Mol Cancer Ther. 2004;3:353-362
© 2004 American Association for Cancer Research
Cassette dosing pharmacokinetics of a library of 2,6,9-trisubstituted purine cyclin-dependent kinase 2 inhibitors prepared by parallel synthesis
Florence I. Raynaud1,
Peter M. Fischer2,
Bernard P. Nutley1,
Phyllis M. Goddard1,
David P. Lane2 and
Paul Workman1
1 Cancer Research UK Centre for Cancer Therapeutics, The Institute of Cancer Research, Sutton, Surrey, United Kingdom and
2 Cyclacel Limited, Dundee, Scotland, United Kingdom
Requests for Reprints: Florence I. Raynaud and Paul Workman, Cancer Research UK Centre for Cancer Therapeutics, The Institute of Cancer Research, 15 Cotswold Road, Sutton, Surrey, SM2 5NG, United Kingdom. Fax: 44-0-2087224324. E-mail: Florence.Raynaud{at}icr.ac.uk, Paul.Workman{at}icr.ac.uk
 |
Abstract
|
|---|
Determination of pharmacokinetic properties in the intact animal remains a major bottleneck in drug discovery. Cassette dosing involves administration of a cocktail of drugs to individual animals. Here we describe the cassette dosing properties of a 107-membered library of 2,6,9-trisubstituted purine cyclin-dependent kinase 2 (CDK2) inhibitors. A three-step parallel synthesis approach produced compounds with purity ranging from 63% to 100%. Cassette dosing was validated by comparing the pharmacokinetic parameters obtained following i.v. administration of a mixture of olomoucine, R-roscovitine (CYC202), and bohemine, each at 16.6 mg/kg, with results for administration of single agents at 50 mg/kg. No significant difference was observed between the pharmacokinetic parameters of agents when dosed in combination compared with those of individual compounds. CYC202 showed the highest area under the curve (AUC) and the longest elimination half-life (t1/2). Further cassettes evaluated the library of trisubstituted purines with CYC202 and purvalanol A included as pharmacokinetic standards in a validated limited sampling strategy. The ratios of pharmacokinetic parameters to that of CYC202 [AUC, maximum concentration (Cmax), and t1/2] remained similar when compounds were tested in two different cassettes or as individual compounds. Following dosing of the same cassette on three different days, there was less than 20% variation in pharmacokinetic parameters between days. The structure-pharmacokinetics relationship showed that the favored purine substituents are benzylamine and veratrylamine at position 6, amino-2 propanol at position 2, and methylpropyl or hydroxyethyl at position 9. Without cassette dosing, this study would have used 3 times as many animals and would have taken 4 times longer, illustrating the power of this method in lead optimization.
Key Words: trisubstituted purine • cassette dosing pharmacokinetics
|
Introduction
|
|---|
The evolution of the drug discovery process over the last decade has been the result of the considerable progress in target validation, production of new chemical entities, and screening methodology (1, 2). Advances in the understanding of cell physiology, genomics, and genetic engineering techniques have facilitated the cloning and expression of newly discovered genes, creating novel therapeutic opportunities (3). The development of robotics has impacted on both screening and chemistry. Traditional medicinal chemistry has been complemented by combinatorial chemistry generating large libraries of compounds for evaluation by high-throughput screening and downstream test cascades.
This increase in the rate of compound production and screening has resulted in a bottleneck in in vivo evaluation at the whole animal level (4, 5). The difficulty in achieving appropriate drug levels in plasma is well known to be a limiting step in lead optimization. The traditional analytical tools used to monitor absorption, distribution, metabolism, and excretion (ADME) properties have proven inadequate in dealing with the large number of compounds that now require evaluation. Advances in liquid chromatography tandem mass spectrometry (LC-MS/MS), which is both selective and sensitive, have allowed, through multiple reaction monitoring (MRM), the simultaneous measurement of several compounds in the same sample (6–8). Different approaches have been taken to improve the throughput of preclinical pharmacokinetics (PK) (6, 9–13). In cassette dosing, also known as n-in-one or cocktail dosing, a combination of drugs is administered to a single animal (11, 12, 13, 14, 15, 16). In the second method, samples are pooled after individual dosing (12). In the third approach, limited sampling strategies (even down to one time point) have been used as a predictor of area under the curve (AUC) in oral dosing studies (13). While the first approach has the inconvenience of potential drug-drug interactions, the second method still relies on extensive animal use, and the third approach would not be suitable for i.v. compounds with rapid clearance.
Although it is known that cocktail dosing is employed extensively in industry to increase throughput and hence overcome the ADME bottleneck, reducing costs and animal usage, there have been very few published examples of detailed validation of the methodology. To our knowledge, there are none in the cancer area. In the present study, we have used cassette dosing to evaluate the PK of a library of 2,6,9-trisubstituted purines. Olomoucine (Fig. 1
) is the parent compound in this class of cyclin-dependent kinase (CDK)-specific ATP antagonists (14). Bohemine and R-roscovitine (CYC202) emerged as improved analogues of OLO from studies designed to probe the structure-activity relationship within this compound class (15, 16). Purvalanol A is the most potent trisubstituted purine in terms of antiproliferative activity known to date (17). Overall, several hundred purine analogues of this type have been reported over the past decade (18–20).
Preclinical evaluation of olomoucine, bohemine, and CYC202 in terms of PK in mice xenografted with human tumors revealed that while all three compounds are taken up rapidly in tumor, liver, spleen, and kidney, plasma clearance is also very rapid (21). This is due in part to cytochrome P450-mediated hepatic metabolism of the side chain alcohol to the corresponding carboxylic acid (22). Furthermore, the in vitro growth inhibitory effect of these compounds on tumor cells depends on the length of exposure and appears to be optimal after about 16 h (i.e., approximately the duration of one cell cycle). The pharmacokinetic parameters observed, when considered in the light of the length of exposure needed for maximum activity, suggested that continuous or repeated administration might be necessary to achieve optimal in vivo antitumor activity. ADME properties are therefore of clear importance in the development of trisubstituted purine CDK2 inhibitors and overall our objective was to evaluate the potential of cassette dosing PK with these agents.
The compound set investigated (Fig. 2
) was designed in the light of known requirements for 2,6,9-trisubstituted purine CDK inhibition with particular emphasis on the substituents that would confer the best pharmacokinetic properties. The following features were incorporated in the library design: (a) modest variation of the size and apolar nature of the aryl/alkyl group on the obligatory secondary amine in R1; (b) narrow variation of R2, where only small aliphatic substituents are tolerated; (c) modest variation and conformational constriction of the optimal hydroxyethylamino motif in R3, as well as stabilization of the carbinol carbon in that substituent toward metabolic oxidation. The specific objectives of the work described in this paper were to evaluate the potential of cassette dosing PK with this class of compounds and to identify structural features that confer relatively favorable pharmacokinetic properties. Both these objectives were achieved. On the basis of the results reported here, future work will evaluate how this pharmacokinetic information can be integrated with structure-activity relationships for CDK inhibition, kinase selectivity, and cellular activity.
View larger version (17K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. Synthesis and structures of novel 2,6,9-trisubstituted purine analogues used in this study. The stepwise introduction of 6-(R1), 9-(R2), and 2-(R3) substituents used in the parallel synthesis of the analogue set starting from 2,6-dichloropurine is shown. Structures of compounds in the set are indicated as follows: {R1R2R3}. For example, the compound 2-methoxyethylamino-6-benzylamino-9-cyclopentylpurine corresponds to member {ADF}.
|
|
|
Materials and Methods
|
|---|
Synthesis
OLO, BOH (14, 16), CYC202 (23), and PUA (18) were synthesized as previously described. The purine analogues shown in Fig. 2 were prepared by simultaneous parallel solution methods, the chemistry for which was adapted from known procedures (14, 16, 18, 24–30). In short: commercial 2,6-dichloropurine (minimum 95% purity; 100 mmol scale) was aminated in refluxing n-butanol with moderate excess of amines R1-H (with auxiliary base triethylamine where necessary) for 4 h. The purine N9-H function in each of the carefully dried products (12.5 mmol scale) was then deprotonated by treatment with a slight excess of fresh sodium hydride in dry dimethylformamide, followed by reaction overnight at 75°C with the appropriate alkyl halides R2-X, under anhydrous conditions. In those cases where R2 is hydroxyethyl (analogues {XCX}), alkylation was performed with 2-bromoethoxy-tert-butyldimethylsilane (99%, Aldrich Chemical Co., Gillingham, Dorset, United Kingdom)?Q1: Please provide city and state (country, if foreign) for Aldrich Chemical Co.> under similar conditions. The O-silyl protecting groups were removed by heating in methanolic HCl after the final amination step. The 2-chloro-6,9-disubstituted purine intermediates were then reacted (2 mmol scale) with large excess of amines R3-H at 160°C for 3 h. Depending on the nature of the amines, sealed flask (no co-solvents) or open vessel (2,5-lutidine as solvent, including excess potassium carbonate addition) methods were used. Intermediates at each stage were isolated/purified by precipitation from suitable aqueous or organic solvents, extraction, or filtration through plugs of silica gel, depending on physical properties and appearance. Final products were purified by flash chromatography (31). Reaction progress was monitored using TLC. Where necessary, success of individual reactions was assessed by 1H nuclear magnetic resonance (NMR) and/or mass spectrometry analysis. The purity of the final products was measured using analytical reversed-phase high-performance (high-pressure) liquid chromatography (HPLC) (Phenomenex IB-SIL 5 µM, C-18 BDS, 250 x 4.6 mm column; 1 ml/min; linear gradient 5–60% acetonitrile (MeCN) over 20 min, then isocratic at 60% acetonitrile in 0.1% aqueous trifluoroacetic acid; 
= 260 nm). Identity was ascertained by electrospray ionization mass spectrometry.
Preparation of Dosing Solutions
Stock solutions of compounds were made in DMSO (50 mM) and diluted to the required concentrations with 1% Tween 20 in saline. The final solution contained 10% DMSO. A few compounds were eliminated from the study on the basis of poor solubility. For quantification purposes, the stock solutions in DMSO were diluted in methanol (MeOH) and subsequently spiked in mouse plasma.
Animals and Treatment Schedules
All experiments were carried out in accordance with the local institutional and national requirements (32). Female Balb C– mice (6 weeks of age) were obtained from Charles River (Margate, United Kingdom) and were acclimatized to the laboratory conditions 2 weeks before the experiment. They were allowed food (Lillico, Betchwork, Surrey, United Kingdom) and water ad libitum. The animals weighed 20 g (±1.2 g) at the time of treatment. Following transient hyperthermia to induce vasodilation, animals were injected into the tail vein with a single dose of 50 mg/kg of OLO, CYC202, or BOH in 10% DMSO, 0.5% Tween 20 in saline, or with a combination of the purines at a total dose of 50 mg/kg (16.6 mg/kg of each drug). Anesthesia was induced immediately before bleeding and maintained with 6% halothane in oxygen at a flow rate of 3 ml/min. Samples were taken at 0.25, 0.5, 1, 2, 4, 6, and 24 h post-dosing (n = 3 animals per time point). Blood was collected by cardiac puncture and was centrifuged for 10 min at 1500 x g. The plasma was decanted and frozen at –20°C until analysis within 2 weeks.
For cassette dosing of novel compounds, 166.6 nmol of each analogue (n = 5) were administered per mouse together with 166.6 nmol of either CYC202 or PUA as the PK standard. This approximates to a total dose of approximately 20 mg/kg, chosen so that solubility of the agents would not limit the number of compounds that could be administered. Plasma was collected at 0.5, 2, 4, and 6 h post-administration (n = 2 animals per time point).
Choice of the Analogues Used in the Cassette
To avoid analytical interference, compounds administered together in the same cassette were selected so as to have at least 2–3 atomic mass unit (AMU) difference. In a given cassette, it was ensured that potential metabolites of the compounds (M + 16, M + 14, M – R1, M – R2) (33) would not have the same molecular weight as any other compound in the cassette. For example, Table 1 shows the compounds administered together in cassette A. CYC202, which has the slowest plasma clearance of the three originally studied analogues, was used as the internal PK standard. In cassettes where one of the compounds had the same molecular weight as that of CYC202, the latter was replaced by PUA, which has superimposable plasma PK properties compared with that of CYC202.1 Cassettes of six compounds, including the PK standard, were dosed at 166.6 nmol per mouse.
Analytical Methods
Compound measurements were performed by LC-MS/MS. The LC system used was a Waters 600 MS, connected to a Finnigan Mat TSQ 700 triple quadrupole mass spectrometer fitted with an electrospray ionization source (Thermoquest, Herts, United Kingdom). The capillary voltage was 4.5 kV and the capillary temperature was 250°C. Chromatography was performed using a 50 x 4.6 mm zwitterionic ABZ column (Supelco, Poole, Dorset, United Kingdom); elution was with a gradient from 80% 0.1% aq formic acid-20% MeOH to 100% MeOH over 5 min, followed by a 3-min isocratic run of 100% MeOH. Standard curves of the combined analytes were made in plasma at levels of 5, 10, 50, 250, 2,500, 25,000, and 125,000 nM. One standard curve included the combination of analytes. Plasma aliquots (100 µl) were added to 30 µl of MeOH solution containing 500 ng/ml internal analytical standard (6-benzylaminopurine for the validation experiment and OLO in the cassettes), incubated for 30 min, and then mixed with 3 volumes of 100% MeOH. Following centrifugation, the supernatant was transferred to HPLC vials and 10 µl were injected onto the column. Detection was achieved with a triple sector mass spectrometer (TSQ700, Thermoquest) by MRM of selected daughter ions of the pseudomolecular ion [M + H]+ through the first quadrupole and following the induced fragmentation in Q2, the appropriate product ion was selected on Q3. The collision gas was set at 1.2 Torr and the collision energy was optimized for each compound. For each cassette, MRM was performed and the analytical interference between compounds was assessed for each analyte in the cassette.
Relative peak areas (versus the internal analytical standard) were monitored and plotted against concentration (Graphpad Software, San Diego, CA) and shown to be linear (r > 0.98). Quality controls were used throughout at concentrations of 1,000 and 10,000 nM and had to be within 15% of nominal concentration.
Pharmacokinetic Calculations
PK parameters were evaluated with the program WinNonLin Professional version 3.2 (non-compartmental analysis model 201, Pharsight, Mountain View, CA). In the different cassettes, the PK parameters, AUC (area under the concentration versus time curves to the last time point), Cmax (maximum concentration extrapolated to t0 by log-linear regression of the first two time points), t1/2 (half-life), and Vss (volume of distribution at steady state), were evaluated. The results are expressed as ratios to the value for the PK standard, allowing easy ranking and comparison of the analogues. Uniform weighting was used for the analysis.
Intra-cassette and Inter-cassette Variability
To evaluate the variability of cassette dosing, the analogues {ACC}, {BDF}, {CAE}, {DCD}, and {EDA}, which had previously been dosed in cassettes F, I, B, V, and E, respectively, were dosed again with CYC202 as internal standard (cassette Z). This cassette was dosed on three consecutive days and LC-MS/MS analysis performed on three different days. Two compounds that were dosed together in cassettes were also dosed individually for comparison ({ADB} and {CDE}).
Statistical Analysis
Comparison between all the PK parameters obtained following cassette or single dosing was carried out by Spearman's ranking test. The percentage of variation between Cmax, AUC, Vss, and t1/2 in the two dosing schedules was evaluated. The differences between AUC across dosing methods and compounds were evaluated with a method used for destructive measurement techniques (34). The Bonferroni-adjusted critical value was z = 1.96 for two groups.
|
Results
|
|---|
Synthesis
Combinatorial solution synthesis of the target 2,6,9-trisubstituted purine analogues using the well-established synthetic route (14, 16, 18, 24–30) outlined in Fig. 2 afforded the desired compounds in overall isolated average chemical yields of 20–50% over the three synthetic steps. Following isolation by precipitation and short-column chromatography, compound purity ranged between 63% and 100%. Average purity as assessed by RP-HPLC analysis was 92% and compound identity was ascertained using LC-MS and 1H NMR methods. The inclusion criteria for compounds in this study were >60% purity (by RP-HPLC), single impurities no more than 5% (by RP-HPLC), and confirmed identity of the main target component. These criteria are somewhat lower than would normally be expected of screening compounds but are consistent with library synthesis procedures where speed and throughput are the main objectives. The results are summarized in Table 2, together with PK data (see below).
Pharmacokinetics
Analytical Methodology. The standard curves for OLO, CYC202, and BOH were linear (r2 > 0.98) over the large dynamic range used (5 nM–125 µM), but this was not the case for some analogues, where linearity could only be obtained up to 10 µM. However, considering the relatively low doses used in the cassette (166 nmol), this dynamic range proved adequate for all compounds. The lowest limit of detection varied according to the compounds and was 5 nM in 10% of cases, 10 nM in 45%, 50 nM in 45%, and 250 nM in 10%. These levels are at least an order of magnitude below the concentration required to induce a 50% decrease in the cell number of A2780 human ovarian carcinoma cells after 96-h exposure as measured by the sulforhodamine B assay of the four standards (21). Of the 107 compounds analyzed, all but 4 gave a mass spectrometric signal at the correct pseudomolecular ion and all but 3 were soluble in the vehicle at the required concentration.
Validation of Cassette Dosing with OLO, CYC202, and BOH. The plasma concentration versus time curves following administration of OLO, CYC202, and BOH in combination or as separate agents were very similar (Fig. 3
). The PK parameters of the compounds administered alone or in the cassette showed a maximum variation of 28% between the two dosing methods (Table 3). The relative ranking of each of the individual PK parameters was maintained with CYC202 showing the highest plasma levels and the longest t1/2, being the only compound still detected 6 h post-administration. The PK parameters of the compounds alone or in the cassette were significantly correlated (Spearman's ranking test r = 0.99). In comparison to single dosing administration, cassette dosing was found to underestimate the CYC202 AUC and to overestimate the OLO and BOH AUCs, while underestimating the Cmax of CYC202 and bohemine and underestimating that for olomoucine. However, the ratios of the PK parameters olomoucine and bohemine to that of CYC202 in the cassettes versus a single agent were very similar, the Cmax ratio of olomoucine showing the highest variations (0.6 as a single agent versus 0.9 in a cassette). Comparing the AUC of each compound as a single agent or following cassette dosing with the Bonferroni-adjusted test showed no significant differences. However, there were significant differences between the AUC values of CYC202 and bohemine and olomoucine, P < 0.05.
View larger version (17K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. Plasma concentration versus time curves following adminstration of 50 mg/kg olomoucine, CYC202, and bohemine as single agents or 16.66 mg/kg of each in a cassette. Cassette dosing concentrations have been adjusted to 50 mg/kg assuming linear PK.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Pharmacokinetics parameters for bohemine, olomoucine, and CYC202 dosed as single agents at 50 mg/kg or in the cassette at 16.66 mg/kg
|
|
Determination of Limited Sampling Times. The PK parameters derived from sampling at 0.5, 2, 4, and 6 h with the more extensive data set described above are shown in Table 4
. The number of time points represents a balance between having sufficient points for accurate pharmacokinetic estimates (n > 3) while minimizing animal use (n < 5). This was validated in experiments where n = 8. The PK parameters derived from this limited sampling strategy were very closely related to those derived from the full sampling set for both single and cassette dosing. The greatest difference was observed for the t1/2 of BOH (1.27 versus 0.97 h). However, the ranking of the compounds was maintained, with CYC202 showing the highest AUC, the highest Cmax, and the longest t1/2.
View this table:
[in this window]
[in a new window]
|
Table 4. Pharmacokinetic parameters derived from the full data set and the limited sampling strategy (0.5, 2, 4, 6 h)
|
|
Reproducibility of PK Standards in Cassette Dosing. The variability of the PK standards was assessed. In 17 cassettes, CYC202 was used as internal PK standard, while in 5 cassettes, which included compounds of m/z = 354 (the pseudomolecular ion for CYC202) or where interference with potential metabolites could be anticipated, PUA was used as the internal PK standard. PUA has a PK profile very similar to that of CYC202 (data not shown). The concentration versus time profiles of the PK standard CYC202 in all of the 17 cassettes in which it was included were determined, and the derived PK parameters are shown in Table 5
. Up to 4-fold variation could be observed in the PK of CYC202 in different cassettes. For example, in cassette F, the AUC was 1,635 nM h compared with 6,475 nM h in cassette N. The t1/2 of CYC202 was shorter than that observed in the single dose study and in some cases could not be evaluated due to the limited number of samples detectable. The pharmacokinetics parameters observed for purvalanol A when used as the internal pharmacokinetics standard also showed some variation, with the AUC ranging from 9,077 nM h in cassette E to 3,353 nM h in cassette L. Although the AUC and Cmax appear lower and the Vss higher with CYC202 was used as internal standard, when compared with purvalanol A, there was no statistically significant difference in the AUCs between CYC202 and PUA. The pharmacokinetics of CYC202 were linear with mean AUC after cassette dosing at 5 mg/kg of 3,763 nM, 6,741 nM after 16.6 mg/kg, and 20,225 nM after 50 mg/kg.
Pharmacokinetics of Library Compounds Dosed in Cassettes. Of the compounds dosed, 15 were undetectable after dosing (Table 2, NSI), although they were stable in plasma following spiking. Another 5 compounds were only detectable at 15 min post-dosing: {ACD}, {ACE), {ACF}, {BCF}, and {EBF}. Some compounds (n = 23) showed an AUC ratio (AUC/AUCCYC202 or purvalanol A) greater than 1: {BBE}, {EDE} {CAD}, {BAD}, {CAB}, {DAB}, {AAB}, {EBB}, {DBB}, {BAF}, {BBF}, {BDF}, {ADF}, {CBC}, {EBC}, {ABC}, {ADC}, {CCA}, {ECF}, {BCC}, {DCA}, {DCD}, and {ACB}. Of these compounds, only {BBE}, {AAB}, and {BAF} exhibited increased half-lives compared with CYC202 and all the others had an increased Cmax. Overall, substitution in position 2 of the purine template with a saturated ring, such as trans-2 aminocyclohexanol or aminomethylcyclohexanol, did not increase the plasma AUC values when compared with CYC202. On the contrary, 4 out of the 20 compounds with an amino-2 propanol substituent had an AUC ratio greater than 1. A methoxyethylamino or a piperidinol substituent led to greater exposure compared to an ethyl-2-hydroxyethylamino group in 3 of 20 cases. At position 6, aromatic rings such as benzylamine and veratrylamine showed the highest AUC ratios (4 of 21 and 4 of 23 compounds with AUC ratios greater than 1), followed by aminoethylmorpholine substitution. Most of the aminoindane and aminofluorene compounds had lower AUC than the PK standard (20 of 22 and 19 of 21, respectively). Finally, hydroxyethyl and methylpropyl substituents at position 9 gave higher AUC ratios than isopropyl or cyclopentyl. An example of plasma concentration versus time curves (cassette A) is shown in Fig. 4
).
Reproducibility of Cassette Dosing. Table 6 shows the reproducibility of dosing the same cassette (Z) on three different days. The PK behavior showed a high degree of reproducibility with a maximum of 20% variation in any parameter between days. The ranking was highly maintained for all the compounds. Table 7 shows the reproducibility of the PK ratios for compounds dosed in different cassettes. Compound {DCD} was not detected in any samples following dosing in cassette Z, while previously in cassette V, there was one detectable time point. This was also the case for {CDA}. The AUC ratios of compound {CAE} were quite different between cassettes F and Z but they were both below 1. Compound {BDF} dosed in the cassettes F and Z showed exactly the same ratios. In summary, whether compounds showed a higher or lower AUC, Cmax, and t1/2 values than CYC202 in one cassette, they behaved similarly when dosed in a different cassette.
|
Discussion
|
|---|
Cassette Dosing Pharmacokinetics
The requirement for early PK and metabolism information as part of the modern discovery process is now well documented (35). It is recognized that the likely success of a drug candidate in preclinical and subsequent clinical and commercial phases of development depends very heavily on appropriate ADME/PK properties. In recent years, various in vitro and in vivo methodologies with improved throughput have accelerated ADME/PK studies, but there still remains a bottleneck, especially as a result of high-throughput screening and combinatorial chemistry. There are several ways to accelerate the PK throughput: decreasing the analytical time (6, 8, 36); reducing the animal numbers, either by dosing several compounds together, or by limiting the number of sampling points, or both (9, 10); using in silico prediction methods (37); using Caco-2 cells for absorbability predictions (38); and employing high-throughput metabolism studies employing microsomal incubations (39). Decreases in analytical time have been achieved through the use of LC-MS/MS with MRM and short columns. MRM allows simultaneous detection of several analytes in the same sample. This can be used either for the analysis of pooled samples from different experiments or alternatively as part of cassette dosing.
Cassette dosing has been introduced in recent years as an efficient way of increasing the PK throughput (6, 9–13). It involves administration of a combination of compounds, usually 5–10, to the same animal using the same dosing solution. Although there are a number of theoretical and practical challenges, cassette dosing is now used quite widely as part of the contemporary drug discovery process, especially in the pharmaceutical industry. Some examples of cassette dosing have been reported in dogs, rats, and mice (8, 40, 41). There are, however, relatively few detailed studies of its application in the scientific literature and to our knowledge, there is no published study of cassette dosing describing the PK screening of potential anticancer agents. In the present study, we have used the approach of cassette dosing in mice to study the pharmacokinetic properties of a series of trisubstituted purine compounds. These were designed as part of a project to develop inhibitors of CDK2. However, purine compounds have potentially broad therapeutic potential (26). One example of this series, CYC202 , has recently entered into phase I and II clinical trials as a potential anticancer drug (20). In addition, a large number of analogues have been synthesized, by ourselves and other groups, providing an important resource for identification of additional drug candidates. The objectives of the study were first to validate the cassette dosing methodology with known analogues and second to apply the validated method to a library of compounds. In all, a total of 107 trisubstituted purine derivatives synthesized by parallel synthesis were investigated and structure-pharmacokinetic relationships were derived from the results.
Initially, single versus cassette dosing was compared for OLO, BOH, and CYC202. Up to 28% variation in the PK parameters was recorded between single and cassette dosing. CYC202 showed consistently the highest AUC with the longest t1/2 and the highest Cmax. This was true when the three compounds were dosed individually or in combination. Further analysis of the data set showed that a limited sampling strategy consisting of 0.5, 2, 4 and 6 h sampling gave an accurate measure and ranking of the PK parameters for the compounds. This strategy was therefore used in subsequent studies.
Initially, it was thought that compounds could be administered at an approximate total dose of 50 mg/kg. However, due to limited solubility of some compounds, the cassettes had to be administered at a total dose of around 20 mg/kg (166.6 nmol of each compound per mouse). Five unknown compounds, together with a PK standard, were dosed simultaneously. There is no general consensus as to how many agents should be given together in cassette dosing. For example, up to 64 compounds have been administered together (9). However, the use of 5–10 compounds seems to be more common (42). In our study, increasing the number of compounds beyond 5 would have meant decreasing the dose administered for solubility reasons and would have required a more sensitive analysis with sample purification, thus increasing time and analytical cost.
The use of a PK standard has two goals. The first is to allow the PK properties for the compounds to be expressed as a ratio relative to the PK standard, providing a normalization that allows easy comparison of compounds between different cassettes. The second is to measure variation in the PK standard, for example due to drug-drug interactions in a cassette. There are no real guidelines as to how much variation is indicative of drug-drug interactions (4, 42). Also, in many studies, there is no assessment of the variability of the PK standard. We observed up to 4-fold variability in the AUC of our PK standard CYC202. Other investigators (41) also reported significant variability (5-fold) in the AUC of the PK standard. In addition, we observed a shorter t1/2 in the standard when compared with the compound dosed alone, which has also been reported in other studies (10). This is possibly the result of administration in the cassette at a lower dose and limitations imposed by the sensitivity of the analytical method.
The observed variability of the PK standard could have several causes. Firstly, it could be explained by the variability at any point of the whole process from animal dosing to analytical measurement. When cassette (Z) was dosed on three consecutive days, there was less than 20% variation between days for any compound. This suggests that the 4-fold variation in AUC observed with CYC202 in different cassettes could be due to drug-drug interactions. The fact that a compound could be eliminated from selection for further studies when it actually has better PK than CYC202 or PUA (false negative) would be a cause of major concern. In reality, all the compounds that were dosed in cassette Z and had previously been dosed in other cassettes, behaved similarly (i.e., had similar AUC ratios and Cmax ratios), suggesting that the test compounds were affected similarly to the PK standard. Of course, the possibility that compounds other than the PK standard could interact together cannot be totally eliminated. However, two other compounds that were dosed both alone and in different cassettes ({ADB} and {CDE}) ranked similarly to CYC202. This added to our confidence in using cassette dosing for ranking purposes.
In total, 10 compounds ranked similarly to the PK standard as single agents or in a different cassette and none of the compounds tested ranked differently. Although cassette dosing has been subject to a recent critical theoretical analysis (42), our studies have satisfied us that the approach can be used to give sufficiently reproducible PK behavior to provide an overall ranking of compounds for screening purposes. It is of course not the intention of cassette dosing to generate highly accurate PK analysis of individual compounds. The goal is efficient prioritization of compounds for further analysis (e.g., testing in a disease model). In addition, rapid feedback can be provided to the medicinal chemists as to which chemical features confer beneficial PK properties.
Structure-Pharmacokinetic Relationships
On the basis of an assessment of the PK properties of the 2,6,9-trisubstituted purine analogues by cassette dosing (Table 1), certain conclusions can be drawn regarding the structure-pharmacokinetic relationships. Of the R1 substituents (Fig. 2): all five R1 substituents A-D were found in compounds with higher drug exposures as measured by AUC. At position R2, there were no particular preferences amongst the four R2 substituents while R3 showed most of the variability in pharmacokinetics.
- R1: B > A > E > D > C
- R2: C > A = B > D
- R3: B = F > C > A = D = E
Prevention of carbinol oxidation, the major metabolic pathway of CYC202 (22), did not result in the expected increased exposure.
With respect to carbinol oxidation, comparative studies with seven analogues have shown a correlation between plasma clearance and percentage compound metabolized in microsomal incubations (data not shown).
|
Conclusions
|
|---|
We have validated the use of cassette dosing for trisubstituted purine derivatives and have used it successfully to compare the PK properties of a library of 2,6,9-trisubstituted purine derivatives prepared by parallel synthesis. The combination of parallel synthesis and cassette dosing proved to be a valuable way of gaining rapid information on the comparative PK properties of this series. This resulted in a significant 4-fold gain of time and has reduced animal usage and costs by 3-fold. To our knowledge, this is the first report using cassette dosing in drug development for oncology. Pharmacokinetics information of this type, when used alongside other pharmacological properties such as enzyme inhibitory activity and antiproliferative effect, should provide the basis for the development of novel trisubstituted purine derivatives as inhibitors of CDK2 or for other mechanisms of action. These studies are in progress. On the basis of our experience, we recommend the use of cassette dosing to evaluate the pharmacokinetic properties of novel classes of anticancer agents (43).
|
Footnotes
|
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
3 Unpublished observations. 
Received 10/31/03;
revised 12/12/03;
accepted 12/23/03.
|
References
|
|---|
Michelson S, Joho K. Drug discovery, drug development and the emerging world of pharmacogenomics: prospecting for information in a data-rich landscape. Curr Opin Mol Ther, 2000;2:651–4.[Medline]
Aherne GW, McDonald E, Workman P. Finding the needle in the haystack: why high-throughput screening is good for your health. Breast Cancer Res, 2002;4:148–54.[CrossRef][Medline]
Workman P. Scoring a bull's-eye against cancer genome targets. Curr Opin Pharmacol, 2001;1:342–52.[CrossRef][Medline]
Floyd CD, Leblanc C, Whittaker M. Combinatorial chemistry as a tool for drug discovery. Prog Med Chem, 1999;36:91–168.[Medline]
Gershell LJ, Atkins J. A brief history of novel drug discovery technologies. Nat Rev, 2003;3:320–7.
Berman J, Halm K, Adkison K, Shaffer J. Simultaneous pharmacokinetic screening of a mixture of compounds in the dog using API LC/MS/MS analysis for increased throughput. J Med Chem, 1997;40:827–9.[CrossRef][Medline]
McLoughlin DA, Olah TV, Gilbert JD. A direct technique for the simultaneous determination of 10 drug candidates in plasma by liquid chromatography-atmospheric pressure chemical ionization mass spectrometry interfaced to a Prospekt solid-phase extraction system. J Pharm Biomed Anal, 1997;15:1893–901.[CrossRef][Medline]
Olah TV, McLoughlin DA, Gilbert JD. The simultaneous determination of mixtures of drug candidates by liquid chromatography/atmospheric pressure chemical ionization mass spectrometry as an in vivo drug screening procedure. Rapid Commun Mass Spectrom, 1997;11:17–23.[CrossRef][Medline]
Beaudry F, Le Blanc JCY, Coutu M, Brown NK. In vivo pharmacokinetic screening in cassette dosing experiments: the use of online Prospekt liquid chromatography/atmospheric pressure chemical ionization tandem mass spectrometry technology in drug discovery. Rapid Commun Mass Spectrom, 1998;12:1216–22.[CrossRef][Medline]
Allen MC, Shah TS, Day WW. Rapid determination of oral pharmacokinetics and plasma free fraction using cocktail approaches: methods and application. Pharm Res, 1998;15:93–7.[CrossRef][Medline]
Adkison KK, Halm KA, Shaffer JE, Drewry D, Sinhababu AK, Berman J. Discovery of a potent and selective 
1A antagonist: utilization of a rapid screening method to obtain pharmacokinetic parameters. Pharm Biotechnol, 1998;11:423–43.[Medline]
Cai Z, Han C, Harrelson S, Fung E, Sinhababu AK. High-throughput analysis in drug discovery: application of liquid chromatography/ion-trap mass spectrometry for simultaneous cassette analysis of -1a antagonists and their metabolites in mouse plasma. Rapid Commun Mass Spectrom, 2001;15:546–50.[Medline]
Cox KA, Dunn-Meynell K, Korfmacher WA, et al. Novel in vivo procedure for rapid pharmacokinetic screening of discovery compounds in rats. Drug Discov Today, 1999;4:232–7.[CrossRef][Medline]
Vesely J, Havlicek L, Strnad M, et al. Inhibition of cyclin-dependent kinases by purine analogues. Eur J Biochem, 1994;224:771–86.[Medline]
Hajduch M, Havlicek L, Vesely J, Novotny R, Mihal V, Strnad M. Synthetic cyclin dependent kinase inhibitors: new generation of potent anti-cancer drugs. Adv Exp Med Biol, 1999;457:341–53.[Medline]
Havlicek L, Hanus J, Vesely J, et al. Cytokinin-derived cyclin-dependent kinase inhibitors: synthesis and cdc2 inhibitory activity of olomoucine and related compounds. J Med Chem, 1997;40:408–12.[CrossRef][Medline]
Gray NS, Wodicka L, Thunissen A-M, et al. Exploiting chemical libraries, structure, and genomics in the search for kinase inhibitors. Science, 1998;281:533–8.[Abstract/Free Full Text]
Chang YT, Gray NS, Rosania GR, et al. Synthesis and application of functionally diverse 2,6,9-trisubstituted purine libraries as CDK inhibitors. Chem Biol, 1999;6:361–75.[CrossRef][Medline]
Gray N, Détivaud L, Doerig C, Meijer L. ATP-site directed inhibitors of cyclin-dependent kinases. Curr Med Chem, 1999;6:859–75.[Medline]
Fischer PM. Recent advances and new directions in the discovery and development of cyclin-dependent kinase inhibitors. Curr Opin Drug Discov Dev, 2001;4:623–34.[Medline]
Raynaud FI, Nutley BP, Goddard P, et al. Pharmacokinetics of the cyclin dependent kinase inhibitors Olomoucine, CYC201 and CYC202 in Balb C– mice after iv administration. Clin Cancer Res, 1999;5:541.
Nutley BP, Fischer P, Raymond F et al. Metabolism of the cyclin dependent kinase inhibitors olomoucine, roscovitine and bohemine in Balb C mice. Proc Am Assoc Cancer Res, 2000;41:702.
Wang S, McClue SJ, Ferguson JR, et al. Synthesis and configuration of the cyclin-dependent kinase inhibitor roscovitine and its enantiomer. Tetrahedron Asymmetry, 2001;12:2891–4.[CrossRef]
Oh C-H, Kim H-K, Lee S-C, et al. Synthesis and biological properties of C-2, C-8, N-9 substituted 6-(3-chloroanilino)-purine derivatives as cyclin-dependent kinase inhibitors. Part II. Arch Pharm, 2001;334:345–50.[CrossRef]
Norman TC, Gray NS, Koh JT, Schultz PG. A structure-based library approach to kinase inhibitors. J Am Chem Soc, 1996;118:7430–1.[CrossRef]
Schow SR, Mackman RL, Blum CL, et al. Synthesis and activity of 2,6,9-trisubstituted purines. Bioorg Med Chem Lett, 1997;7:2697–702.[CrossRef]
Imbach P, Capraro H-G, Furet P, Mett H, Meyer T, Zimmermann J. 2,6,9-Trisubstituted purines: optimization towards highly potent and selective Cdk1 inhibitors. Bioorg Med Chem Lett, 1999;9:91–6.[CrossRef][Medline]
Legraverend M, Ludwig O, Bisagni E, et al. Synthesis and in vitro evaluation of novel 2,6,9-trisubstituted purines acting as cyclin-dependent kinase inhibitors. Bioorg Med Chem, 1999;7:1281–93.[CrossRef][Medline]
Fiorini MT, Abell C. Solution-phase synthesis of 2,6,9-trisubstituted purines. Tetrahedron Lett, 1998;39:1827–30.[CrossRef]
Dorff PH, Garigipati RS. Novel solid-phase preparation of 2,6,9-trisubstituted purines for combinatorial library generation. Tetrahedron Lett, 2001;42:2771–3.[CrossRef]
Still WC, Kahn M, Mitra A. Rapid chromatographic technique for preparative separations with moderate resolution. J Org Chem, 1978;43:2923–5.[CrossRef]
Workman P, Balmain A, Hickman JA, et al. UKCCCR guidelines for the welfare of animals in experimental neoplasia. Lab Anim, 1988;22:195–201.[Free Full Text]
Nutley BP, Raynaud FI, Wilson SC, et al. Comparative metabolism of the cyclin dependent kinase inhibitor roscovitine and the deuterated analogue d9-roscovitine in Balb C– mice. Clin Cancer Res, 2000;6:318.
Bailer AJ. Testing for the equality of area under the curves when using destructive measurement techniques. J Pharmacokinet Biopharm, 1988;16:303–9.[CrossRef][Medline]
Eddershaw PJ, Dickins M. Advances in in vitro drug metabolism screening. Pharm Sci Technol Today, 1999;2:13–9.[CrossRef][Medline]
Romanyshyn L, Tiller PR, Hop CECA. Bioanalytical applications of "fast chromatography" to high-throughput liquid chromatography/tandem mass spectrometric quantitation. Rapid Commun Mass Spectrom, 2000;14:1662–8.[Medline]
Ekins S, Waller CL, Swaan PW, Cruciani G, Wrighton SA, Wikel JH. Progress in predicting human ADME parameters in silico. J Pharmacol Toxicol Methods, 2001;44251–72.
Bu H-Z, Poglod M, Micetich RG, Khan JK. High-throughput Caco-2 cell permeability screening by cassette dosing and sample pooling approaches using direct injection/on-line guard cartridge extraction/tandem mass spectrometry. Rapid Commun Mass Spectrom, 2000;14:523–8.[CrossRef][Medline]
Bertrand M, Jackson P, Walther B. Rapid assessment of drug metabolism in the drug discovery process. Eur J Pharm Sci, 2000;11:S61–72.
Frick LW, Higton DM, Wring SA, Wells-Knecht KJ. Cassette dosing: rapid estimation of in vivo pharmacokinetics. Med Chem Res, 1998;8:472–7.
Shaffer JE, Adkison KK, Halm K, Hedeen K, Berman J. Use of "N-in-One" dosing to create an in vivo pharmacokinetics database for use in developing structure-pharmacokinetic relationships. J Pharm Sci, 1999;88:313–8.[CrossRef][Medline]
White RE, Manitpisitkul P. Pharmacokinetic theory of cassette dosing in drug discovery screening. Drug Metab Dispos, 2001;29:957–66.[Abstract/Free Full Text]
Smith NF, Hayes A, James K, et al. Pharmacokinetic and metabolism studies of a novel synthetic series of heat shock protein 90 (HSP90) inhibitors. Clin Cancer Res, 2003;9:239.
This article has been cited by other articles:
|
|
|
|
 
K. Bettayeb, H. Sallam, Y. Ferandin, F. Popowycz, G. Fournet, M. Hassan, A. Echalier, P. Bernard, J. Endicott, B. Joseph, et al.
N-&-N, a new class of cell death-inducing kinase inhibitors derived from the purine roscovitine
Mol. Cancer Ther.,
September 1, 2008;
7(9):
2713 - 2724.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
S. Mohapatra, D. Coppola, A. I. Riker, and W. J. Pledger
Roscovitine Inhibits Differentiation and Invasion in a Three-Dimensional Skin Reconstruction Model of Metastatic Melanoma
Mol. Cancer Res.,
February 1, 2007;
5(2):
145 - 151.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
N. F. Smith, F. I. Raynaud, and P. Workman
The application of cassette dosing for pharmacokinetic screening in small-molecule cancer drug discovery
Mol. Cancer Ther.,
February 1, 2007;
6(2):
428 - 440.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
N. F. Smith, A. Hayes, K. James, B. P. Nutley, E. McDonald, A. Henley, B. Dymock, M. J. Drysdale, F. I. Raynaud, and P. Workman
Preclinical pharmacokinetics and metabolism of a novel diaryl pyrazole resorcinol series of heat shock protein 90 inhibitors.
Mol. Cancer Ther.,
June 1, 2006;
5(6):
1628 - 1637.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
P. WORKMAN
Drugging the Cancer Kinome: Progress and Challenges in Developing Personalized Molecular Cancer Therapeutics
Cold Spring Harb Symp Quant Biol,
January 1, 2005;
70(0):
499 - 515.
[Abstract]
[PDF]
|
|
|
|
|
|
|
B. P. Nutley, F. I. Raynaud, S. C. Wilson, P. M. Fischer, A. Hayes, P. M. Goddard, S. J. McClue, M. Jarman, D. P. Lane, and P. Workman
Metabolism and pharmacokinetics of the cyclin-dependent kinase inhibitor R-roscovitine in the mouse
Mol. Cancer Ther.,
January 1, 2005;
4(1):
125 - 139.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
