Anal. Chem. 2002, 74, 591-599
A Validated Liquid Chromatography/Tandem Mass Spectrometry Assay for cis-Amminedichloro(2-methylpyridine)platinum(II) in Human Plasma Ultrafiltrate Tomoyuki Oe,† Ye Tian,† Peter J. O’Dwyer,† David W. Roberts,‡ Michael D. Malone,‡ Christopher J. Bailey,‡ and Ian A. Blair*,†
Center for Cancer Pharmacology, Department of Pharmacology, University of Pennsylvania, 1254 BRB II/III, 421 Curie Boulevard, Philadelphia, Pennsylvania 19104-6160, and DMPK Department, AstraZeneca U.K., Mereside, Alderley Park, Macclesfield, Cheshire SK10 4 TG, United Kingdom.
The clinical use of platinum drugs as anticancer agents has encountered problems when relating pharmacokinetic profiles with efficacy and toxicity is attempted. This has been mainly due to the lack of specific and sensitive analytical methodology to examine concentrations of the unbound drug in plasma. The presence of a carbocyclic ring on the new drug, cis-amminedichloro(2-methylpyridine)platinum(II) (ZD0473) suggested that it would be possible to develop the first stable isotope dilution LC/ MS assay for a platinum drug in human plasma ultrafiltrate samples. The dichloro form of the drug exists in equilibrium with at least two aquated forms in plasma. The molecular form of the drug, therefore, depends on the length of time that the plasma sample is maintained at room temperature before freezing. Therefore, we have developed a method that quantitatively converts the aquated species back to the dichloro form of the parent drug so that a single molecular species can be analyzed. Selected reaction monitoring was performed on the transition of m/z 393 [M + NH4]+ to m/z 304 [M + NH4 NH3 - 2 × HCl]+ for ZD0473, and m/z 400 [M + NH4]+ to m/z 310 [M + NH4 - NH3 - HCl - 2HCl]+ for [2H7]ZD0473. The standard curves were fitted to a quadratic regression over the range from 10 to 5000 ng/ mL in human plasma ultrafiltrate. The lower limit of quantitation for ZD0473 was 10 ng/mL for 100 µL of plasma ultrafiltrate. This simple, rapid, reliable, and sensitive method of quantitation had excellent accuracy and precision. The method provided adequate sensitivity for the analysis of plasma ultrafiltrate samples from a phase II study in which ZD0473 was administered to patients as an intravenous infusion at a dose of 150 mg/ m2. The platinum agent ZD0473 {cis-amminedichloro(2-methylpyridine)platinum(II)} (Figure 1) is a novel sterically hindered * Corresponding author. Phone: 215-573-9885. Fax: 215-573-9889. E-mail:
[email protected]. † University of Pennsylvania. ‡ AstraZeneca U.K.. 10.1021/ac010792v CCC: $22.00 Published on Web 01/08/2002
© 2002 American Chemical Society
Figure 1. Structures of platinum anticancer drugs.
antitumor agent that was designed to circumvent platinum drug resistance induced by cisplatin (Figure 1).1 ZD0473 was chosen from a panel of sterically hindered pyridine platinum complexes because of its reduced reactivity with sulfur ligands, unique DNA binding properties, and its ability to circumvent several of the major resistance mechanisms that are observed in cisplatinresistant tumor cell lines,2 including acquired cisplatin resistance of human ovarian carcinoma cell models.3 ZD0473 hydrolyses more slowly in water than cisplatin and is less reactive with sulfhydryl ligands.4 The 2-methylpyridine ligand that is attached to the platinum has allowed the development of a specific analytical method for the molecule, thus providing plasma ultrafiltrate concentration data based on ZD0473 rather than total platinum. Pharmacokinetic guided dosing strategies, such as those derived for carboplatin,5 may be possible using these data. We have shown previously that the platinum(IV) drug JM216 was extensively metabolized by in vivo reduction to the platinum(II) compound, JM118 (Figure 1).6,7 A substantial portion of the drug was excreted (1) Holford, J.; Raynaud, F.; Murrer, B. A.; Grimaldi, K.; Hartley, J. A.; Abrams, M.; Kelland, L. R. Anticancer Drug Des. 1998, 13, 1-18. (2) Raynaud, F. I.; Boxall, F. E.; Goddard, P. M.; Valenti, M.; Jones, M.; Murrer, B. A.; Abrams, M.; Kelland, L. R. Clin. Cancer Res. 1997, 3, 22063-74. (3) Holford, J.; Sharp, S. Y.; Murrer, B. A.; Abrams, M.; Kelland, L. R. Br. J. Cancer 1998, 77, 366-73. (4) Kelland, L. R. Cisplatin; Lippert, B., Ed.; Verlag Helvetica Chimica Acta: Zu ¨ rich, 1999; pp 498-521. (5) van den Bongard, D. H. J.; Mathot, R. A.; Beijnen, J. H.; Schellens, J. H. M. Clin. Pharmacokin. 2000, 39, 345-67.
Analytical Chemistry, Vol. 74, No. 3, February 1, 2002 591
as cyclohexylamine,8 although it was never established whether the cyclohexylamine arose from metabolism of the parent drug or from the JM118 metabolite. A number of other studies have shown that platinum drugs with carbocyclic rings, such as carboplatin and oxaliplatin (Figure 1), can undergo biotransformations in vivo.5,9-11 Pharmacokinetic studies of platinum drugs have typically relied upon atomic absorption spectrophotometry12,13 or more sensitive ICPMS methodology13,14 for quantitation of Pt in plasma ultrafiltrate. Because of the potential for biotransformation and the presence of inactive low-molecular-weight (20 (Figure 6C). No interfering peaks were observed in LC/SRMMS chromatograms for the LQC, MQC, or HQC, and the intensity of the internal standard channel remained relatively constant. The mean retention time for the analyte was 6.97 min (precision 0.1%), and the internal standard was 6.92 min (precision 0.1%) during 3 days of validation. This showed that the method was very robust. Calibration Curves. The calibration curves were prepared by plotting the peak area ratio between ZD0473 and [2H7]ZD0473 against the concentration of ZD0473 in the plasma ultrafiltrate.
All calibration curves were fitted (r2 ) 0.9999 a quadratic regression) in the range of 10-5000 ng/mL (Figure 7). The accuracy of the back-calculated values from theoretical values ranged from 95.9 to 104.8%. Precision and Accuracy. The lower limit of quantitation of the ZD0473 assay was 10 ng/mL (250 pg/on column). The signalto-noise ratio (S/N) was >20, and it showed a symmetrical and robust peak. The accuracy over the three validation days for the LLQ was 110.9%, and the precision was 12.1% (Table 1). Accuracy was in the range 101.1-107.7% and precision was 5.0% or better for the other QC samples (Table 1). It was anticipated that there would be occasional clinical samples with an inadequate amount of plasma ultrafiltrate. For these samples, a 1-day validation was conducted in which 50-µL plasma ultrafiltrate QC samples were diluted with drug-free plasma ultrafiltrate. Accuracy and precision were similar to those found for the 100-µL samples (Tables 1 and 2). Therefore, all of the QC samples readily met the criteria established by Shah et al. for assay validation.28 The LLQ, LQC, MQC, HQC, and UQC validation samples from day 1 were reanalyzed after 3 days of storage on the autosampler at 4 °C. The accuracy for the LLQ, LQC, MQC, HQC, and UQC Analytical Chemistry, Vol. 74, No. 3, February 1, 2002
597
Table 2. Accuracy and Precision for 2-Fold Diluteda LQC, MQC, and HQC Samplesb
dil. LQC (40 ng/mL) dil. MQC (400 ng/mL) dil. HQC (4000 ng/mL)
mean (ng/mL)
accuracy (%)
precision (%)
43.1 419.6 4152
107.6 104.9 103.8
5.1 2.8 1.1
a The samples (50 µL) were diluted with 50 µL of drug-free plasma ultrafiltrate. b n ) 5.
Table 3. Accuracy and Precision of LQC, MQC, and HQC Samples after 3 Freeze/Thaw Cyclesa,b
LQC (40 ng/mL) MQC (400 ng/mL) HQC (4000 ng/mL)
mean (ng/mL)
accuracy (%)
precision (%)
45.7 418.7 4035
114.1 104.7 100.9
4.5 0.8 1.0
a The freeze thaw cycle consisted of thawing for 1 h at room temperature followed by freezing for 23 h or longer at -80 °C. b n ) 5.
were 108.0, 103.6, 102.9, 102.5, and 100.2%, respectively. The precision for the LLQ, LQC, MQC, HQC, and UQC were 7.9, 3.8, 0.7, 0.5, and 1.2%, respectively. These data confirmed the samples were stable enough to be stored on the autosampler for up to 3 days. After 3 months of storage, the accuracy for the LQC, MQC, and HQC were 110.4, 110.2, and 107.3%, respectively. The precision for the LQC, MQC, and HQC were 3.3, 2.8, and 2.1%, respectively. These data confirmed that ZD0473 was stable in plasma ultrafiltrate for at least 3 months if the samples were stored at -80 °C. LQC, MQC and HQC samples were analyzed after 3 freeze/ thaw cycles (Table 3). The freeze/thaw cycle consisted of thawing for 1 h at room temperature, followed by freezing for 23 h or longer at -80 °C. The accuracy for the LQC, MQC, and HQC were 114.1, 104.7, and 100.9%, respectively. Similarly, the precision for the LQC, MQC, and HQC were 4.5, 0.8, and 1.0%, respectively. Therefore, samples could be reanalyzed on two further occasions if repeat analyses were required. Suppression of ESI Signal from Plasma Ultrafiltrate. The effect of suppression of ionization from constituents of the plasma ultrafiltrate was demonstrated through days 1-3 of assay validation. The response of the MQC samples (400 ng/mL) was compared with the response from standard samples containing 400 ng of ZD0473 and 2500 ng of [2H7]ZD0473 in 1 mL of 0.15 M sodium chloride. The response of ZD0473 and [2H7]ZD0473 through 3 days of validation were 90-99% and 91-99%, respectively, of the control values. Therefore, impurities that could have caused suppression of the ESI signal eluted at a different retention time from the analytes on the YMC-AQ column. Application of the LC/MS Method to a Clinical Study. One representative cancer patient received a 150 mg/m2 dose of ZD0473 as a 1-h infusion and blood samples were taken at selected points before and after completion of the infusion. Plasma ultrafiltrate was then prepared so that protein-free drug concentrations could be determined. After conversion of all aquated species back to the parent dichloro form of ZD0473, stable isotope dilution 598 Analytical Chemistry, Vol. 74, No. 3, February 1, 2002
Figure 8. LC/ESI-SRM-MS analysis of plasma ultrafiltrate from a cancer patient infused with 150 mg/m2 ZD0473: (A) predose sample, (B) 8-h sample (54.2 ng/mL), and (C) Cmax sample at 0.92 h diluted 1:10 with blank human plasma ultrafiltrate (7228 ng/mL).
analyses were performed by LC/SRM-MS. No drug was detected in plasma ultrafiltrate at the start of the infusion (Figure 8A). It could be detected for up to 8 h after beginning of the infusion (Figure 8B), and the concentration at this time-point (54.2 ng/ mL) was well above the LLQ (20 ng/mL). The maximum concentration (Cmax) of ZD0473 in plasma ultrafiltrate from this patient (7228 ng/mL) was observed 0.92 h after the start of the infusion (Figure 8C). No ZD0473 was detected in the 24-h or 72-h samples (data not shown). A full pharmacokinetic study is currently under way in order to determine whether there is a correlation between the area under the plasma ultrafiltrate concentration/time curve (AUC) and the observed toxicity (thrombocytopenia). Such studies were not possible before the development of this LC/MS assay. CONCLUSION We have developed the first LC/MS method for the quantitative analysis of a platinum anticancer drug. Platinum drugs have found wide utility in the treatment of cancer.34,35 However, it can be difficult to develop rational dosing strategies for platinum drugs because of analytical problems associated with quantitation of the intact drug entities. The ease with which ligand exchange occurs after plasma samples have been collected adds further complications. The availability of a validated LC/MS assay for the new drug, cis-amminedichloro(2-methylpyridine)platinum(II) (ZD0473) in which the intact drug is analyzed will allow accurate and precise pharmacokinetic parameters to be obtained during phase II/III studies. This should lead to the rapid development of rational (34) Judson, I.; Kelland, L. R. Drugs 2000, 59, 29-36. (35) Johnson, S. W.; Stevenson, J. P.; O’Dwyer, P. J. Principles and Practice of Oncology, 6th ed.; DeVita, V. T., Jr., Hellman, S., Rosenberg, S. A., Eds.; Lippincott: Philadelphia, 2001; pp 376-387.
dosing strategies. The method is based upon the use of stable isotope dilution LC/ESI-SRM-MS. Plasma ultrafiltrate samples required no cleanup or derivatization. The standard curves based on the peak area ratio against the internal standard were fitted to a quadratic regression over the range 10-5000 ng/mL in human plasma ultrafiltrate. The lower limit of quantitation (LLQ) for ZD0473 was 10 ng/mL. This simple, rapid, reliable, and sensitive method of quantitation had excellent accuracy and precision for ZD0473. Only 100 µL of sample was required, but as little as 50 µL could be analyzed if necessary. The 19-min total run time (allowing for column reequilibration) made it possible to analyze 50 samples with a calibration curve in a single overnight run. The
method was sensitive enough to permit a pharmacokinetic study of ZD0473 disposition after administration of an intravenous dose of the drug to humans. It will, therefore, be employed for more detailed pharmacokinetic studies of ZD0473. ACKNOWLEDGMENT We acknowledge the support of the University of Pennsylvania Cancer Center NIH grant P30 CA16520. Received for review July 16, 2001. Accepted November 20, 2001. AC010792V
Analytical Chemistry, Vol. 74, No. 3, February 1, 2002
599
