Journal of Chromatography B, 969 (2014) 117–122
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Development and validation of a HILIC–MS/MS method for quantification of decitabine in human plasma by using lithium adduct detection Wenyi Hua ∗ , Thomas Ierardi, Michael Lesslie, Brian T. Hoffman, Daniel Mulvana Bioanalytical and ADME Labs, Quintiles, 19 Brown Road, Ithaca, NY 14850, USA
a r t i c l e
i n f o
Article history: Received 18 February 2014 Accepted 7 August 2014 Available online 15 August 2014 Keywords: Lithiated adducts Alkali adducts HILIC Tandem mass spectrometry (MS/MS) Anomerizaton Decitabine
a b s t r a c t A highly sensitive, selective, and rugged quantification method was developed and validated for decitabine (5-aza-2 -deoxycytidine) in human plasma treated with 100 g/mL of tetrahydrouridine (THU). Chromatographic separation was accomplished using hydrophilic interaction liquid chromatography (HILIC) and detection used electrospray ionization (ESI) tandem mass spectrometry (MS/MS) by monitoring lithiated adducts of the analytes as precursor ions. The method involves simple acetonitrile precipitation steps (in an ice bath) followed by injection of the supernatant onto a Thermo Betasil Silica100, 100 × 3.0 mm, 5 m LC column. Protonated ([M+H]+ ), sodiated ([M+Na]+ ), and lithiated ([M+Li]+ ) adducts as precursor ions for MS/MS detection were evaluated for best sensitivity and assay performance. During initial method development abundant sodium [M+Na]+ and potassium [M+K]+ adducts were observed while the protonated species [M+H]+ was present at a relative abundance of less than 5% in Q1. The alkali adducts were not be able to be minimized by the usual approach of increasing acid content in mobile phases. Significant analyte/internal standard (IS) co-suppression and inter-lot response differences were observed when using the sodium adduct as the precursor ion for quantification. By adding 2 mM lithium acetate in aqueous mobile phase component, the lithium adduct effectively replaced other cationic species and was successfully used as the precursor ion for selected reaction monitoring (SRM) detection. The method demonstrated the separation of anomers and from other endogenous interferences using a 3-min gradient elution. Decitabine stock, working solution stabilities were investigated during method development. Three different peaks, including one from anomerization, were observed in the SRM transition of the analyte when it was in neutral aqueous solution. The assay was validated over a concentration range of 0.5–500 ng/mL (or 0.44–440 pg injected on column) in 50 L of human plasma. The accuracy and precision were within 8.6% relative error and 6.3% coefficient of variation, respectively. Decitabine was stable in THU treated human plasma for at least 68 days and after 5 freeze–thaw cycles when stored at −70 ◦ C. Stability of decitabine in THU treated human whole blood, matrix factor and recovery were also evaluated during method validation. The method was successfully used for clinical sample analysis. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Decitabine (5-aza-2 -deoxycytidine, Fig. 1) is a drug used in the treatment of patients with myelodysplastic syndromes and has demonstrated promising activity in the treatment of acute leukemia, in particular acute myeloid leukemia [1,2]. At high doses, decitabine exerts primarily a cytotoxic effect, but at lower doses its hypomethylating activity predominates, resulting in gene
∗ Corresponding author. Tel.: +1 607 330 9831. E-mail addresses: [email protected], [email protected] (W. Hua). http://dx.doi.org/10.1016/j.jchromb.2014.08.012 1570-0232/© 2014 Elsevier B.V. All rights reserved.
re-expression and cellular differentiation of leukemic cells [3]. However, the pharmacokinetic (PK) profile of decitabine at lower doses (e.g., 15 mg/m2 ) in humans remained elusive until recently, in partially due to challenges of quantifying and characterizing decitabine in biological matrices [2,4,5]. Decitabine is unstable at physiological temperature and pH. The degradation of decitabine results a plethora of products, formed by hydrolytic opening, deformylation of the triazine ring, anomerization, and possibly other changes in the sugar ring structure [4–6]. Additionally, it is susceptible to deamination by high levels of cytidine deaminase, as occurs with other nucleoside analogs in human plasma [1]. The pharmacological and toxicological properties of the
118
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
NH2
13
C N
N
HC HO CH2 O HC
C
CH HO
Corporation (PA, USA). Human whole blood (K2 EDTA) was collected from Quintiles volunteers in accordance with current company policies on informed consent and ethical approval.
15
NH2
N CH
N
15
N
13
O
H C HO CH2 O HC CH
CH2
A
C
15
HO
15
N
2.2. Preparation of stock solutions, calibration standards, and quality controls (QCs)
C O
CH CH2
B
Fig. 1. Chemical structures of (A) decitabine (-anomer), C8 H12 N4 O4 , monoisotopic mass: 228.09; (B) IS [13 C2 , 15 N4 ]-decitabine (mixture of ␣- and -anomer), 13 C2 C6 15 N4 H12 O4 , monoisotopic mass: 234.09.
degradation products are yet largely unknown. Decitabine is also a highly polar compound with log P of approximately −2.16 [7]. Due to its high hydrophilicity, it is difficult to retain on a reversedphase (RP) column unless highly aqueous mobile phase eluents are used, resulting in lower sensitivity in LC/MS detection. Postcolumn addition of organic solvent to enhance ionization after RP LC separation was reported [4]. HILIC with unique selectivity has been demonstrated as an alternative separation technique for small polar compounds [8,9] including several anticancer drugs [10]. In addition, high content of organic solvent, typically acetonitrile, can provide increased sensitivity for mass spectrometric detection. A few analytical methods have been reported for quantifying decitabine in biological matrices. Early methods based on L1210 cell kills [11] and LC–UV techniques [12,13] suffered from lack of sensitivity and selectivity and could not be applied for PK characterization at lower doses. In recent years, more sensitive methods have been reported using protein precipitation (PPT) or mixedmode cation exchange (MCX) solid-phase extraction (SPE) coupled with RP-LC–MS/MS with lower limits of quantification (LLOQs) in a range of 1–10 ng/mL (or as low as 14 pg injected on-column) in rat or human plasma [3,4,14,15]. These methods used the protonated molecule as precursor ion for MS/MS detection, except Xu et al. [14] who reported use of sodiated adducts for MS/MS detection. In this study, we developed and validated a highly sensitive and selective HILIC–MS/MS method to analyze decitabine in human plasma by using lithiated adducts. To stabilize decitabine in human whole blood or plasma, cytidine deaminase inhibitor THU at 100 g/mL was added into samples as previously described [15]. With a simple PPT, a LLOQ of 0.5 ng/mL (or 0.44 pg injected on-column) in 50 L of human plasma was easily achieved. 2. Experimental 2.1. Chemical and reagents Decitabine (5-aza-2 -deoxycytidine, >97%) was purchased from Sigma–Aldrich (MO, USA). [13 C2 , 15 N4 ]-decitabine (mixture of anomers, >98.6%, Fig. 1) used as IS was purchased from Alsachim (Strasbourg, France). Acetonitrile, methanol, water were HPLC grade and were obtained from Honeywell Burdick & Jackson (MI, USA). Dimethylformamide (DMF), dimethyl sulfoxide (DMSO), ammonium formate, and lithium acetate dihydrate were purchased from Sigma–Aldrich (MO, USA). THU was purchased from EMD Millipore (MA, USA). Formic acid was purchased from EMD Chemicals Inc. (NJ, USA). Trifluoroacetic acid (TFA) was purchased from Fisher Scientific (PA, USA). Human plasma treated with K2 EDTA was obtained from Bioreclamation (NY, USA) or Biological Specialty
Stock solutions of decitabine and [13 C2 , 15 N4 ]-decitabine were prepared at a concentration of 0.5 mg/mL in DMF and stored at −20 ◦ C. Standard working solutions (SWS) at 20,000 ng/mL of decitabine in DMF were used for the preparation of calibration curves and QCs. Control human plasma used for this study was treated with THU at a concentration of 100 g/mL. Spiked calibration standards in control human plasma were prepared from serial dilutions at the following concentrations: 0.5, 1.0, 5.0, 50.0, 100, 250, 425, and 500 ng/mL. The QCs were spiked at the following levels: 0.5 (LLOQ), 1.5 (QC1), 15 (GMQC1), 150 (QC2), and 400 ng/mL (QC3). A 20-fold dilution QC was prepared at a concentration of 8000 ng/mL. Internal standard working solution (ISWS) was prepared in DMF at 100 ng/mL and stored at −20 ◦ C. 2.3. Sample preparation Sample preparation was performed in an ice bath (equivalent to 4 ◦ C). Protein precipitation in 1 mL 96-well plates was used for sample preparation. Briefly, 50 L aliquots of plasma samples were mixed with 20 L of ISWS and followed by a 500 L addition of acetonitrile. The plate was vortexed and centrifuged for 5 min. Supernatant was transferred into a clean 1 mL 96-well plate and a volume of 10 L was injected onto LC–MS/MS instrument. 2.4. Instrument analysis The HPLC system was equipped with an SCL-10A pump controller, two LC-10AD pumps (Shimadzu, MD, USA) and PAL autosampler (LEAP Technologies, NC, USA). The HPLC system was interfaced to an API5000 tandem mass spectrometer (Applied Biosystems-MDS Sciex, ON, Canada). The chromatographic separation was performed on a Thermo Betasil Silica-100, 100 × 3.0 mm, 5 m HPLC column (Thermo Scientific, MA, USA) at ambient temperature. The mobile phases (MP) used consisted of 100:1:0.5:0.4 water/1 M ammonium formate in water/formic acid/0.5 M lithium acetate in water as MP A and acetonitrile as MP B. The elution gradient was from 90% B initial to 50% B in 3 min at 0.5 mL/min flow rate. The LC column was flushed with 40% B at 1.8 mL/min and equilibrated to the initial condition. Water and 90:10 acetonitrile/water were used for autosampler washes to minimize carryover. The LC eluent was introduced to the mass spectrometer using the ESI interface in positive ion mode. The ion spray voltage, turbo heater temperature, and declustering potential were set at 2500 V, 650 ◦ C, and 60 V, respectively. The quantification was performed using SRM detection with the transitions of m/z 235.1→119.1 for decitabine and m/z 241.1→125.0 for IS. The optimized collision energy was at 22 eV for both decitabine and IS. Due to the extensive signal of decitabine and limited amount of - in IS, a collision energy of 30 eV was used for detuning decitabine while the optimized value was used for the IS during validation. 2.5. Method validation To validate the method, tests of the selectivity, linearity, intraand inter-assay accuracy and precision, sample dilution, extraction recovery, matrix effects, freeze/thaw and long-term stability in plasma samples as well as working solution and stock solution stability were conducted in compliance with the Food and Drug Administration guidelines [16].
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
251.1 [M+Na]+
100%
Max. 2.9e6 cps
Relative Ion Abundance (%)
90% 80% 70% 60% 50% 40% 30% 20% 252.1
10%
229.2 [M+H]+
200
220
210
230
240
250 m/z
260
266.9 [M+K]+ [M+ACN+Na]+ 292.2 270
280
290
300
A
235.1 [M+Li]+
100%
Max. 2.7e6 cps
Relative Ion Abundance (%)
90% 80% 70% 60% 50% 40% 30%
236.2 234.3
20% 10% 200
220
210
230
240
250 m/z
260
270
280
290
300
B
Fig. 2. Q1 and MS/MS full scan spectra of decitabine (A) without lithium acetate added in MP A; (B) with 2 mM lithium acetate added in MP A. Q1 full scan spectra were acquired by injection with gradient elution described in Section 2.4.
The selectivity was verified using six individual blank human plasma samples. Whether any endogenous peaks were interfering with the peak of decitabine or IS in the SRM chromatograms was investigated using blank plasma samples without spiking the analyte and IS.
2.0e6
Self-suppression across the curve
↑
Inter-lot variability
↑
IS Peak Area (counts)
1.8e6 1.6e6 1.4e6 1.2e6 1.0e6 8.0e5 6.0e5
Calibration Standard (1 Matrix Lot) Inter-lot Zero Samples (6 Matrix Lots) QC Samples (2 Matrix Lots) Control Blank or Zero Sample (1 Matrix Lot)
4.0e5 2.0e5 0 5
15
25
35
45
55
65
75
85
95
Injection Number Fig. 3. IS response plot when using [M+Na]+ as precursor ion with MRM transition 251.1→135.0 for decitabine and 257.1→141.0 for IS. MPA: 100:1:0.5 water/1 M ammonium formate in water/formic acid. MP B: acetonitrile.
119
The intra- and inter-assay accuracy and precision were evaluated using six replicates of QC samples at the concentrations described in Section 2.2. Long term storage stability was evaluated at −20 ◦ C and −70 ◦ C for up to 68 days, 5 freeze/thaw-cycle (−20 ◦ C, −70 ◦ C) and bench-top up to 29 h at 4 ◦ C were evaluated for stability of decitabine in THU treated human plasma with QC1, QC3, and dilution QC. The processed sample stability for decitabine was determined by quantifying the extracted, stored QC samples (QC1, QC2, and QC3) up to 173 h at 4 ◦ C against freshly extracted standard curves. In the reproducibility and stability tests, the mean accuracy was expressed as RE (%) and the precision expressed as coefficient of variation (CV, %). In addition to the stability of decitabine in THU treated plasma, the stability of decitabine in THU treated (100 g/mL) whole blood was evaluated by spiking decitabine at a concentration of 250 ng/mL. The baseline plasma sample (T = 0) was harvested by centrifugation of the whole blood sample after 10-min incubation and immediately frozen at −20 ◦ C. This sample was compared to samples harvested after storage of whole blood for 1 h (T = 1 h) at ambient and in an ice bath. In this study, the control matrix used for analytical and stability QC samples described above were prepared by adding THU to plasma and then spiking decitabine. Since the amount of THU that ends up in the plasma from clinical study samples cannot be determined, additional QC samples (QC1, QC3, and dilution QC) were prepared in a manner to more closely simulate study samples where whole blood was collected in tubes containing THU at 100 g/mL and then spun down to plasma followed by spiking decitabine. These QCs were stored at −20 ◦ C and −70 ◦ C for 0 and 48 days for additional long-term storage stability. The extraction recoveries were measured by comparing the peak area ratios of decitabine QCs at all four concentrations (QC1, GMQC1, QC2, QC3) in samples spiked after extraction (post-extract) with those of samples spiked before extraction (pre-extract). Matrix effects (matrix factor) were determined by comparing the peak areas of analyte in spiked plasma extracts at all four QC concentration levels to those in plasma-free extracts in six different sources of plasma.
3. Results and discussion 3.1. Optimization of chromatographic condition and sample preparation Both RP and HILIC LC separations were evaluated during method development. Several LC–MS methods for decitabine were published [3,4,12–15] but none of them employed HILIC separation. Since decitabine is a strong hydrophilic compound, reversed phase C18 and C8 columns which are marketed as suitable for analysis of polar compounds were initially tested. Decitabine was poorly retained on these columns unless using very high percentage of aqueous phase, and suffered from poor selectivity and sensitivity. HILIC columns, including Betasil silica, 5 m (100 × 3.0 mm, 50 × 2.1 mm), Kinetex HILIC 2.6 m (50 × 2.1 mm), and Atlantis Silica HILIC 5 m (2.1 × 50 mm) were then evaluated. Betasil silica 5 m (100 × 3.0 mm) was chosen for the method based on optimal retention time (RT), selectivity, and peak shape. Ammonium formate, ammonium acetate with or without acid were evaluated as aqueous MP additives for best sensitivity, selectivity and chromatographic reproducibility. Ammonium formate and ammonium acetate without pH adjustment both improved peak shape and separation relative to use of water. Increased sensitivity was achieved when ammonium formate and formic acid were used. The stable isotope-labeled IS used in our method was a mixture of anomers with ␣-anomer the majority. Decitabine, however is
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
3.3. Stock, working solutions stability Stock solution, SWS stabilities were evaluated in different solvents during method development. Stock solutions of 0.5 mg/mL of were dissolved in DMSO, DMF, and 50:50 methanol/water. After 18 days of storage at −20 ◦ C, there was 33% decrease for the stock solution prepared in 50:50 methanol/water and no sign of degradation for those prepared in DMSO or DMF. DMF, 0.1 N HCL in
1.2e6 8.0e5 4.0e5 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
A
2.17
790 Decitabine 600 400 200 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
B Intensity (cps)
3.2. Optimization of mass spectrometric conditions
2.17 β-anomer
7.0e5 Decitabine 5.0e5
2.30 α-anomer 2.46
3.0e5 1.0e5 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
C
2.5e5
Intensity (cps)
The protonated molecule m/z 229 was mostly used as the precursor ion in previously published methods. Xu et al. [14] reported formation of mainly sodium adducts of decitabine in full-scan positive ESI MS, which were used as precursor ions for quantification of decitabine in rat plasma. In our study, we also found that decitabine formed predominantly sodium adducts at m/z 251.1, and potassium adducts at m/z 267.1 (Fig. 2A). Initial effort was focused on suppression of sodium and potassium adduction by increasing acid content in aqueous MP. Formic acid (0.5%, 2.5%, and 5%) and TFA (0.5%, 1%, and 2.5%) in aqueous MP were evaluated for minimizing adduct formation. However, sodium adducts still predominated at the highest formic acid and TFA concentrations and no significant increase of protonated ions in Q1 was observed. Quantitation using sodium adducts (SRM transition m/z 251.1→135.0 for decitabine and m/z 257.1→141.0 for IS) was evaluated. The sodium adduction was maintained in the collision-induced dissociation (CID) fragmentation. Significant analyte/IS co-suppression, and inter-lot response differences were observed even though accuracy and precision were acceptable. Fig. 3 shows the IS responses from a 96-injection run that included 6 inter-lot LLOQs, all levels of QCs and standard curves when using [M+Na]+ as precursor ion for quantitation. The QCs and standard curves were prepared in two different sources of human plasma. The IS response across the run had 31% of precision, which is unexpectedly high for a PPT method. Since lithium adduction has been previously shown as an effective means of ionizing weakly basic compounds [17,18], it was considered a possible means of displacing the sodium and potassium adducts. Addition of 2 mM of lithium acetate to the aqueous component of MP effectively formed entirely [M+Li]+ in Q1. The most abundant product ion (m/z 119.0) (Fig. 2B) for decitabine and m/z 125.0 for IS) was used for quantitation. These species maintained the lithium adduction in the CID fragmentation. When the same run was re-injected with 2 mM lithium acetate MP a consistent IS response was achieved in SRM detection with 7% of precision.
2.13 β-anomer
1.6e6 Decitabine
Intensity (cps)
the -anomer. It was found that improved quantitation (better precision) was achieved if the IS anomers were chromatographically separated and only the -anomer was used for quantitation. Both PPT and cation exchange SPE were investigated for sample preparation. For SPE, Oasis MCX 96-well plate (10 mg) and Evolute CX 96-well plate (25 mg) were evaluated using 50 L sample aliquots. The recoveries from both SPE plates were excellent and were consistent with previously reported data [4]. PPT using acetonitrile without drying down step was used due to its simplicity and cost-effectiveness, and most importantly, due to high sensitivity achieved by HILIC separation. Sample preparation in an ice bath is necessary due to insufficient stability of analytes in the matrix. Bench-top stability of spiked QCs was evaluated during initial method development at two concentrations (QC1 and dilution QC). After 6 h of storage at ambient temperature, 18% of degradation was observed for dilution QC. After 20 h of storage at ambient temperature, 37% and 63% of degradation was observed for QC1 and dilution QC, respectively. No significant degradation was observed for the same QCs after 20 h of storage at 4 ◦ C.
Intensity (cps)
120
2.27 α-anomer
[13C215N4]Decitabine
1.5e5 β-anomer 2.14 5.0e4 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
D
Fig. 4. Chromatograms of 20,000 ng/mL decitabine in (A) DMF; (B) 0.1 N HCl(aq); (C) water stressed at 37 ◦ C for 17 h; (D) [13 C2 , 15 N4 ]-decitabine (mixture of ␣- and -anomer).
water, and water were evaluated as diluents for SWS. To enable quick evaluation, SWS prepared in above diluents were stored at −20 ◦ C, ambient, and at 37 ◦ C for 17 h. These solutions were then compared to the ones that were freshly prepared. Decitabine prepared in DMF was stable under all the storage condition including at 37 ◦ C for 17 h. Degradation of decitabine in 0.1 N HCl in water, and water was large dependant on storage temperature (see Supplementary Table S1). No significant degradation was observed when the solutions were stored at −20 ◦ C, while after storage at 37 ◦ C for 17 h, 100% and 59% degradation was observed for 0.1 N HCl in water, and water, respectively. Degradation products were different in 0.1 N HCl and water based on the chromatographic evidence. Fig. 4A–C presents chromatograms of decitabine in DMF, 0.1 N HCl in water, and water after storage at 37 ◦ C for 17 h. A total of four peaks (RT = 1.94, 2.17, 2.30, 2.46 min) in the same SRM transition of decitabine (m/z 235.1→119.0) were identified for decitabine in water that was stress at 37 ◦ C for 17 h (Fig. 4C). Rogstad et al. [5] proposed that at neutral pH in aqueous phase, both base-catalyzed events (hydrolytic ring-opening) and acid-catalyzed events (sugar ring-opening followed by anomerization and reclosure) can take place simultaneously for decitabine, resulting ␣- and -anomers
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
Intensity (cps)
8000 2.44 α-anomer
4000 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
Intensity (cps)
2.30 β-anomer
1.15e4 Decitabine
3000
121
1.58
Decitabine 0.99
2000 1000 0
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
5000
D 2.44 α-anomer
Decitabine
3000 2000 1000 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
0.98
2000 1000 0
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2. 2 2 .4 2.6 2.8 Time (min)
E 2.21 α-anomer
2.30 β-anomer Intensity (cps)
Decitabine Intensity (cps)
2.08 β-anomer
3000
B 2000 1000 0
1.58
4000 Decitabine
4000
Intensity (cps)
Intensity (cps)
A
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
1.2e5 8.0e4 β-anomer 2.08 4.0e4 0
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
F
C
Fig. 5. Chromatograms of extracted (A) clinical study sample T = 2 h; (B) clinical study sample T = 3 h; (C) long-term stability QC1 (after 48 days at −70 ◦ C); (D) control blank; (E) LLOQ standard; (F) [13 C2 , 15 N4 ]-decitabine (mixture of ␣- and -anomer). Note: Retention time of A, B, C is later than that of D, E, F since they are from different analysis batches. See insert for IS transition of A, B, C.
of furanoside and pyranoside derivatives. In his study, only two peaks at m/z 229 proposed as ␣- and -anomers of decitabine were identified by LC–UV and QTOF/MS from stressed (37 ◦ C for 24 h) decitabine in 100 mM potassium phosphate (pH 7.4). Our results confirmed anomerization since one of the peaks eluted at the same RT of ␣-anomer of the IS. This was also observed in the study samples (Fig. 5A and B) but not in the long-term stability QCs (Fig. 5C) that were prepared in a manner closely simulating study samples (see Section 2.5). To our knowledge, detection of peaks eluting at RT = 1.94, 2.46 min as degradation products have not been reported. Since these two peaks had same product ions in SRM detection, they are likely the ␣- and -anomers of pyranoside derivatives with triazine ring (formed by reclosure of hydrolytic ring-opening) as proposed by Rogstad et al. [5]. Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jchromb.2014.08.012.
3.4. Method validation
Table 1 Accuracy and precision of decitabine in THU treated human plasma from three validation runs.
Treated plasma lot (n = 6)
Decitabine concentration (ng/mL)
0.5 (LLOQ) 1.5 (QC1) 15 (GMQC1) 150 (QC2) 400 (QC3) 8000 (dilution QC, 20-fold)
Intra-assay (n = 6)
Inter-assay (n = 18)
RE (%)
CV (%)
RE (%)
CV (%)
8.6 4.7 0.7 0.7 −1.8 −2.6
4.9 2.8 1.3 1.6 1.6 0.6
6.4 5.3 1.3 0 −2.3 −1.1
6.3 3.0 1.3 1.3 1.8 1.9
3.4.1. Selectivity Fig. 5D and E shows the typical chromatograms of a blank plasma sample and a blank plasma sample spiked with decitabine at the LLOQ. No chromatographic interferences were detected in six different sources of plasma that were evaluated. 3.4.2. Linearity and LLOQ Standard curves were performed in duplicate in THU treated plasma. In all cases the regression coefficient was >0.997. Decitabine was linear over the range of 0.5–500 ng/mL with a weighting of 1/x2 . All coefficients of determination (r2 ) were ≥0.998.
Table 2 Inter-lot precision and accuracy for decitabine. Decitabine concentration (ng/mL) LLOQ (0.5)
ULOQ (500)
1 2 3 4 5 6
0.493 0.474 0.500 0.469 0.556 0.542
494 484 477 485 479 490
Mean CV (%) RE (%)
0.506 7.1 1.2
485 1.3 −3.0
122
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
Table 3 Stability of decitabine in THU treated (100 g/mL) human plasma under various storage conditions. Plasma concentration (ng/mL)
1.5 (QC1) 150 (QC2) 400 (QC3) 8000 (dilQC4)
Long-term storage (n = 4)
Five freeze–thaw cycles (n = 4)
4 ◦ C bench-top (n = 4)
Processed sample at 4 ◦ C (n = 8)
-20 ◦ C (21 days)
-70 ◦ C (68 days)
-20 ◦ C
29 h
173 h
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
−6.0 NA −9.0 −8.9
4.0 NA 1.3 0.3
4.0 NA −1.8 −1.3
2.4 NA 2.2 1.5
-2.7 NA −11.3 −10.0
3.8 NA 1.0 2.1
−0.7 NA −6.5 −6.6
3.6 NA 2.3 3.0
−6.7 NA −12.5 −16.0
2.9 NA 1.8 1.7
−2.0 −6.0 −9.5 NA
3.1 1.5 2.7 NA
3.4.3. Precision and accuracy Table 1 summarizes the results for intra- and inter-day precision and accuracy for decitabine measured by QCs. The intra- and interday precision were all below 15% with a maximum CV % of 6.3% and a maximum bias (RE %) of 8.6% for accuracy was calculated. 3.4.4. Recovery and matrix effect The recoveries of decitabine extracted from THU treated plasma were 69.6 ± 11.1%, 73.2 ± 3.3%, 77.0 ± 2.6%, and 77.0 ± 2.2% at the concentrations of 1.5, 15, 150, and 400 ng/mL, respectively (n = 6). The recovery of IS was 80.8 ± 1.6% at concentration of 100 ng/mL (n = 6). The matrix factor of decitabine from six different sources of human plasma samples at concentrations of 1.5, 15, 150, and 400 ng/mL (n = 6) were in the range of 103–112% with CV% below 10%, indicating no matrix-dependent quantitation issues. In addition, 6 different lots of plasma were spiked with analyte at the LLOQ and ULOQ with resulting CV ≤ 7.1% and mean bias within ±3% (Table 2). 3.4.5. Effect of hemolyzed and lipemic plasma To evaluate whether real world sample conditions would quantify accurately, plasma containing 2% and 5% lysed blood as well a plasma containing high triglyceride levels were spiked with analyte at concentrations of 1.5 and 400 ng/mL. Acceptable quantitation was achieved in all cases (see Supplementary Table S2). Supplementary Table S2 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jchromb.2014.08.012. 3.4.6. Stability Decitabine in THU treated human plasma at 1.5, 400, and 8000 ng/mL was found stable at -20 ◦ C for at least 21 days, at -70 ◦ C for at least 68 days, after 5 freeze–thaw cycles when stored at 20 ◦ C or -70 ◦ C, at approximately 4 ◦ C (bench-top, equivalent to an ice bath) under yellow light after 29 h (Table 3) although some loss is evident in the bench-top samples. This loss may be a function of the aqueous instability rather than due to cytidine deaminase activity, since the latter would be expected to be suppressed by THU. Decitabine in processed samples was stable at 4 ◦ C for at least 173 h. The stock solution at 0.5 mg/mL and SWS at 20,000 ng/mL in DMF were stable for at least 49 days at -20 ◦ C and at least 30 h at 4 ◦ C. Decitabine was stable in THU treated human whole blood after storage for 1 h at ambient and in an ice bath (Supplementary Table S3). For the QC samples (1.5, 400, and 8000 ng/mL) prepared in the control plasma that were centrifuged from THU treated human whole blood, stability was established for decitabine for up to 48 days at -70 ◦ C (Fig. 4C, data not shown) but not at -20 ◦ C. Supplementary Table S3 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jchromb. 2014.08.012.
-70 ◦ C
4. Conclusion A fast, sensitive and rugged HILIC–MS/MS method has been developed and validated for the analysis of decitabine in THU treated human plasma. With a simple step of protein precipitation, our method achieved 0.5 ng/mL LLOQ with 50 L sample volume (or 0.44 pg on-column). The use of lithium adducts enabled reproducible analysis in multiple sources of matrix while maintaining sensitivity requirements for the analysis of clinical samples. Further improvements in sensitivity could be gained based on our preliminary evaluation of MCX SPE extraction should it be required (e.g. reduced sample volumes in pediatric samples). However, the simplicity of PPT extraction enables low cost and reproducible sample analysis while maintaining high sensitivity. The ability to separate anomers is important since ␣-anomer was observed in clinical study samples. Acknowledgements The authors would like to acknowledge J. Johnson, G. Lal, C. Jordaens, B. Beckhorn, M. Mann, A. Drelick, J. Urda, C. Tilley, C. Reeves, E. King, E. Button, M. Welser for sample preparation and LC/MS analysis. S. Swift-Spencer is thanked for formatting figures and tables. References [1] E. Jabbour, J.-P. Issa, G. Garcia-Manero, H. Kantarjian, Cancer 112 (2008) 2341–2351. [2] J. Bryan, H. Kantarjian, G. Garcia-Manero, E. Jabbour, Expert Opin. Drug Metab. Toxicol. 7 (5) (2011) 661–672. [3] A.F. Cashen, A.K. Shah, L. Todt, N. Fisher, J. DiPersio, Cancer Chemother. Pharmacol. 61 (2008) 759–766. [4] Z. Liu, G. Marcucci, J.C. Byrd, M. Grever, J. Xiao, K.K. Chan, Rapid Commun. Mass Spectrom. 20 (2006) 1117–1126. [5] D.K. Rogstad, J.L. Herring, J.A. Theruvathu, A. Burdzy, C.C. Perry, J.W. Neidigh, L.C. Sowers, Chem. Res. Toxicol. 22 (2009) 1194–1204. [6] K.T. Lin, R.L. Momparler, G.E. Rivard, J. Pharm. Sci. 70 (1981) 1228–1232. [7] http://www.chemicalize.org/structure/#!mol=Dacogen [8] W. Naidong, J. Chromatogr. B 796 (2003) 209–224. [9] P. Hemström, K. Irgum, J. Sep. Sci. 29 (2006) 1784–1821. [10] S. Nussbaumer, P. Bonnabry, J.-L. Veuthey, S. Fleury-Souverain, Talanta 85 (2011) 2265–2289. [11] G.G. Chabot, G.E. Rivard, R.L. Momparler, Cancer Res. 43 (1983) 592–597. [12] J.M. Covey, D.S. Zaharko, Eur. J. Cancer Clin. Oncol. 21 (1985) 109–117. [13] K.T. Lin, R.L. Momparler, G.E. Rivard, J. Chromatogr. 345 (1985) 162–167. [14] H. Xu, S. Lv, M. Qiao, Y. Fu, X. Jiang, Y. Jin, C. Li, B. Yuan, J. Chromatogr. B 899 (2012) 81–85. [15] K. Patel, S.M. Guichard, D.I. Jodrell, J. Chromatogr. B 863 (2008) 19–25. [16] Guidance for Industry Bioanalytical Method Validation, U.S. Department of Health and Human Services, Food and Drug Administration Centre for Drug Evaluation and Research (CDER), Centre for Veterinary Medicine (CM), May 2001, http://www.fda.gov/cder/guidance/4252fnl.htm [17] J. Adams, M.L. Gross, J. Am. Chem. Soc. 108 (1986) 6912–6915. [18] B.M. Ham, J.T. Jacob, M.M. Keese, R.B. Cole, J. Mass Spectrom. 39 (2004) 1321–1336.
Contents lists available at ScienceDirect
Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb
Development and validation of a HILIC–MS/MS method for quantification of decitabine in human plasma by using lithium adduct detection Wenyi Hua ∗ , Thomas Ierardi, Michael Lesslie, Brian T. Hoffman, Daniel Mulvana Bioanalytical and ADME Labs, Quintiles, 19 Brown Road, Ithaca, NY 14850, USA
a r t i c l e
i n f o
Article history: Received 18 February 2014 Accepted 7 August 2014 Available online 15 August 2014 Keywords: Lithiated adducts Alkali adducts HILIC Tandem mass spectrometry (MS/MS) Anomerizaton Decitabine
a b s t r a c t A highly sensitive, selective, and rugged quantification method was developed and validated for decitabine (5-aza-2 -deoxycytidine) in human plasma treated with 100 g/mL of tetrahydrouridine (THU). Chromatographic separation was accomplished using hydrophilic interaction liquid chromatography (HILIC) and detection used electrospray ionization (ESI) tandem mass spectrometry (MS/MS) by monitoring lithiated adducts of the analytes as precursor ions. The method involves simple acetonitrile precipitation steps (in an ice bath) followed by injection of the supernatant onto a Thermo Betasil Silica100, 100 × 3.0 mm, 5 m LC column. Protonated ([M+H]+ ), sodiated ([M+Na]+ ), and lithiated ([M+Li]+ ) adducts as precursor ions for MS/MS detection were evaluated for best sensitivity and assay performance. During initial method development abundant sodium [M+Na]+ and potassium [M+K]+ adducts were observed while the protonated species [M+H]+ was present at a relative abundance of less than 5% in Q1. The alkali adducts were not be able to be minimized by the usual approach of increasing acid content in mobile phases. Significant analyte/internal standard (IS) co-suppression and inter-lot response differences were observed when using the sodium adduct as the precursor ion for quantification. By adding 2 mM lithium acetate in aqueous mobile phase component, the lithium adduct effectively replaced other cationic species and was successfully used as the precursor ion for selected reaction monitoring (SRM) detection. The method demonstrated the separation of anomers and from other endogenous interferences using a 3-min gradient elution. Decitabine stock, working solution stabilities were investigated during method development. Three different peaks, including one from anomerization, were observed in the SRM transition of the analyte when it was in neutral aqueous solution. The assay was validated over a concentration range of 0.5–500 ng/mL (or 0.44–440 pg injected on column) in 50 L of human plasma. The accuracy and precision were within 8.6% relative error and 6.3% coefficient of variation, respectively. Decitabine was stable in THU treated human plasma for at least 68 days and after 5 freeze–thaw cycles when stored at −70 ◦ C. Stability of decitabine in THU treated human whole blood, matrix factor and recovery were also evaluated during method validation. The method was successfully used for clinical sample analysis. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Decitabine (5-aza-2 -deoxycytidine, Fig. 1) is a drug used in the treatment of patients with myelodysplastic syndromes and has demonstrated promising activity in the treatment of acute leukemia, in particular acute myeloid leukemia [1,2]. At high doses, decitabine exerts primarily a cytotoxic effect, but at lower doses its hypomethylating activity predominates, resulting in gene
∗ Corresponding author. Tel.: +1 607 330 9831. E-mail addresses: [email protected], [email protected] (W. Hua). http://dx.doi.org/10.1016/j.jchromb.2014.08.012 1570-0232/© 2014 Elsevier B.V. All rights reserved.
re-expression and cellular differentiation of leukemic cells [3]. However, the pharmacokinetic (PK) profile of decitabine at lower doses (e.g., 15 mg/m2 ) in humans remained elusive until recently, in partially due to challenges of quantifying and characterizing decitabine in biological matrices [2,4,5]. Decitabine is unstable at physiological temperature and pH. The degradation of decitabine results a plethora of products, formed by hydrolytic opening, deformylation of the triazine ring, anomerization, and possibly other changes in the sugar ring structure [4–6]. Additionally, it is susceptible to deamination by high levels of cytidine deaminase, as occurs with other nucleoside analogs in human plasma [1]. The pharmacological and toxicological properties of the
118
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
NH2
13
C N
N
HC HO CH2 O HC
C
CH HO
Corporation (PA, USA). Human whole blood (K2 EDTA) was collected from Quintiles volunteers in accordance with current company policies on informed consent and ethical approval.
15
NH2
N CH
N
15
N
13
O
H C HO CH2 O HC CH
CH2
A
C
15
HO
15
N
2.2. Preparation of stock solutions, calibration standards, and quality controls (QCs)
C O
CH CH2
B
Fig. 1. Chemical structures of (A) decitabine (-anomer), C8 H12 N4 O4 , monoisotopic mass: 228.09; (B) IS [13 C2 , 15 N4 ]-decitabine (mixture of ␣- and -anomer), 13 C2 C6 15 N4 H12 O4 , monoisotopic mass: 234.09.
degradation products are yet largely unknown. Decitabine is also a highly polar compound with log P of approximately −2.16 [7]. Due to its high hydrophilicity, it is difficult to retain on a reversedphase (RP) column unless highly aqueous mobile phase eluents are used, resulting in lower sensitivity in LC/MS detection. Postcolumn addition of organic solvent to enhance ionization after RP LC separation was reported [4]. HILIC with unique selectivity has been demonstrated as an alternative separation technique for small polar compounds [8,9] including several anticancer drugs [10]. In addition, high content of organic solvent, typically acetonitrile, can provide increased sensitivity for mass spectrometric detection. A few analytical methods have been reported for quantifying decitabine in biological matrices. Early methods based on L1210 cell kills [11] and LC–UV techniques [12,13] suffered from lack of sensitivity and selectivity and could not be applied for PK characterization at lower doses. In recent years, more sensitive methods have been reported using protein precipitation (PPT) or mixedmode cation exchange (MCX) solid-phase extraction (SPE) coupled with RP-LC–MS/MS with lower limits of quantification (LLOQs) in a range of 1–10 ng/mL (or as low as 14 pg injected on-column) in rat or human plasma [3,4,14,15]. These methods used the protonated molecule as precursor ion for MS/MS detection, except Xu et al. [14] who reported use of sodiated adducts for MS/MS detection. In this study, we developed and validated a highly sensitive and selective HILIC–MS/MS method to analyze decitabine in human plasma by using lithiated adducts. To stabilize decitabine in human whole blood or plasma, cytidine deaminase inhibitor THU at 100 g/mL was added into samples as previously described [15]. With a simple PPT, a LLOQ of 0.5 ng/mL (or 0.44 pg injected on-column) in 50 L of human plasma was easily achieved. 2. Experimental 2.1. Chemical and reagents Decitabine (5-aza-2 -deoxycytidine, >97%) was purchased from Sigma–Aldrich (MO, USA). [13 C2 , 15 N4 ]-decitabine (mixture of anomers, >98.6%, Fig. 1) used as IS was purchased from Alsachim (Strasbourg, France). Acetonitrile, methanol, water were HPLC grade and were obtained from Honeywell Burdick & Jackson (MI, USA). Dimethylformamide (DMF), dimethyl sulfoxide (DMSO), ammonium formate, and lithium acetate dihydrate were purchased from Sigma–Aldrich (MO, USA). THU was purchased from EMD Millipore (MA, USA). Formic acid was purchased from EMD Chemicals Inc. (NJ, USA). Trifluoroacetic acid (TFA) was purchased from Fisher Scientific (PA, USA). Human plasma treated with K2 EDTA was obtained from Bioreclamation (NY, USA) or Biological Specialty
Stock solutions of decitabine and [13 C2 , 15 N4 ]-decitabine were prepared at a concentration of 0.5 mg/mL in DMF and stored at −20 ◦ C. Standard working solutions (SWS) at 20,000 ng/mL of decitabine in DMF were used for the preparation of calibration curves and QCs. Control human plasma used for this study was treated with THU at a concentration of 100 g/mL. Spiked calibration standards in control human plasma were prepared from serial dilutions at the following concentrations: 0.5, 1.0, 5.0, 50.0, 100, 250, 425, and 500 ng/mL. The QCs were spiked at the following levels: 0.5 (LLOQ), 1.5 (QC1), 15 (GMQC1), 150 (QC2), and 400 ng/mL (QC3). A 20-fold dilution QC was prepared at a concentration of 8000 ng/mL. Internal standard working solution (ISWS) was prepared in DMF at 100 ng/mL and stored at −20 ◦ C. 2.3. Sample preparation Sample preparation was performed in an ice bath (equivalent to 4 ◦ C). Protein precipitation in 1 mL 96-well plates was used for sample preparation. Briefly, 50 L aliquots of plasma samples were mixed with 20 L of ISWS and followed by a 500 L addition of acetonitrile. The plate was vortexed and centrifuged for 5 min. Supernatant was transferred into a clean 1 mL 96-well plate and a volume of 10 L was injected onto LC–MS/MS instrument. 2.4. Instrument analysis The HPLC system was equipped with an SCL-10A pump controller, two LC-10AD pumps (Shimadzu, MD, USA) and PAL autosampler (LEAP Technologies, NC, USA). The HPLC system was interfaced to an API5000 tandem mass spectrometer (Applied Biosystems-MDS Sciex, ON, Canada). The chromatographic separation was performed on a Thermo Betasil Silica-100, 100 × 3.0 mm, 5 m HPLC column (Thermo Scientific, MA, USA) at ambient temperature. The mobile phases (MP) used consisted of 100:1:0.5:0.4 water/1 M ammonium formate in water/formic acid/0.5 M lithium acetate in water as MP A and acetonitrile as MP B. The elution gradient was from 90% B initial to 50% B in 3 min at 0.5 mL/min flow rate. The LC column was flushed with 40% B at 1.8 mL/min and equilibrated to the initial condition. Water and 90:10 acetonitrile/water were used for autosampler washes to minimize carryover. The LC eluent was introduced to the mass spectrometer using the ESI interface in positive ion mode. The ion spray voltage, turbo heater temperature, and declustering potential were set at 2500 V, 650 ◦ C, and 60 V, respectively. The quantification was performed using SRM detection with the transitions of m/z 235.1→119.1 for decitabine and m/z 241.1→125.0 for IS. The optimized collision energy was at 22 eV for both decitabine and IS. Due to the extensive signal of decitabine and limited amount of - in IS, a collision energy of 30 eV was used for detuning decitabine while the optimized value was used for the IS during validation. 2.5. Method validation To validate the method, tests of the selectivity, linearity, intraand inter-assay accuracy and precision, sample dilution, extraction recovery, matrix effects, freeze/thaw and long-term stability in plasma samples as well as working solution and stock solution stability were conducted in compliance with the Food and Drug Administration guidelines [16].
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
251.1 [M+Na]+
100%
Max. 2.9e6 cps
Relative Ion Abundance (%)
90% 80% 70% 60% 50% 40% 30% 20% 252.1
10%
229.2 [M+H]+
200
220
210
230
240
250 m/z
260
266.9 [M+K]+ [M+ACN+Na]+ 292.2 270
280
290
300
A
235.1 [M+Li]+
100%
Max. 2.7e6 cps
Relative Ion Abundance (%)
90% 80% 70% 60% 50% 40% 30%
236.2 234.3
20% 10% 200
220
210
230
240
250 m/z
260
270
280
290
300
B
Fig. 2. Q1 and MS/MS full scan spectra of decitabine (A) without lithium acetate added in MP A; (B) with 2 mM lithium acetate added in MP A. Q1 full scan spectra were acquired by injection with gradient elution described in Section 2.4.
The selectivity was verified using six individual blank human plasma samples. Whether any endogenous peaks were interfering with the peak of decitabine or IS in the SRM chromatograms was investigated using blank plasma samples without spiking the analyte and IS.
2.0e6
Self-suppression across the curve
↑
Inter-lot variability
↑
IS Peak Area (counts)
1.8e6 1.6e6 1.4e6 1.2e6 1.0e6 8.0e5 6.0e5
Calibration Standard (1 Matrix Lot) Inter-lot Zero Samples (6 Matrix Lots) QC Samples (2 Matrix Lots) Control Blank or Zero Sample (1 Matrix Lot)
4.0e5 2.0e5 0 5
15
25
35
45
55
65
75
85
95
Injection Number Fig. 3. IS response plot when using [M+Na]+ as precursor ion with MRM transition 251.1→135.0 for decitabine and 257.1→141.0 for IS. MPA: 100:1:0.5 water/1 M ammonium formate in water/formic acid. MP B: acetonitrile.
119
The intra- and inter-assay accuracy and precision were evaluated using six replicates of QC samples at the concentrations described in Section 2.2. Long term storage stability was evaluated at −20 ◦ C and −70 ◦ C for up to 68 days, 5 freeze/thaw-cycle (−20 ◦ C, −70 ◦ C) and bench-top up to 29 h at 4 ◦ C were evaluated for stability of decitabine in THU treated human plasma with QC1, QC3, and dilution QC. The processed sample stability for decitabine was determined by quantifying the extracted, stored QC samples (QC1, QC2, and QC3) up to 173 h at 4 ◦ C against freshly extracted standard curves. In the reproducibility and stability tests, the mean accuracy was expressed as RE (%) and the precision expressed as coefficient of variation (CV, %). In addition to the stability of decitabine in THU treated plasma, the stability of decitabine in THU treated (100 g/mL) whole blood was evaluated by spiking decitabine at a concentration of 250 ng/mL. The baseline plasma sample (T = 0) was harvested by centrifugation of the whole blood sample after 10-min incubation and immediately frozen at −20 ◦ C. This sample was compared to samples harvested after storage of whole blood for 1 h (T = 1 h) at ambient and in an ice bath. In this study, the control matrix used for analytical and stability QC samples described above were prepared by adding THU to plasma and then spiking decitabine. Since the amount of THU that ends up in the plasma from clinical study samples cannot be determined, additional QC samples (QC1, QC3, and dilution QC) were prepared in a manner to more closely simulate study samples where whole blood was collected in tubes containing THU at 100 g/mL and then spun down to plasma followed by spiking decitabine. These QCs were stored at −20 ◦ C and −70 ◦ C for 0 and 48 days for additional long-term storage stability. The extraction recoveries were measured by comparing the peak area ratios of decitabine QCs at all four concentrations (QC1, GMQC1, QC2, QC3) in samples spiked after extraction (post-extract) with those of samples spiked before extraction (pre-extract). Matrix effects (matrix factor) were determined by comparing the peak areas of analyte in spiked plasma extracts at all four QC concentration levels to those in plasma-free extracts in six different sources of plasma.
3. Results and discussion 3.1. Optimization of chromatographic condition and sample preparation Both RP and HILIC LC separations were evaluated during method development. Several LC–MS methods for decitabine were published [3,4,12–15] but none of them employed HILIC separation. Since decitabine is a strong hydrophilic compound, reversed phase C18 and C8 columns which are marketed as suitable for analysis of polar compounds were initially tested. Decitabine was poorly retained on these columns unless using very high percentage of aqueous phase, and suffered from poor selectivity and sensitivity. HILIC columns, including Betasil silica, 5 m (100 × 3.0 mm, 50 × 2.1 mm), Kinetex HILIC 2.6 m (50 × 2.1 mm), and Atlantis Silica HILIC 5 m (2.1 × 50 mm) were then evaluated. Betasil silica 5 m (100 × 3.0 mm) was chosen for the method based on optimal retention time (RT), selectivity, and peak shape. Ammonium formate, ammonium acetate with or without acid were evaluated as aqueous MP additives for best sensitivity, selectivity and chromatographic reproducibility. Ammonium formate and ammonium acetate without pH adjustment both improved peak shape and separation relative to use of water. Increased sensitivity was achieved when ammonium formate and formic acid were used. The stable isotope-labeled IS used in our method was a mixture of anomers with ␣-anomer the majority. Decitabine, however is
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
3.3. Stock, working solutions stability Stock solution, SWS stabilities were evaluated in different solvents during method development. Stock solutions of 0.5 mg/mL of were dissolved in DMSO, DMF, and 50:50 methanol/water. After 18 days of storage at −20 ◦ C, there was 33% decrease for the stock solution prepared in 50:50 methanol/water and no sign of degradation for those prepared in DMSO or DMF. DMF, 0.1 N HCL in
1.2e6 8.0e5 4.0e5 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
A
2.17
790 Decitabine 600 400 200 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
B Intensity (cps)
3.2. Optimization of mass spectrometric conditions
2.17 β-anomer
7.0e5 Decitabine 5.0e5
2.30 α-anomer 2.46
3.0e5 1.0e5 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
C
2.5e5
Intensity (cps)
The protonated molecule m/z 229 was mostly used as the precursor ion in previously published methods. Xu et al. [14] reported formation of mainly sodium adducts of decitabine in full-scan positive ESI MS, which were used as precursor ions for quantification of decitabine in rat plasma. In our study, we also found that decitabine formed predominantly sodium adducts at m/z 251.1, and potassium adducts at m/z 267.1 (Fig. 2A). Initial effort was focused on suppression of sodium and potassium adduction by increasing acid content in aqueous MP. Formic acid (0.5%, 2.5%, and 5%) and TFA (0.5%, 1%, and 2.5%) in aqueous MP were evaluated for minimizing adduct formation. However, sodium adducts still predominated at the highest formic acid and TFA concentrations and no significant increase of protonated ions in Q1 was observed. Quantitation using sodium adducts (SRM transition m/z 251.1→135.0 for decitabine and m/z 257.1→141.0 for IS) was evaluated. The sodium adduction was maintained in the collision-induced dissociation (CID) fragmentation. Significant analyte/IS co-suppression, and inter-lot response differences were observed even though accuracy and precision were acceptable. Fig. 3 shows the IS responses from a 96-injection run that included 6 inter-lot LLOQs, all levels of QCs and standard curves when using [M+Na]+ as precursor ion for quantitation. The QCs and standard curves were prepared in two different sources of human plasma. The IS response across the run had 31% of precision, which is unexpectedly high for a PPT method. Since lithium adduction has been previously shown as an effective means of ionizing weakly basic compounds [17,18], it was considered a possible means of displacing the sodium and potassium adducts. Addition of 2 mM of lithium acetate to the aqueous component of MP effectively formed entirely [M+Li]+ in Q1. The most abundant product ion (m/z 119.0) (Fig. 2B) for decitabine and m/z 125.0 for IS) was used for quantitation. These species maintained the lithium adduction in the CID fragmentation. When the same run was re-injected with 2 mM lithium acetate MP a consistent IS response was achieved in SRM detection with 7% of precision.
2.13 β-anomer
1.6e6 Decitabine
Intensity (cps)
the -anomer. It was found that improved quantitation (better precision) was achieved if the IS anomers were chromatographically separated and only the -anomer was used for quantitation. Both PPT and cation exchange SPE were investigated for sample preparation. For SPE, Oasis MCX 96-well plate (10 mg) and Evolute CX 96-well plate (25 mg) were evaluated using 50 L sample aliquots. The recoveries from both SPE plates were excellent and were consistent with previously reported data [4]. PPT using acetonitrile without drying down step was used due to its simplicity and cost-effectiveness, and most importantly, due to high sensitivity achieved by HILIC separation. Sample preparation in an ice bath is necessary due to insufficient stability of analytes in the matrix. Bench-top stability of spiked QCs was evaluated during initial method development at two concentrations (QC1 and dilution QC). After 6 h of storage at ambient temperature, 18% of degradation was observed for dilution QC. After 20 h of storage at ambient temperature, 37% and 63% of degradation was observed for QC1 and dilution QC, respectively. No significant degradation was observed for the same QCs after 20 h of storage at 4 ◦ C.
Intensity (cps)
120
2.27 α-anomer
[13C215N4]Decitabine
1.5e5 β-anomer 2.14 5.0e4 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
D
Fig. 4. Chromatograms of 20,000 ng/mL decitabine in (A) DMF; (B) 0.1 N HCl(aq); (C) water stressed at 37 ◦ C for 17 h; (D) [13 C2 , 15 N4 ]-decitabine (mixture of ␣- and -anomer).
water, and water were evaluated as diluents for SWS. To enable quick evaluation, SWS prepared in above diluents were stored at −20 ◦ C, ambient, and at 37 ◦ C for 17 h. These solutions were then compared to the ones that were freshly prepared. Decitabine prepared in DMF was stable under all the storage condition including at 37 ◦ C for 17 h. Degradation of decitabine in 0.1 N HCl in water, and water was large dependant on storage temperature (see Supplementary Table S1). No significant degradation was observed when the solutions were stored at −20 ◦ C, while after storage at 37 ◦ C for 17 h, 100% and 59% degradation was observed for 0.1 N HCl in water, and water, respectively. Degradation products were different in 0.1 N HCl and water based on the chromatographic evidence. Fig. 4A–C presents chromatograms of decitabine in DMF, 0.1 N HCl in water, and water after storage at 37 ◦ C for 17 h. A total of four peaks (RT = 1.94, 2.17, 2.30, 2.46 min) in the same SRM transition of decitabine (m/z 235.1→119.0) were identified for decitabine in water that was stress at 37 ◦ C for 17 h (Fig. 4C). Rogstad et al. [5] proposed that at neutral pH in aqueous phase, both base-catalyzed events (hydrolytic ring-opening) and acid-catalyzed events (sugar ring-opening followed by anomerization and reclosure) can take place simultaneously for decitabine, resulting ␣- and -anomers
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
Intensity (cps)
8000 2.44 α-anomer
4000 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
Intensity (cps)
2.30 β-anomer
1.15e4 Decitabine
3000
121
1.58
Decitabine 0.99
2000 1000 0
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
5000
D 2.44 α-anomer
Decitabine
3000 2000 1000 0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
0.98
2000 1000 0
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2. 2 2 .4 2.6 2.8 Time (min)
E 2.21 α-anomer
2.30 β-anomer Intensity (cps)
Decitabine Intensity (cps)
2.08 β-anomer
3000
B 2000 1000 0
1.58
4000 Decitabine
4000
Intensity (cps)
Intensity (cps)
A
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Time (min)
1.2e5 8.0e4 β-anomer 2.08 4.0e4 0
2.0 2.2 2.4 2.6 2.8
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 Time (min)
F
C
Fig. 5. Chromatograms of extracted (A) clinical study sample T = 2 h; (B) clinical study sample T = 3 h; (C) long-term stability QC1 (after 48 days at −70 ◦ C); (D) control blank; (E) LLOQ standard; (F) [13 C2 , 15 N4 ]-decitabine (mixture of ␣- and -anomer). Note: Retention time of A, B, C is later than that of D, E, F since they are from different analysis batches. See insert for IS transition of A, B, C.
of furanoside and pyranoside derivatives. In his study, only two peaks at m/z 229 proposed as ␣- and -anomers of decitabine were identified by LC–UV and QTOF/MS from stressed (37 ◦ C for 24 h) decitabine in 100 mM potassium phosphate (pH 7.4). Our results confirmed anomerization since one of the peaks eluted at the same RT of ␣-anomer of the IS. This was also observed in the study samples (Fig. 5A and B) but not in the long-term stability QCs (Fig. 5C) that were prepared in a manner closely simulating study samples (see Section 2.5). To our knowledge, detection of peaks eluting at RT = 1.94, 2.46 min as degradation products have not been reported. Since these two peaks had same product ions in SRM detection, they are likely the ␣- and -anomers of pyranoside derivatives with triazine ring (formed by reclosure of hydrolytic ring-opening) as proposed by Rogstad et al. [5]. Supplementary Table S1 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jchromb.2014.08.012.
3.4. Method validation
Table 1 Accuracy and precision of decitabine in THU treated human plasma from three validation runs.
Treated plasma lot (n = 6)
Decitabine concentration (ng/mL)
0.5 (LLOQ) 1.5 (QC1) 15 (GMQC1) 150 (QC2) 400 (QC3) 8000 (dilution QC, 20-fold)
Intra-assay (n = 6)
Inter-assay (n = 18)
RE (%)
CV (%)
RE (%)
CV (%)
8.6 4.7 0.7 0.7 −1.8 −2.6
4.9 2.8 1.3 1.6 1.6 0.6
6.4 5.3 1.3 0 −2.3 −1.1
6.3 3.0 1.3 1.3 1.8 1.9
3.4.1. Selectivity Fig. 5D and E shows the typical chromatograms of a blank plasma sample and a blank plasma sample spiked with decitabine at the LLOQ. No chromatographic interferences were detected in six different sources of plasma that were evaluated. 3.4.2. Linearity and LLOQ Standard curves were performed in duplicate in THU treated plasma. In all cases the regression coefficient was >0.997. Decitabine was linear over the range of 0.5–500 ng/mL with a weighting of 1/x2 . All coefficients of determination (r2 ) were ≥0.998.
Table 2 Inter-lot precision and accuracy for decitabine. Decitabine concentration (ng/mL) LLOQ (0.5)
ULOQ (500)
1 2 3 4 5 6
0.493 0.474 0.500 0.469 0.556 0.542
494 484 477 485 479 490
Mean CV (%) RE (%)
0.506 7.1 1.2
485 1.3 −3.0
122
W. Hua et al. / J. Chromatogr. B 969 (2014) 117–122
Table 3 Stability of decitabine in THU treated (100 g/mL) human plasma under various storage conditions. Plasma concentration (ng/mL)
1.5 (QC1) 150 (QC2) 400 (QC3) 8000 (dilQC4)
Long-term storage (n = 4)
Five freeze–thaw cycles (n = 4)
4 ◦ C bench-top (n = 4)
Processed sample at 4 ◦ C (n = 8)
-20 ◦ C (21 days)
-70 ◦ C (68 days)
-20 ◦ C
29 h
173 h
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
RE (%)
CV (%)
−6.0 NA −9.0 −8.9
4.0 NA 1.3 0.3
4.0 NA −1.8 −1.3
2.4 NA 2.2 1.5
-2.7 NA −11.3 −10.0
3.8 NA 1.0 2.1
−0.7 NA −6.5 −6.6
3.6 NA 2.3 3.0
−6.7 NA −12.5 −16.0
2.9 NA 1.8 1.7
−2.0 −6.0 −9.5 NA
3.1 1.5 2.7 NA
3.4.3. Precision and accuracy Table 1 summarizes the results for intra- and inter-day precision and accuracy for decitabine measured by QCs. The intra- and interday precision were all below 15% with a maximum CV % of 6.3% and a maximum bias (RE %) of 8.6% for accuracy was calculated. 3.4.4. Recovery and matrix effect The recoveries of decitabine extracted from THU treated plasma were 69.6 ± 11.1%, 73.2 ± 3.3%, 77.0 ± 2.6%, and 77.0 ± 2.2% at the concentrations of 1.5, 15, 150, and 400 ng/mL, respectively (n = 6). The recovery of IS was 80.8 ± 1.6% at concentration of 100 ng/mL (n = 6). The matrix factor of decitabine from six different sources of human plasma samples at concentrations of 1.5, 15, 150, and 400 ng/mL (n = 6) were in the range of 103–112% with CV% below 10%, indicating no matrix-dependent quantitation issues. In addition, 6 different lots of plasma were spiked with analyte at the LLOQ and ULOQ with resulting CV ≤ 7.1% and mean bias within ±3% (Table 2). 3.4.5. Effect of hemolyzed and lipemic plasma To evaluate whether real world sample conditions would quantify accurately, plasma containing 2% and 5% lysed blood as well a plasma containing high triglyceride levels were spiked with analyte at concentrations of 1.5 and 400 ng/mL. Acceptable quantitation was achieved in all cases (see Supplementary Table S2). Supplementary Table S2 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.jchromb.2014.08.012. 3.4.6. Stability Decitabine in THU treated human plasma at 1.5, 400, and 8000 ng/mL was found stable at -20 ◦ C for at least 21 days, at -70 ◦ C for at least 68 days, after 5 freeze–thaw cycles when stored at 20 ◦ C or -70 ◦ C, at approximately 4 ◦ C (bench-top, equivalent to an ice bath) under yellow light after 29 h (Table 3) although some loss is evident in the bench-top samples. This loss may be a function of the aqueous instability rather than due to cytidine deaminase activity, since the latter would be expected to be suppressed by THU. Decitabine in processed samples was stable at 4 ◦ C for at least 173 h. The stock solution at 0.5 mg/mL and SWS at 20,000 ng/mL in DMF were stable for at least 49 days at -20 ◦ C and at least 30 h at 4 ◦ C. Decitabine was stable in THU treated human whole blood after storage for 1 h at ambient and in an ice bath (Supplementary Table S3). For the QC samples (1.5, 400, and 8000 ng/mL) prepared in the control plasma that were centrifuged from THU treated human whole blood, stability was established for decitabine for up to 48 days at -70 ◦ C (Fig. 4C, data not shown) but not at -20 ◦ C. Supplementary Table S3 related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jchromb. 2014.08.012.
-70 ◦ C
4. Conclusion A fast, sensitive and rugged HILIC–MS/MS method has been developed and validated for the analysis of decitabine in THU treated human plasma. With a simple step of protein precipitation, our method achieved 0.5 ng/mL LLOQ with 50 L sample volume (or 0.44 pg on-column). The use of lithium adducts enabled reproducible analysis in multiple sources of matrix while maintaining sensitivity requirements for the analysis of clinical samples. Further improvements in sensitivity could be gained based on our preliminary evaluation of MCX SPE extraction should it be required (e.g. reduced sample volumes in pediatric samples). However, the simplicity of PPT extraction enables low cost and reproducible sample analysis while maintaining high sensitivity. The ability to separate anomers is important since ␣-anomer was observed in clinical study samples. Acknowledgements The authors would like to acknowledge J. Johnson, G. Lal, C. Jordaens, B. Beckhorn, M. Mann, A. Drelick, J. Urda, C. Tilley, C. Reeves, E. King, E. Button, M. Welser for sample preparation and LC/MS analysis. S. Swift-Spencer is thanked for formatting figures and tables. References [1] E. Jabbour, J.-P. Issa, G. Garcia-Manero, H. Kantarjian, Cancer 112 (2008) 2341–2351. [2] J. Bryan, H. Kantarjian, G. Garcia-Manero, E. Jabbour, Expert Opin. Drug Metab. Toxicol. 7 (5) (2011) 661–672. [3] A.F. Cashen, A.K. Shah, L. Todt, N. Fisher, J. DiPersio, Cancer Chemother. Pharmacol. 61 (2008) 759–766. [4] Z. Liu, G. Marcucci, J.C. Byrd, M. Grever, J. Xiao, K.K. Chan, Rapid Commun. Mass Spectrom. 20 (2006) 1117–1126. [5] D.K. Rogstad, J.L. Herring, J.A. Theruvathu, A. Burdzy, C.C. Perry, J.W. Neidigh, L.C. Sowers, Chem. Res. Toxicol. 22 (2009) 1194–1204. [6] K.T. Lin, R.L. Momparler, G.E. Rivard, J. Pharm. Sci. 70 (1981) 1228–1232. [7] http://www.chemicalize.org/structure/#!mol=Dacogen [8] W. Naidong, J. Chromatogr. B 796 (2003) 209–224. [9] P. Hemström, K. Irgum, J. Sep. Sci. 29 (2006) 1784–1821. [10] S. Nussbaumer, P. Bonnabry, J.-L. Veuthey, S. Fleury-Souverain, Talanta 85 (2011) 2265–2289. [11] G.G. Chabot, G.E. Rivard, R.L. Momparler, Cancer Res. 43 (1983) 592–597. [12] J.M. Covey, D.S. Zaharko, Eur. J. Cancer Clin. Oncol. 21 (1985) 109–117. [13] K.T. Lin, R.L. Momparler, G.E. Rivard, J. Chromatogr. 345 (1985) 162–167. [14] H. Xu, S. Lv, M. Qiao, Y. Fu, X. Jiang, Y. Jin, C. Li, B. Yuan, J. Chromatogr. B 899 (2012) 81–85. [15] K. Patel, S.M. Guichard, D.I. Jodrell, J. Chromatogr. B 863 (2008) 19–25. [16] Guidance for Industry Bioanalytical Method Validation, U.S. Department of Health and Human Services, Food and Drug Administration Centre for Drug Evaluation and Research (CDER), Centre for Veterinary Medicine (CM), May 2001, http://www.fda.gov/cder/guidance/4252fnl.htm [17] J. Adams, M.L. Gross, J. Am. Chem. Soc. 108 (1986) 6912–6915. [18] B.M. Ham, J.T. Jacob, M.M. Keese, R.B. Cole, J. Mass Spectrom. 39 (2004) 1321–1336.
