Abstract
Little information regarding the metabolic pathways of pharmaceutical agents administered to dogs, or the inhibition of those metabolic pathways, is available. Without this information, it is difficult to assess how combinations of drugs, whether new or old or approved or nonapproved, may increase the risk for metabolic drug-drug interactions in dogs. Because mammalian xenobiotic metabolism pathways often involve the hepatic cytochrome P450 (P450) monooxgenases, canine liver microsome P450 inhibition screens were tested to evaluate the potential metabolic drug interaction risk of commonly used veterinary medicines. A probe substrate cocktail was developed for four of the five major hepatic canine P450s and used to evaluate their inhibition by 45 canine therapeutic agents in a single-point IC50 screen. Moderate inhibitors (>25%) were further characterized with an automated ninepoint IC50 assay that identified ketoconazole, clomipramine, and loperamide as submicromolar CYP2D15 inhibitors. Additional inhibitors belonged to the antiemetic, antimitotic, and anxiolytic therapeutic classes. According to the marker activities, the relative frequency of P450 inhibition by isoform followed the sequence CYP2D15 > CYP2B11 > CYP2C21/41 > CYP3A12/26 > CYP1A1/2. The findings presented suggest there is some overlap in canine and human P450 inhibition specificity. However, occasional differences may give human drugs used off-label in dogs unexpected P450 inhibition profiles and, therefore, cause an unexpected drug-drug interaction risk.
The number of reports describing the basic enzymology of drug-metabolizing enzymes in nonhuman species has increased the last decade. Notable examples include the characterization of several homologs of the major human xenobiotic-metabolizing cytochromes P450 that were cloned from canine hepatocytes (Roussel et al., 1998; Shou et al., 2003). Single nucleotide polymorphisms within canine P450s that alter activity have also been identified, as have P450 isoforms that appear to only be expressed in a fraction of the beagle population (Sakamoto et al., 1995; Blaisdell et al., 1998). More recently, three hepatic feline P450s cDNAs were characterized, two of which were expressed and characterized (Tanaka et al., 2005, 2006). Whereas some of these reports bring a better understanding of when metabolism in preclinical species may or may not reflect the human situation (Martignoni et al., 2006; Turpeinen et al., 2007), there have been few reports on how specific P450s are inhibited by or metabolize pharmaceuticals used in the animal health field.
For instance, in the development of new therapies for dogs, prediction of whether concomitant medications in the patient population may inhibit metabolic clearance pathways would be helpful in the assessment of the metabolic drug-drug interaction (DDI) potential. However, there is a lack of canine P450 inhibition data and understanding of basic canine enzymology that has caused in vitro canine drug interaction assessment to lag behind that of human health endeavors. Yet, the importance of characterizing canine P450 inhibition profiles of current and future drugs is arguably increasing. Because several canine therapeutic areas require long-term pharmaceutical intervention on the time scale of months or years (e.g., parasitology, pain and inflammation, infectious disease, behavior, cancer, cardiovascular, and endocrinology), prolonged exposure and accumulation, in addition to avid binding to drug-metabolizing enzymes, increases the risk for pharmacokinetic DDIs. In addition, the growing market for canine medicines devoted to improving quality of life rather than treating a specific disease is expected to increase the number of dogs receiving pharmaceutical agents on a regular basis (Riviere, 2007). Approvals of fluoxetine for canine separation anxiety (Simpson et al., 2007) and dirlotapide for overweight dogs (Wren et al., 2007) are recent examples. When coupled with the favorable market that has emerged for companion animals (Evans and Chapple, 2002), these observations suggest that dogs will be increasingly more likely to be administered multiple medications and for longer periods of time.
The role of hepatic P450s in metabolic drug interactions is well established under circumstances in which an inhibitor decreases the metabolism of a compound whose clearance is largely determined by metabolism. Therefore, canine liver (beagle) P450 inhibition screens were developed to evaluate the potential metabolic drug interaction risk of commonly used veterinary medicines. It is envisioned that the inhibition screens could be readily integrated into the lead optimization process early in the drug discovery process.
Materials and Methods
Chemicals. The majority of FDA-approved and nonapproved drugs used in canine veterinary practice were purchased from Sigma-Aldrich (St. Louis, MO) under the Sigma, Aldrich, Pestanal, and Fluka brands. All compounds were certified at ≥95% purity, or were acquired at the highest purity available. The salt forms of all compounds are listed below in Table 3. Cefpodoxime was received as a reference standard from Pfizer, Inc. (New York, NY). Methylprednisolone was purchased from Steraloid (Wilton, NH). Eight proprietary compounds and deracoxib were obtained from Pfizer Animal Health (Kalamazoo, MI). Canine P450 marker substrates (phenacetin, temazepam, diclofenac, bufuralol, and midazolam) and the metabolites 4-acetamidophenol and oxazepam, NADPH, DMSO, and monobasic and dibasic potassium phosphate were obtained from Sigma-Aldrich. 4′-OH-diclofenac, 1′-OH-bufuralol, and 1′-OH-midazolam were purchased from BD Biosciences (San Jose, CA). High-performance liquid chromatography-grade solvents were purchased from Honeywell Burdick & Jackson (Morristown, NJ). Formic acid (99.9%) was obtained from Pierce Chemicals (Rockford, IL).
Preparation of Substrate and Inhibitor Solutions. Marker substrates for canine P450s were dissolved in 10% acetonitrile or DMSO and then diluted with water to obtain concentrated stocks (10–20 mM) that were later diluted in reaction buffer. Stock inhibitor solutions (10 mM) were initially prepared by dissolving the inhibitors in 50:50 acetonitrile-potassium phosphate buffer (0.1 M, pH 7.4) except for l-thyroxine and cyclosporin, which were dissolved in water containing 1 molar equivalent of sodium hydroxide and 100% DMSO, respectively. All solutions of inhibitors were then prepared by diluting a 10 mM stock into phosphate buffer (0.1 M, pH 7.4) to make 1.0 or 0.1 mM solutions. Cyclosporin was only soluble up to concentrations near 30 μM in buffer. Addition of cyclosporin solutions containing liver microsomes did not appear to improve solubility based on the partial inhibition of midazolam hydroxylase activity.
Enzymes. Canine liver microsomes (20 mg/ml in 250 mM sucrose) were obtained from Xenotech LLC (Lanexa, KS). Microsomes were pooled from four male and four female beagles.
General Enzyme Incubation Method. Enzyme incubations (0.2 ml) were carried out in 0.1 M potassium phosphate, pH 7.4, with 0.1 mg/ml microsomal protein in 96-well V-bottom 30-mm-tall plates in a shaking water bath held at 37°C. MgCl2 was not included in reactions because its effects have been shown to be substrate dependent and, in some cases, inhibitory. A 10 mM stock of NADPH dissolved in buffer was used to initiate all incubations to give a final concentration of 1 mM after a 4-min preincubation/equilibration time. The length of the incubations was 8 min during which metabolite formation was demonstrated to remain linear. All incubations were quenched with 0.1 ml of acetonitrile containing 50 ng/ml triazolam as an internal standard. Organic solvents used to dissolve substrates and inhibitors only exceeded a concentration of 0.05% (v/v) in incubations that contained cyclosporine. Experiments with cyclosporin were conducted with equal volumes of DMSO in all reactions and controls because of its low solubility. DMSO did not exceed 2% of the final reaction volume in these cases.
Enzyme Kinetics. The enzyme kinetics for several P450 activities in canine liver microsomes were characterized. Ranges of substrate concentrations were as follows: 10 to 300 μM for phenacetin (CYP1A), 2 to 400 μM for temazepam (CYP2B), 5 to 500 μM for diclofenac (CYP2C), 0.5 to 500 μM for bufuralol (CYP2D), and 0.1 to 100 μM for midazolam (CYP3A). Except for midazolam, all kinetic profiles were fit using nonlinear regression to the Michaelis-Menten equation using SigmaPlot v9.0 (Systat Software, Inc., San Jose, CA). Formation of 1′-OH-midazolam demonstrated substrate inhibition and was fit to eq. 1:
Metabolite Assay. After quenching reactions, plates were dried down with nitrogen at 37°C. Samples were then dissolved in 0.1 ml of reaction buffer (0.1 M potassium phosphate, pH 7.4) by vigorous mixing in a plate vortexer for 1 min. Metabolites of the P450 substrates were assayed using a nonvalidated LC-tandem mass spectrometry method on a Thermo triple-quadrupole TSQ Quantum mass spectrometer (Thermo Scientific, Waltham, MA). The transitions monitored by LC-tandem mass spectrometry are noted below in Table 1. Samples (10 μl) were injected onto a Phenomenex 5 μ polar-RP high-performance liquid chromatography column (50 × 2 mm) equilibrated with 5% acetonitrile in water containing 0.1% formic acid (mobile phase solution A). A linear gradient from 100% mobile phase A to 100% mobile phase solution B (100% acetonitrile containing 0.1% formic acid) was initiated from 0.1 to 0.4 min and held at 100% B until 3.5 min before returning to 100% A for equilibration. Retention times for the metabolites 4-acetamidophenol, 1′-OH-bufuralol, oxazepam, 1′-OH-midazolam, and 4′-OH-diclofenac were 1.31, 3.27, 3.47, 3.49, and 3.51 min, respectively. The retention time for the internal standard triazolam was 3.53 min.
Evaluation of a Canine P450 Probe Substrate Cocktail. Substrates whose selectivity was previously characterized with recombinant canine P450s (Shou et al., 2003), except for midazolam, were coincubated in pair combinations and as cocktails containing four or five substrates. The inhibition of metabolite formation for each microsomal activity was then characterized to determine whether the potential probe substrates could be combined into a cocktail to measure multiple P450 activities in the same microsomal incubation. Substrate concentration was always held equal to the Km of each compound (see Table 2).
Single-Point IC50 for Canine Microsomal P450s. A single concentration of drug (3 μM) was used in a high-throughput approach to bin canine medicines by their potential for P450-mediated drug-drug interactions. The methodology is based largely on the work of Gao et al. (2002) wherein the inhibition at 3 μM inhibitor could be correlated to the determined IC50 of the inhibitor.
IC50 Assay. An automated assay configured on a BioMek 2000 automated workstation (Beckman Coulter, Fullerton, CA) was used to determine IC50 values using nine concentrations of inhibitor. The BioMek diluted and mixed the inhibitors with reaction buffer to 0.001, 0.01, 0.1, 1.0, 3.0, 10, 30, 100, and 300 μM or to 0.0001, 0.001, 0.01, 0.1, 0.3, 1.0, 3.0, 10, and 30 μM for potent inhibitors, when the single-point inhibition was >80%. Incubations were conducted in duplicate and quenched under the conditions stated above. Reaction velocities were averaged and normalized to unity at the lowest inhibitor concentration before fitting to a four-parameter logistic sigmoidal function using SigmaPlot: The highest inhibitor concentration used typically produced greater than ∼75% inhibition for the weakest inhibitors so that the beginning of curvature leading to the bottom baseline was evident. Use of a fixed-bottom curve function appeared to overestimate IC50 values for curves with small Hill slopes.
A custom aluminum plate holder on the deck that was adapted to the height of the 96-well plate described above was used for reactions conducted on the BioMek 2000 system. The custom holder was adapted for a temperature-regulated water circulator that was connected with ¼-inch tubing to circulate water internally through the base of the holder. Water was added to the aluminum plate holder for direct contact with the plate, and the temperature setting of the water circulator was adjusted to ensure that the temperature of reactions inside the plate wells was 37°C using a digital thermocouple.
Results
Kinetics and Cocktail Compatibility of Canine P450 Probe Substrates. The steady-state kinetics of the potential canine P450 marker substrates phenacetin, temazepam, diclofenac, bufuralol, and midazolam were evaluated in pooled canine (beagle) liver microsomes (Table 1). As demonstrated before in canine microsomes (Regmi et al., 2007), midazolam displayed substrate inhibition with canine microsomes in the formation of the 1′-OH metabolite greater than 10 μM. The substrate inhibition was characterized by a Ksi = 28 ± 4 μM, which presumably is indicative of a second substrate binding event that alters the orientation of the first midazolam. Kinetic constants were on the order of those reported earlier with minor differences in phenacetin O-demethylation velocities (Lu et al., 2005) and the diclofenac Michaelis constant (Shou et al., 2003) (Table 1).
As a means to develop a higher throughput method for measuring canine P450 inhibition, substrates were evaluated for their compatibility in cocktails. Combining multiple canine P450 probe substrates in cocktails demonstrated that several substrates could be coincubated without significantly inhibiting the metabolism of each other (Table 2). When a cocktail of phenacetin, diclofenac, bufuralol, and midazolam was used, only bufuralol metabolism was inhibited more than 10%. Including temazepam in the cocktail inhibited the turnover of diclofenac, a less selective substrate for CYP2C in the dog. The broad-spectrum control inhibitor miconazole gave similar IC50 values with single substrates and the substrate cocktail (Table 3).
Selection of Canine Veterinary Medicines for DDI Screening. Forty-five FDA-approved and nonapproved canine therapeutic agents were identified from internal sources at Pfizer Animal Health. First, a collection of roughly 60 products ranked by their market value from the year 2005 was organized, and the nutriceutical products were then eliminated. For example, although human P450 2B6 has been demonstrated to bind some fairly polar natural products (Foti et al., 2007), glucosamine, chondroitin, and S-adenosylmethionine would most likely be low risk for metabolic DDIs, unless they were found to be inducing agents of P450. Biological agents such as insulin were eliminated. Several topical products with low absorption were also eliminated (e.g., neomycin). Next, lists containing all medications administered to dogs from two Pfizer Animal Health clinical studies were combined. Emphasis was placed on compounds that give systemic exposure and are administered chronically when possible, as these would have the highest chance for metabolic DDIs. Several drugs limited to single-use applications, such as for anesthesia and postoperative pain relief, were omitted. Additional compounds were eliminated if there were already multiple analogs for the same indication. Finally, eight proprietary pharmaceutical compounds currently under investigation across six discovery projects at Pfizer Animal Health were included to compare the potential drug interaction liability of new drug leads.
Single-Point IC50 Screen. Five canine P450 marker substrates could be used to measure activity levels that are presumed to represent the P450s 1A1/2, 2B11, 2C21/41, 2D15, and 3A12/26 in only two reactions: one containing phenacetin, diclofenac, bufuralol, and midazolam and one containing temazepam (Table 2). As an initial P450 metabolic DDI screen, the five probe substrates were used at their Km concentration in microsomes, whereas canine therapeutic agents were tested at a single concentration of 3 μM. Gao et al. (2002) previously observed that a 3 μM concentration of a new chemical entity being studied as a potential P450 inhibitor correlated better to actual IC50 values than 1 or 10 μM. Except in cases for which the total dose is large or a compound is an irreversible inhibitor of P450 (Obach et al., 2006), an IC50 of ≤1 μM against a P450 has been suggested to be a useful rule of thumb in predicting the need for either more detailed DDI analysis or in vivo DDI studies. The single-point inhibitor screen was adopted to provide a simple and quick assessment of canine P450 inhibition, knowing that each discovery project possesses unique DDI risk profiles. As a percentage of all inhibitors screened, 15% of compounds were CYP1A1/2 inhibitors, 29% were CYP2B11 inhibitors, 20% were CYP2C21/41 inhibitors, 42% were CYP2D15 inhibitors, and 16% were CYP3A12/26 inhibitors using 10% inhibition as the cutoff value (Table 3). Broad-spectrum inhibitors of P450s and/or potent inhibitors may be worth further examination with respect to canine DDI risk. The antifungal ketoconazole inhibited all of the tested P450 activities in microsomes ≥25% (Table 3). Vincristine and ivermectin inhibited several activities, but to a lesser extent than the antifungals. Clomipramine, fluoxetine, and three proprietary compounds decreased bufuralol hydroxylase activity >50%. In addition, loperamide appeared to be a relatively potent inhibitor of P450 2B11, 2D15, and 3A12/26 according to marker activities.
Automated IC50 Screen. A relationship between the IC50 value of a compound for P450s and the inhibition produced by a single concentration of that compound was sought. Any correlation would support using the latter screen as a higher throughput method to identify compounds with increased drug-drug interaction risk in early discovery efforts (IC50 <1 μM). The IC50 screen was adapted to a liquid handling system so that all pipetting was automated and so that multiple plates could be run sequentially with the preparation of only one sample well for microsomes, NADPH, and quenching solution. Compounds that produced greater than 25% inhibition in the single 3 μM concentration screen were further analyzed to determine their IC50 values using a single substrate. Incomplete IC50 curves were observed when some compounds produced <25% inhibition in the single concentration screen. Therefore, it was assumed that roughly 25% inhibition was needed to more accurately define IC50 across the 0.0001 to 300 μM inhibitor concentration range. The automated inhibitor titration experiment conducted with liver microsomes allowed coverage of a wide range of IC50 values. IC50 values for the microsomal P450 activities appeared to weakly correlate with the degree of inhibition observed with the single inhibitor concentration experiments (Fig. 1). Actual IC50 values for five compounds were greater than 5-fold different from those predicted by the single-point assay according to the exponential regression. The inhibitions of phenacetin deethylation and midazolam hydroxylation by ketoconazole and midazolam hydroxylation by loperamide were overestimated. Meanwhile, the inhibitions of temazepam demethylation by miconazole and bufuralol hydroxylation by loperamide were under-predicted compared with the determined IC50 values. Sixty-three percent of the predicted IC50 values were within 2-fold of the actual IC50 value.
There appeared to be some trends within the P450 isoforms inhibited and the potency of inhibitors within therapeutic areas despite the limited data set. The antiemetics, antifungals, anxiolytics, and the antidiarrheal loperamide were the most potent canine P450 inhibitors. P450 2D15 was associated with more of the low IC50 values than were the other isoforms according to the use of bufuralol 1′-hydroxylase as a marker activity in microsomes.
Discussion
Screens for the inhibition of drug-metabolizing enzymes have become routine in human health in assessing the risk for metabolic drug interactions. Efforts are often made early in discovery to engineer out features of new drug leads that confer potent CYP3A4, CYP2D6, CYP2C9, and CYP2C19 inhibition. These human isoforms are well known for their roles in clearance and metabolic drug-drug interactions, and, in some cases, for their polymorphic nature, which can lead to population pharmacokinetic variability. However, in the animal health field, many details regarding P450 structure-function relationships and P450 expression levels in various tissues are lacking. As a step toward identifying canine therapeutic agents with potential DDI liabilities, canine liver P450 inhibition screens were evaluated with drugs commonly used in canine veterinary medicine.
Although recombinant P450s are often used in early-stage DDI screening, only a few canine isoforms are commercially available and only then with rat P450 oxidoreductase. Because the relative importance of individual canine hepatic P450s in drug metabolism is not yet well characterized, the use of canine liver microsomes should allow the measurement of several important P450 activities. The use of microsomes, however, requires that marker or so-called probe substrates have been sufficiently characterized so that the conditions in which they remain selective for particular P450s can be maintained.
The first hurdle in assessing canine P450 inhibition was choosing probe substrates for the major hepatic P450 isoforms. Cohen et al. (2003) pointed out some of the limitations of fluorogenic probe substrates, which may be exacerbated when microsomes are used. However, a disadvantage with nonfluorescent substrates is the increased analysis time required for LC-mass spectrometry detection of metabolites. By combining probe substrates into a cocktail, metabolites from all of the substrates included can be quantitated in a single LC-mass spectrometry run as long as substrates are selective and do not inhibit other P450 isoforms. Phenacetin, temazepam, diclofenac, and bufuralol were chosen as potential canine P450 probe substrates based on work conducted at Merck Research Laboratories (Shou et al., 2003; Lu et al., 2005) (Table 1). Midazolam was chosen as a marker for CYP3A activity. The comparative metabolism of midazolam by recombinant canine P450s has not been reported, but 1′-hydroxylation is a common marker activity of CYP3A isoforms evaluated in most mammalian liver microsomes including dogs (Kuroha et al., 2002; Regmi et al., 2007). Other substrates evaluated by Shou et al. (2003) that were not considered here include diazepam, dextromethorphan, and testosterone. These substrates produce multiple metabolites by multiple P450s, indicating that they may be more apt to inhibit multiple enzymes.
Coincubation of a substrate cocktail without the CYP2B11 probe temazepam demonstrated some promise. Temazepam inhibited diclofenac hydroxylation (Table 2). In addition, diclofenac was previously shown to be a high-affinity, low-velocity substrate for CYP2B11. Because the selectivity of diclofenac hydroxylation by CYP2C21/41 over CYP2B11 is somewhat low to begin with, the combination of temazepam and diclofenac may be confounding in a cocktail system (Shou et al., 2003). Furthermore, temazepam was a modest heteroactivator of midazolam hydroxylation and might mask inhibition of CYP3A12. Bufuralol hydroxylation was somewhat inhibited in the presence of midazolam, but the selectivity of this reaction catalyzed by CYP2D15 is expected to help compensate for this (Table 1). Clearly there is room for improvement in finding a more selective CYP2C21 probe and a cocktail-compatible CYP2B11 probe and in further assessing midazolam selectivity. However, as discussed below, the two microsomal incubations, one containing the cocktail and another containing temazepam, could be used to quickly assess DDI potential of new chemical entities (NCEs) for the five major canine drug-metabolizing P450s.
When a lead series is beginning to undergo absorption, distribution, metabolism, excretion, and toxicity evaluation in early drug discovery, an expedient in vitro DDI assay is desirable (Gao et al., 2002). By correlating the single-point inhibition to determined IC50 values using the same conditions and substrates, estimates of IC50 values for subsequent NCEs can be obtained (Fig. 1). In an in vitro DDI screening paradigm, assigning an IC50 threshold near 1 μM is expected to identify NCEs with the highest potential for DDIs with a few exceptions as outlined by Obach et al. (2006). The DDI risk should be weighed against all other facets of a project, such as the indication and target population. However, the 1 μM cutoff value is a straightforward indicator of whether more definitive inhibition experiments are needed. The IC50 correlation produced with beagle liver microsomes estimated that a 1 μM IC50 would be achieved at ∼60% inhibition in the single-point IC50 screen. The poor IC50 estimates for some inhibitors may have resulted from using microsomes and a substrate cocktail rather than recombinant P450s and a single substrate as originally reported (Gao et al., 2002). Automation used to carry out full IC50 determination was also beneficial for improving efficiency and reproducibility. In addition, based on the favorable results of miconazole with the substrate cocktail, as many as 16 IC50 values per plate could be determined for four NCEs using duplicate reactions with the substrate cocktail. When CYP2B was included, two NCEs could be evaluated per plate using separate reactions for CYP2B/temazepam.
With some basic P450 inhibition assays in place, an in vitro DDI screening process for canine discovery projects could be implemented. The probe substrates used here suggested that use of the single-point IC50 assay would be a reliable way to identify potent P450 inhibitors as later confirmed with nine-point IC50 determination. Furthermore, the substrates, even in a cocktail, were sensitive to compounds known to be human P450 substrates or inhibitors. Similarities with human P450s included the apparent binding of CYP1A by ondansetron, CYP2B by diazepam, CYP2C by piroxicam, CYP2D by clomipramine and fluoxetine, and CYP3A by cyclosporin and loperamide. A particularly interesting finding was the high number of potential CYP2B11 and CYP2D15 (according to marker activities) inhibitors and the relative lack of inhibitors of diclofenac and midazolam hydroxylation. Several inhibitors of CYP2D were also identified in a limited set of canine discovery leads within Pfizer.
Most compounds appeared to present little DDI risk based solely on the IC50 value, although plasma drug concentrations, dosing regimen, and mechanism-based inactivation (Venkatakrishnan and Obach, 2007) must also be considered. Most anti-inflammatory nonsteroidal anti-inflammatory drugs did not demonstrate CYP2C inhibition, as was the case with celecoxib, a canine CYP2D substrate (Paulson et al., 1999). Steroids, although often P450 substrates, usually do not have high affinity for drug-metabolizing P450s. With a few exceptions, cephalosporins, sulfonamides, fluoroquinolones (Regmi et al., 2005, 2007), anthracyclines, and cyclophosphamide (Huang et al., 2000) are not potent P450 inhibitors. H2 receptor antagonists have low affinity for P450s, but the high doses administered can lead to DDIs. Erythromycin also has low affinity for P450s but is given in large doses and is a P450 inactivator (Okudaira et al., 2007).
Only propofol, ondansetron, ketoconazole, cyclosporin, vincristine, clomipramine, diazepam, fluoxetine, and loperamide possessed low micromolar or nanomolar IC50s. The DDI potential of some of these compounds was briefly reviewed (Trepanier, 2006). Except for propofol, which was confirmed to be a CYP2B11 substrate (Hay Kraus et al., 2000), several of the compounds listed above may be administered for at least several days. For example, ketoconazole may be administered for months with high doses (5–40 mg/kg/day) and variable exposures (Plumb, 2005) and is capable of increasing the area under the curve of intravenous midazolam up to 4-fold (Kuroha et al., 2002). Ultimately, Ki values would be determined for the other potent inhibitors identified here. Ki, along with in vivo drug exposures that result from likely efficacious doses, could then be used to decide whether in vivo drug interaction studies might be warranted (Obach et al., 2005, 2006; Brown et al., 2006).
In summary, the single-point IC50 and automated multipoint IC50 assays may serve as efficient “first-round” screens in the in vitro assessment of metabolic DDI risk for dogs. The accuracy of forecasting the magnitude of the DDI in vivo will be much reduced because of knowledge gaps surrounding general canine drug metabolism and breed effects. Still, screening schemes can be proposed to help identify NCEs with more obvious DDI risk, again, depending on the indication and the knowledge of possible interacting drugs.
Acknowledgments
We thank Drs. Dawn Merritt and Steven Cox for helpful discussion.
Footnotes
-
D.A. was supported by the Pfizer, Inc. Internship Program.
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.108.021196.
-
ABBREVIATIONS: P450, cytochrome P450; DDI, drug-drug interaction; FDA, U.S. Food and Drug Administration; DMSO, dimethyl sulfoxide; OH, hydroxy; LC, liquid chromatography; NCE, new chemical entity.
- Received February 25, 2008.
- Accepted April 28, 2008.
- The American Society for Pharmacology and Experimental Therapeutics