 35 % of the reanalyzed samples were outside the acceptance criteria (±15.0 %; ±20.0 % at the LLOQ level) when compared to their results obtained by M1. Also, the reanalysis showed an overall trend towards lower anidulafungin concentrations with a mean bias of approximately 13 %. However, we could not associate the variation in the relative response of D1 (expressed as % of the D1peak area to anidulafungin peak area in the same sample) in incurred samples from the three clinical studies with conditions related to sample collection, storage or handling. Moreover, D1 was not present in the QC samples stored for at least the same length of time, at the same temperature and analyzed together with incurred samples. While the peak of the degradant/open-ring anidulafungin D1 could be seen in older IS solutions, freshly prepared anidulafungin stocks, as well as QCs and calibrators in plasma contained only traces of D1. The slow chemical degradation of anidulafungin generating D1 in validation QC samples stored at −70 °C for frozen long-term stability experiments was well within experimental error and acceptance criteria. The discrepancy in content of D1 between incurred samples and laboratory-prepared QCs confirmed that the redevelopment of the new method (M2) for separating D1 from anidulafungin chromatographically was necessary. Bioanalytical methods are validated generally using calibrators and QCs prepared in matrix fortified with freshly prepared analyte solutions. This explains why, with only traces of the degradant present in calibration standards and validation QCs, both of our anidulafungin LC-MS/ MS methods, M1 and M2, were successfully validated, regardless of the chromatography. Yet only the M2 is set up to quantify anidulafungin in incurred samples without interference caused by potential presence of D1. Following the validation, M2 was used to reassay the samples from the two studies where the lower anidulafungin concentration trend was observed and is now being used for plasma sample analysis in anidulafungin clinical studies. Human plasma collection For the youngest pediatric population, 0.2–0.5 mL of blood for each time point is collected and an intravenous or capillary blood collection system (Microvette 500 from Sarstedt or similar) may be used. A blood volume of 0.2 mL is sufficient to harvest enough plasma for the analysis and one reanalysis either for repeat due to a failure, or incurred sample reproducibility. For older children and adults or when infrequent sampling is required, 1–2 mL of blood for each time point is collected into standard commercial collection

Anidulafungin

2051

Table 2 Chromatographic conditions for methods M1 and M2 Method

M1

M2

HPLC column Guard column Mobile phase composition A

Xbridge C18, 2.12×20 mm Xbridge C18, 2.1×10 mm

BDS Hypersil C8, 2.1×100 mm Javelin BDS Hypersil C8, 2.1×20 mm

B

1 M ammonium bicarbonate Water Methanol Ammonium hydroxide 1 M ammonium bicarbonate

Retention time Cycle time

Water Acetonitrile 1.4 min 5 min

Fig. 3 Anidulafungin LLOQ calibration standard (50 ng/mL nominal concentration) in human plasma, chromatogram by LC-MS/MS method M2

vol 1 90 10 0.6 1

0.1 % formic acid in water

0.1 % formic acid in acetonitrile

5 95 6.3 min 10.5 min

2052

tubes. Heparin was used as an anticoagulant; sodium and lithium were successfully cross validated as counter ions. Chromatography Difficulties in obtaining base line separation between anidulafungin and D1 by attempts to modify chromatographic conditions of M1 prompted redevelopment of the chromatography (Table 2). The Xbridge C18 2.1×20-mm chromatographic column and the basic bicarbonate-andacetonitrile-based mobile phase were replaced by a BDS Hypersil C8 2.1×100 mm column and an acidic formic acid–acetonitrile-based mobile phase. Anidulafungin transformation is accelerated at basic pH [10, 11], therefore, in M2 the anidulafungin plasma extracts were diluted with 50 mM ammonium acetate, pH 4.0. The resulting chromatographic conditions of method M2 successfully separate D1 from anidulafungin and reduce the transformation of anidulafungin in extracts. Fourfold increase in anidulafungin Fig. 4 a Anidulafungin: an incurred plasma sample, by LCMS/MS method M1. Anidulafungin and D1 are not separated in time; they both coelute at 1.35 min. b Anidulafungin: the incurred plasma sample from Fig. 4a, by LC-MS/MS method M2. Anidulafungin elutes at 6.08 min and is baseline separated from D1 which elutes at 5.54 min

T. Alebic-Kolbah, M.S. Modesitt

retention time on the column, from ~1.4 to ~6.1 min, only doubled the cycle time. A representative chromatogram of an anidulafungin LLOQ standard in human plasma analyzed by method M2 is given in Fig. 3. Example chromatograms of the same incurred anidulafungin plasma sample analyzed by methods M1 and M2 are shown in Fig. 4. In Fig. 4a, in the chromatogram obtained by method M1, D1 coelutes with anidulafungin at 1.35 min. In Fig. 4b, obtained by method M2, D1 elutes at 5.54 min and is separated from anidulafungin which elutes at 6.08 min. Validation and incurred sample reproducibility The above-described procedures for determination of anidulafungin in human plasma were validated according to the CDER Bioanalytical Guidance [12] with respect to limits of quantification, intra- and inter-assay precision and accuracy, selectivity, and stability.

Anidulafungin

2053

Fig. 4 (continued)

Selectivity The method selectivity and ionization effects were successfully evaluated in ten different lots of human plasma fortified with anidulafungin at the QCL (150 ng/mL) level. Precision and accuracy Under the above-described conditions, a linear relationship was established by plotting anidulafungin concentrations against the peak area ratio of anidulafungin to the internal standard. This new HPLC-MS/MS method (M2) is applicable to quantitation within a nominal anidulafungin range of 50 to 20,000 ng/mL and requires a 50-μL human plasma aliquot. A linear, 1/concentration squared weighted, least-squares regression algorithm is used to quantitate incurred samples. The inter-assay accuracy and precision in sodium heparin human plasma ranged from −4.33 to 0.0386 % and from 2.14 to 7.32 %, respectively (Table 3). Dilution integrity for tenfold dilution of anidulafungin at 30,000 ng/mL with blank human plasma into the valid curve range was also

confirmed. The mean apparent anidulafungin and IS recoveries were 103 and 109 %, respectively. The calibration curve range and placement of QCs spanned and adequately reflected the anidulafungin concentrations of the incurred study samples. Stability Freeze and thaw matrix stability for anidulafungin in human sodium heparin plasma was demonstrated for six cycles thawed on wet ice and frozen at −70 °C, and for three cycles thawed on wet ice and frozen at −20 °C, for each freeze/ thaw cycle. Thawed matrix storage stability was demonstrated for 26 h on wet ice. Long-term frozen matrix storage stability for anidulafungin in sodium heparin human plasma was established for 740 days at both −70 and −20 °C. For anidulafungin in lithium heparin human plasma, LTS was confirmed for 498 days at −70 °C. Processed extract stability, also known as autoinjector stability, was demonstrated for 151 h at 15 °C in supernatant - ammonium acetate (50 mM, pH 4.0; 1:3, v/v).

2054

T. Alebic-Kolbah, M.S. Modesitt

Table 3 Inter-assay precision and accuracy of anidulafungin in sodium heparin human plasma Run ID 1

2

3

N Theoretical Concentration Mean S.D. %C.V. % Difference from Theoretical

QCLLOQ (ng/mL)

QCL (ng/mL)

QCML (ng/mL)

QCM (ng/mL)

QCMH (ng/mL)

QCH (ng/mL)

45.0 49.3 47.1 48.9 55.7 47.8 45.6 48.5

149 147 137 155 146 153 138 142

744 738 784 745 718 752 735 727

1,920 1,970 1,930 1,930 1,830 1,880 1,940 1,960

6,180 5,870 5,970 6,120 6,100 6,110 5,840 5,980

14,800 15,000 15,300 15,200 14,800 14,800 14,800 14,900

48.4 45.5 44.1 47.3 46.0 41.5 48.1 53.1 53.5 45.5 18 50.0 47.8 3.50 7.32

144 142 153 137 149 148 164 143 135 137 18 150 146 7.66 5.26

782 694 739 720 720 711 726 733 704 714 18 750 732 23.6 3.23

2,040 1,970 1,950 1,980 1,900 2,030 1,910 2,000 1,970 1,930 18 2,000 1,950 50.1 2.58

5,950 6,160 5,840 5,930 5,950 6,180 6,010 5,950 6,130 5,770 18 6,000 6,000 129 2.14

14,900 14,600 15,200 14,500 14,600 14,700 14,300 14,500 13,800 14,100 18 15,000 14,700 382 2.60

−4.33

−2.97

−2.34

−2.73

Incurred sample reproducibility From an ongoing anidulafungin study, 47 samples (10 % of the total number of samples analyzed) were reassayed for ISR. Samples were chosen over the range of concentrations from Cmax to the terminal elimination phase in the PK profile. All 47 samples confirmed their original concentration within the ±20 % limits, with 43 out of 47 confirming the original concentration within the ±15 % limits. The ISR results met the acceptance recommendations for incurred samples from Crystal City III Workshop [13]. D1 in incurred anidulafungin plasma samples The open-ring form of anidulafungin, D1, was not available as a pure standard substance and, therefore, it was prepared as described under “Preparation of standards” section. Damle et al. [11] had used a similar procedure to generate D1 and estimated its concentrations in vitro and in vivo. They compared the disappearance of anidulafungin and formation of D1 in phosphate-buffered saline at pH 8 during 96 h at ambient temperature. Assuming that D1 was stable in the incubation mixture and comparing the peak areas of

0.0386

−1.95

anidulafungin and D1, they estimated the difference in MS response sensitivity between the two compounds; D1 had about 1/4 of the response of anidulafungin [11]. In our hands and in similar experimental conditions, D1 showed approximately only about 1/8 of the MS response of anidulafungin. Most likely, D1 is identical to the anidulafungin degradant observed recently by another group of authors developing a TDM assay for anidulafungin [8]. D1 is not an active form and our aim was not to quantify it in the incurred samples; our aim was to ensure that its presence was not interfering with the quantification of anidulafungin. Therefore, we used it as an external marker or system suitability sample with the sole purpose of confirming its retention time and assuring its base line separation from anidulafungin.

Conclusions A rugged, accurate, precise and reproducible LC-MS/MS method for the analysis of anidulafungin in human plasma was validated in a linear range from 50 to 20,000 ng/mL.

Anidulafungin

This method separates the isobaric open-ring form of anidulafungin (D1) from anidulafungin. It requires a 50-μL aliquot of human plasma and is suitable for pediatric studies where limited volume of blood for PK sampling is available. The anidulafungin frozen plasma stability was confirmed for 740 days at both −70 and −20 °C. The method is presently used for analysis of anidulafungin in plasma samples from several clinical studies. Acknowledgment The bioanalytical assay described in this manuscript was developed in collaboration with PPD and validated by PPD. This work was funded by Pfizer Inc. Michael Modesitt is a full-time employee of PPD. Tanja Alebic-Kolbah is a full-time employee of Pfizer. The authors would like to thank Ping Liu from Pfizer for her engagement and help during procedure development for PK sample collection and manuscript preparation. Gregory S Walker, Sharon Ripp, and Gregory L Weber from Pfizer are acknowledged for valuable contribution and discussion related to the anidulafungin open-ring form. Generous help in editing and peer-reviewing of the manuscript by Ileana Ionita from Pfizer is also acknowledged and highly appreciated. Thanks are due to Nada Kurt Stojkovic for reviewing the manuscript.

References 1. Raasch RH (2004) Anidulafungin: review of a new echinocandin antifungal agent. Expert Rev Anti Infect Ther 2(4): 499–508 2. Dowell JA, Knebel W, Ludden T, Stogniew M, Krause D, Henkel T (2004) Population pharmacokinetic analysis of anidulafungin, an echinocandin antifungal. J Clin Pharmacol 44(6):590–598. doi:10.1177/009127000426564444/6/590 3. Dowell JA, Schranz J, Baruch A, Foster G (2005) Safety and pharmacokinetics of coadministered voriconazole and anidulafungin. J Clin Pharmacol 45(12):1373–1382. doi:10.1177/ 0091270005281234 4. Crandon JL, Banevicius MA, Fang AF, Crownover PH, Knauft RF, Pope JS, Russomanno JH, Shore E, Nicolau DP, Kuti JL (2009) Bronchopulmonary disposition of intravenous voriconazole and anidulafungin given in combination to healthy adults. Antimicrob Agents Chemother 53(12):5102–5107. doi:10.1128/ AAC.01042-09

2055 5. Sutherland CA, Nicolau DP, Kuti JL (2011) Development of an HPLC method for the determination of anidulafungin in human plasma and saline. J Chromatogr Sci 49(5):397–400 6. Farowski F, Cornely OA, Vehreschild JJ, Hartmann P, Bauer T, Steinbach A, Ruping MJ, Muller C (2010) Quantitation of azoles and echinocandins in compartments of peripheral blood by liquid chromatography–tandem mass spectrometry. Antimicrob Agents Chemother 54(5):1815–1819. doi:10.1128/AAC.01276-09 7. Decosterd LA, Rochat B, Pesse B, Mercier T, Tissot F, Widmer N, Bille J, Calandra T, Zanolari B, Marchetti O (2010) Multiplex ultra-performance liquid chromatography–tandem mass spectrometry method for simultaneous quantification in human plasma of fluconazole, itraconazole, hydroxyitraconazole, posaconazole, voriconazole, voriconazole-N-oxide, anidulafungin, and caspofungin. Antimicrob Agents Chemother 54(12):5303–5315. doi:10.1128/ AAC.00404-10 8. Martens-Lobenhoffer J, Rupprecht V, Bode-Boger SM (2011) Determination of micafungin and anidulafungin in human plasma: UV- or mass spectrometric quantification? J Chromatogr B Analyt Technol Biomed Life Sci 879(22):2051–2056. doi:10.1016/ j.jchromb.2011.05.033 9. Balkovec JM, Black RM, Hammond ML, Heck JV, Zambias RA, Abruzzo G, Bartizal K, Kropp H, Trainor C, Schwartz RE et al (1992) Synthesis, stability, and biological evaluation of watersoluble prodrugs of a new echinocandin lipopeptide. Discovery of a potential clinical agent for the treatment of systemic candidiasis and Pneumocystis carinii pneumonia (PCP). J Med Chem 35 (1):194–198 10. Norris T, VanAlsten J, Hubbs S, Ewing M, Cai W, Jorgensen ML, Bordner J, Jensen GO (2008) Commercialization and late-stage development of a semisynthetic antifungal API: anidulafungin/Dfructose (Eraxis). Org Process Res Dev 12(3):447–455 11. Damle BD, Dowell JA, Walsky RL, Weber GL, Stogniew M, Inskeep PB (2009) In vitro and in vivo studies to characterize the clearance mechanism and potential cytochrome P450 interactions of anidulafungin. Antimicrob Agents Chemother 53(3):1149– 1156. doi:10.1128/AAC.01279-08 12. US Department of Health and Human Services. Food and Drug Administration (FDA). Center for Drug Evaluation and Research (CDER). Guidance for Industry. Bioanalytical Method Validation. http://www.fda.gov/cder/guidance/4252fnl.htm. Accessed December 2008 13. Fast DM, Kelley M, Viswanathan CT, O’Shaughnessy J, King SP, Chaudhary A, Weiner R, DeStefano AJ, Tang D (2009) Workshop report and follow-up—AAPS workshop on current topics in GLP bioanalysis: assay reproducibility for incurred samples—implications of Crystal City recommendations. AAPS J 11(2):238–241. doi:10.1208/s12248-009-9100-9