During the last 10-15 years, cytochrome P450 (CYP) 2C8 has emerged as an important drug-metabolizing enzyme. CYP2C8 is highly expressed in human liver and is known to metabolize more than 100 drugs. CYP2C8 substrate drugs include amodiaquine, cerivastatin, dasabuvir, enzalutamide, imatinib, loperamide, montelukast, paclitaxel, pioglitazone, repaglinide, and rosiglitazone, and the number is increasing. Similarly, many drugs have been identified as CYP2C8 inhibitors or inducers. In vivo, already a small dose of gemfibrozil, i.e., 10% of its therapeutic dose, is a strong, irreversible inhibitor of CYP2C8. Interestingly, recent findings indicate that the acyl-β-glucuronides of gemfibrozil and clopidogrel cause metabolism-dependent inactivation of CYP2C8, leading to a strong potential for drug interactions. Also several other glucuronide metabolites interact with CYP2C8 as substrates or inhibitors, suggesting that an interplay between CYP2C8 and glucuronides is common. Lack of fully selective and safe probe substrates, inhibitors, and inducers challenges execution and interpretation of drug-drug interaction studies in humans. Apart from drug-drug interactions, some CYP2C8 genetic variants are associated with altered CYP2C8 activity and exhibit significant interethnic frequency differences. Herein, we review the current knowledge on substrates, inhibitors, inducers, and pharmacogenetics of CYP2C8, as well as its role in clinically relevant drug interactions. In addition, implications for selection of CYP2C8 marker and perpetrator drugs to investigate CYP2C8-mediated drug metabolism and interactions in preclinical and clinical studies are discussed.
Cytochrome P450 (CYP) 2C8 accounts for approximately 6–7% of the total hepatic CYP content (Rowland Yeo et al., 2004; Inoue et al., 2006; Rostami-Hodjegan and Tucker, 2007; Achour et al., 2014). The importance of CYP2C8 causing variation in drug response via drug-drug interactions and pharmacogenetic polymorphisms has been recognized only for the last 10–15 years. In the beginning of the millennium, the pharmacokinetic drug-drug interaction between the fibric acid derivative gemfibrozil and the 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitor cerivastatin, a CYP2C8 substrate, resulting in rhabdomyolysis cases and fatalities brought attention to the importance of CYP2C8 in drug metabolism (Backman et al., 2002; Staffa et al., 2002; Wang et al., 2002; Chang et al., 2004; Huang et al., 2008). The event was the onset of a broadening scientific interest in CYP2C8, promptly convincing drug regulatory authorities to acknowledge CYP2C8 as one of the major drug-metabolizing CYP enzymes.
Drugs that were introduced into clinical use before the role of CYP2C8 was recognized may have deficient or incorrect product information regarding their interaction potential (Neuvonen, 2012). One example is the leukotriene receptor antagonist montelukast. Early preclinical studies concluded that CYP2C9 and CYP3A4 are the most important enzymes involved in its metabolism, whereas the role of CYP2C8 was not evaluated (Chiba et al., 1997). In interaction studies performed more than a decade later, the strong CYP2C8 inhibitor gemfibrozil increased the plasma exposure to montelukast almost fivefold, whereas the strong CYP3A4 inhibitor itraconazole had no significant effect on its pharmacokinetics (Karonen et al., 2010, 2012). The clinical findings were corroborated by in vitro data, showing that CYP2C8 is the main enzyme involved in the oxidative metabolism of montelukast (Filppula et al., 2011; VandenBrink et al., 2011).
The large, trifurcated active site cavity of CYP2C8 is able to accommodate substrates of different shapes and sizes (Schoch et al., 2008). Today, CYP2C8 is known to participate in the metabolism of more than 100 drugs, including amodiaquine, cerivastatin, dasabuvir, enzalutamide, imatinib, loperamide, montelukast, paclitaxel, pioglitazone, repaglinide, and rosiglitazone. The number of drugs that are identified as CYP2C8 substrates or inhibitors, as well as CYP2C8-mediated drug-drug interactions is continuously increasing. The strong CYP2C8 inhibition by gemfibrozil observed in vivo is due to its acyl-β-glucuronide metabolite, which is a potent mechanism-based inhibitor of CYP2C8 (Ogilvie et al., 2006). Also other glucuronide metabolites of drugs were recently found to interact with CYP2C8 either as substrates or inhibitors. For instance, the acyl-β-d-glucuronide metabolite of clopidogrel is a metabolism-dependent inhibitor of CYP2C8, causing more than a fivefold increase in the plasma exposure to repaglinide in healthy subjects (Tornio et al., 2014). In another recent in vitro study, CYP2C8 metabolized the glucuronide metabolite of desloratadine to its pharmacologically active 3-hydroxydesloratadine metabolite (Kazmi et al., 2015). These and other data suggest that an interplay between CYP2C8 and glucuronide metabolites may be more a rule than an exception.
There are several common nonsynonymous variations in the CYP2C8 gene (Daily and Aquilante, 2009; Aquilante et al., 2013b). For example, the CYP2C8*3 allele has been associated with decreased metabolism of several substrates, e.g., paclitaxel, in vitro (Dai et al., 2001). In contrast, clinical data indicate that the CYP2C8*3 allele is often associated with increased metabolism of CYP2C8 substrates, such as repaglinide (Niemi et al., 2003c). Thus, although complete lack of function variants in CYP2C8 are rare, the possible substrate dependency of the functional consequences of the common CYP2C8 variants and their potential clinical significance have raised a lot of interest toward CYP2C8 pharmacogenetics.
The structural properties, regulation of expression, pharmacogenetics, substrates, inhibitors, and physiologic roles of CYP2C8 have been thoroughly examined and discussed in several previous reviews (Kirchheiner et al., 2005; Totah and Rettie, 2005; Garcia-Martin et al., 2006; Gil and Gil Berglund, 2007; Agundez et al., 2009; Chen and Goldstein, 2009; Daily and Aquilante, 2009; Lai et al., 2009; Aquilante et al., 2013b; Fleming, 2014; Xiaoping et al., 2013). Thus, the reader will be directed to these earlier works for some previously known aspects related to CYP2C8. Herein, our intention is to review and update the current knowledge on substrates, inhibitors, inducers, and pharmacogenetics of CYP2C8, as well as its role in clinically relevant drug interactions. In addition, implications for selection of CYP2C8 marker and perpetrator drugs to investigate CYP2C8-mediated drug metabolism and interactions in preclinical and clinical studies are discussed.
II. Basic Characteristics of Cytochrome P450 2C8
A. Genomic Organization and Transcriptional Regulation
The CYP2C8 enzyme is encoded by the CYP2C8 gene, which is located on the chromosome 10q24 in the 2C gene cluster centromere-2C18-2C19-2C9-2C8-telomere in close proximity of the CYP2C9 gene (Fig. 1; Gray et al., 1995; Klose et al., 1999). CYP2C8 is the smallest of the human CYP2C genes; it spans a 31-kb region and contains 9 exons (Klose et al., 1999; Lai et al., 2009). It shares 74% sequence homology with CYP2C9 (Daily and Aquilante, 2009).
The transcriptional regulation of CYP2C8 is mediated via several transcriptional factors and distinct nuclear receptors that can activate the respective responsive elements within the 5′-flanking promoter region of the gene (Ferguson et al., 2005; Johnson and Stout, 2005; Kojima et al., 2007; Chen and Goldstein, 2009). Such factors/receptors include the constitutive androstane receptor (CAR), pregnane X receptor (PXR), vitamin D receptor (VDR), glucocorticoid receptor (GR), hepatic nuclear factor-4α (HNF4α), HNF3γ, CCAAT/enhancer-binding protein α (C/EBPα), and retinoic acid-related orphan receptors (RORs) (Fig. 1; Ferguson et al., 2005; Chen and Goldstein, 2009; Rana et al., 2010; Aquilante et al., 2013b). Although HNF4α, HNF3γ, C/EBPα, and RORs seem to mainly regulate the constitutive expression of CYP2C genes in liver, the other receptors are more important to the xenobiotic-mediated induction of CYP2C8 expression, as described in more detail in section V.C.
After activation by endo- or xenobiotics, CAR, PXR, and VDR form heterodimers with the retinoid X receptor, whereas GR forms homodimers (Chen and Goldstein, 2009). These dimers are thereafter recognized by specific response elements within the CYP2C8 promoter. By using in vitro gel shift assays, responsive elements/motifs within the CYP2C8 promoter regions have been identified for CAR, PXR, and GR (Gerbal-Chaloin et al., 2002; Ferguson et al., 2005; Chen and Goldstein, 2009).
After activation, the orphan nuclear receptor HNF4α binds as a homodimer to a DR1 type element and also to the Hep-G2 specific P450 factor-1 motif (Venepally et al., 1992), whereas HNF3γ binds to DNA as a monomer (Bort et al., 2004). At least two Hep-G2 specific P450 factor-1 motifs and several putative HNF3γ binding sites have been identified within the promoter of CYP2C8 (Bort et al., 2004; Ferguson et al., 2005; Chen and Goldstein, 2009). RORs are constitutively active orphan nuclear receptors, which have natural ligands, such as all-trans-retinoic acid that can influence their activity. Also RORs seem to be involved in the constitutive regulation of CYP2C8, and at least two ROR responsive elements have been identified in the gene promoter (Chen et al., 2009). Transcriptional regulation of CYP2C8 has been reviewed thoroughly by Chen and Goldstein (2009).
B. Protein Structure
The crystal structure of CYP2C8 was resolved in 2004 (Schoch et al., 2004). A single CYP2C8 crystal diffracted to 2.7 Å and had the molecular weight approximated to 54 kDa. Interestingly, CYP2C8 crystallized as a symmetric dimer formed by interactions between the helix F to G regions of the two monomers. Two palmitic acid molecules were bound in the dimer interface, stabilizing the dimer. Thus, the two fatty acids may form a peripheral binding site, which may affect the structural dynamics of the active site and influence reactions catalyzed by CYP2C8. The active site volume of CYP2C8 was estimated to 1,438 Å3 (Schoch et al., 2004), which is similar to that of CYP3A4 (1,386 Å3) but larger than those of CYP1A2 (375 Å3), CYP2A6 (260 Å3), CYP2C9 (∼470 Å3), CYP2D6 (∼540 Å3), and CYP2E1 (190 Å3) (Williams et al., 2003; Yano et al., 2004; Rowland et al., 2006; Sansen et al., 2007; Porubsky et al., 2008). Although CYP3A4 has a uniformly distributed active site cavity, that of CYP2C8 is trifurcated, resembling a T or Y shape (Schoch et al., 2008). The bottom branch of the cavity provides access to the heme, and the two other terminate in solvent and substrate access channels that exit the active site cavity on either side of the helix B-C loop.
In the X-ray crystallography study, the N-terminal anchor domains of both CYP2C8 molecules were located on the same side of the dimer, indicating an orientation compatible with membrane binding (Schoch et al., 2004). The proximal surfaces of each protein were roughly parallel, suggesting that they are accessible for interaction with the membrane-bound CYP oxidoreductase. Another study demonstrated that the dimeric structure observed in the crystal structure of CYP2C8 may also be present in membrane-bound native CYP2C8 (Hu et al., 2010). The signal anchor/linker regions of native CYP2C8 formed a second dimerization interface, and it was suggested that this interaction is required for the formation of the dimer of the native protein. Although direct evidence for a functional significance of the dimerization is lacking, such interactions have been shown to affect activities of other CYPs and membrane proteins in the endoplasmic reticulum (Hu et al., 2010).
According to the X-ray crystallographic data, the active site cavity of CYP2C8 is capable of binding structurally diverse substrates without major changes in its tertiary structure (Schoch et al., 2008). Ligands of CYP2C8 may bind to the active site differently, filling the cavity either partially or completely or occupying it with two molecules simultaneously. For instance, montelukast, a large anionic molecule with a tripartite structure, complemented the size and shape of the whole active-site cavity. The linearly shaped troglitazone molecule occupied the upper portion of the cavity, leaving a significant part of the cavity empty, whereas two molecules of 9-cis-retinoic acid were simultaneously present in the substrate-binding cavity of CYP2C8 (Schoch et al., 2008). The interactions between CYP2C8 and its substrates were predominantly hydrophobic.
In addition, the distal region of the CYP2C8 active site cavity contains a number of polar amino acid side chains and exposed peptide backbone hydrogen bond donors and acceptors (Schoch et al., 2008). Accordingly, for example, the residues Ser-100, Ser-103, Asn-204, Asn-217, and Arg-241 form hydrogen bonds involved in the binding of the CYP2C8 substrates retinoic acid, troglitazone and montelukast. Of note, a pronounced side chain movement was observed in crystallized complexes with troglitazone and retinoic acid, where Arg-241 was reoriented to the inside of the cavity, where it could provide a strong, charge-stabilized hydrogen bond with the substrate. Interestingly, according to computational docking simulations, the glucuronide moieties of gemfibrozil 1-O-β glucuronide and clopidogrel acyl 1-β-d-glucuronide are oriented toward the same hydrophilic area in the active site close to helix B′, where Ser-100 and Ser-103 reside (Fig. 2; Baer et al., 2009; Tornio et al., 2014).
Although the large active site of CYP2C8 and diversity of its substrates (section III) may complicate the use of a general pharmacophore model, analysis of eight CYP2C8 substrates showed that the majority of these compounds contained a terminal anionic or polar group ∼13 Å from the oxidation site, and one or two secondary polar moieties ∼4.5 Å and ∼8.5 Å from the oxidation site (Melet et al., 2004). The pharmacophore model and previously reported homology models for CYP2C8 have been comprehensively reviewed by Lai and colleagues (2009).
According to meta-analyses, the mean hepatic CYP2C8 concentration approximates to 22–24 pmol/mg and 14 pmol/mg in adult Caucasian and Japanese livers, respectively (Rowland Yeo et al., 2004; Inoue et al., 2006; Rostami-Hodjegan and Tucker, 2007; Achour et al., 2014). The interindividual variability of CYP2C8 protein expression in liver is high, with coefficients of variation of 68–95%. The protein expression level of CYP2C8 seems to be highly correlated with both its enzyme activity and messenger ribonucleic acid (mRNA) expression level (Ohtsuki et al., 2012).
Hepatic CYP2C8 mRNA and protein are expressed early in the prenatal development and reaches adult levels already in early childhood (Treluyer et al., 1997; Blanco et al., 2000; Naraharisetti et al., 2010; Cizkova et al., 2014; Johansson et al., 2014). CYP2C8 seems to be the predominant CYP2C isoform in fetal livers (Hakkola et al., 1994; Nishimura et al., 2003; Johansson et al., 2014). In a recent study, CYP2C8 mRNA was expressed in all fetal tissues studied (adrenal, kidney, liver, and lung tissue), whereas CYP2C9 mRNA was restricted to the liver (Johansson et al., 2014). Another study detected CYP2C8, CYP2C9, and CYP2C19 protein in fetal liver, intestine, and kidney (Cizkova et al., 2014). One explanation for the role of CYP2C8 in the fetus during early pregnancy may be a need for CYP2C8 to metabolize endogenous compounds such as retinoic acids and hence protect the fetus from retinoic acid-induced embryotoxicity (Johansson et al., 2014).
In adults, CYP2C8 mRNA has been detected in numerous extrahepatic tissues, including the adrenal gland, arteries, brain, duodenum, heart, kidney, lung, mammary gland, ovary, prostate, retina; testis, and uterus, but not in placenta (Zeldin et al., 1995; Mace et al., 1998; McFayden et al., 1998; Klose et al., 1999; Thum and Borlak, 2000; Nishimura et al., 2003; Delozier et al., 2007; Dutheil et al., 2009; Capozzi et al., 2014). CYP2C8 protein has been detected in heart, hepatocytes, kidney, salivary ducts, small and large intestine, adrenal cortical cells, and tonsils (Läpple et al., 2003; Enayetallah et al., 2004; Delozier et al., 2007; Cizkova et al., 2014). The expression of CYP2C8 and other CYP enzymes has recently been reviewed by Shahabi et al. (2014).
Analysis of liver samples has recently shown that a nearly full-length form of CYP2C8 (wild type) and an N-terminal truncated splice variant 3 are expressed in mitochondria (Bajpai et al., 2014). Although the wild-type protein was detected only at low levels in mitochondria (<25%), variant 3 was primarily targeted to mitochondria and minimally to the endoplasmic reticulum. Interestingly, although molecular modeling showed that both the heme binding pocket and the substrate binding cavity were nearly intact in variant 3, it was unable to catalyze paclitaxel 6-hydroxylation in human hepatocellular liver carcinoma cells. However, it did metabolize smaller substrates such as arachidonic acid and dibenzylfluorescein. Furthermore, the variant generated higher levels of reactive oxygen species and showed a higher level of mitochondrial respiratory dysfunction than wild type CYP2C8, suggesting that the mitochondrially targeted variant 3 may contribute to oxidative stress in tissues (Bajpai et al., 2014).
In living organisms, CYP enzymes undergo natural degradation that can be described as a first-order process (Yang et al., 2008). Therefore, the expression level of the enzyme is determined by the rate of enzyme synthesis and the degradation half-life of the enzyme. The extent and dose and time dependency of enzyme induction and inactivation are thus also dependent on the degradation half-life. Based on clinical studies with the CYP2C8 inactivator gemfibrozil, the degradation (turnover) half-life of CYP2C8 is approximately 22 hours (Backman et al., 2009).
III. Substrates of Cytochrome P450 2C8
CYP2C8 participates in the metabolism of numerous drugs and some endogenous and natural compounds. It catalyzes a variety of oxidative reactions, in particular hydroxylations, N-demethylations, and N-deethylations (Tables 1–4). Because of its large, sinuous active site, CYP2C8 can accommodate substrates of different sizes and structures. The molecular weight of drugs significantly metabolized by CYP2C8 (>20%; Table 1) ranges from 206 to 854 g/mol, with a median of 451 g/mol (Fig. 3).
Most, if not all, of the drugs significantly metabolized by CYP2C8 are also substrates of other CYP enzymes (Table 1), with about 75% being metabolized by CYP3A4 and ∼30% by CYP2C9. However, the metabolic products generated by CYP2C8 and CYP3A4 are often different. For instance, CYP2C8 metabolizes paclitaxel to 6α-hydroxypaclitaxel, whereas CYP3A4 exclusively generates 3′-hydroxypaclitaxel (Rahman et al., 1994), suggesting that compounds that are substrates of both CYP2C8 and CYP3A4 bind differently to their active sites.
CYP2C8 is involved in the metabolism of more than 100 clinically used drugs (Table 1). Typical substrate drugs of CYP2C8 include anticancer, antidiabetic, antimalarial, and lipid-lowering agents (Fig. 4). Interestingly, some glucuronide metabolites of drugs interact with CYP2C8.
1. Anticancer Agents.
The antimicrotubule agent paclitaxel with a molecular weight of 853.9 g/mol is one of the largest substrates of CYP2C8. In vitro, paclitaxel is primarily metabolized by CYP2C8 to its main 6α-hydroxy metabolite and by CYP3A4 to 3′-phenyl-hydroxypaclitaxel, and the further metabolism of these metabolites results in the formation of 6α, 3′-p-dihydroxypaclitaxel (Cresteil et al., 1994, 2002; Harris et al., 1994; Kumar et al., 1994; Rahman et al., 1994). Paclitaxel 6α-hydroxylation is recommended by drug authorities as a marker reaction for CYP2C8 activity in vitro (EMA, 2012b; http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/ucm093664.htm), and it has been widely used in preclinical studies. In a mass balance study in patients, metabolites accounted for about 40% of the total systemic drug exposure, and the excreted 6α-hydroxypaclitaxel corresponded to almost one-third of the administered dose (Walle et al., 1995), suggesting that ∼30–40% of a paclitaxel dose is converted by CYP2C8 to 6α-hydroxy paclitaxel.
Cabazitaxel, a taxane approved in 2010, is also metabolized by CYP2C8 to a small extent in vitro (FDA, 2010a), whereas there is conflicting data regarding the role of CYP2C8 in the metabolism of docetaxel. One in vitro study demonstrated that docetaxel at high concentrations was metabolized by CYP2C8, but no enzyme kinetic parameters were determined (Komoroski et al., 2005). An earlier study, however, suggested that CYP2C8 is unable to accommodate docetaxel in its active site because of the absence of a side chain in the docetaxel molecule (Cresteil et al., 2002). The side chain, which is present in the paclitaxel molecule, is required for a correct orientation of it into the active site of CYP2C8.
The androgen receptor antagonist enzalutamide, indicated for treatment of castration-resistant prostate cancer, is mainly metabolized by CYP2C8 and CYP3A4/5 to enzalutamide M6 in vitro (FDA, 2012k). Then, M6 is further metabolized by CYP2C8 to the active metabolite N-demethyl enzalutamide (M2), which accounts for approximately 50% of the total drug exposure in plasma. CYP2C8 seems to be the predominant enzyme involved in enzalutamide pharmacokinetics also in vivo (section VI.C.5; Gibbons et al., 2015).
Several protein kinase inhibitors are metabolized by CYP2C8 to various degrees in vitro (Table 1). In vitro, the majority (60–70%) of dabrafenib, a selective BRAF inhibitor, is metabolized by CYP2C8, and a smaller part by CYP3A4 (∼25%) and CYP2C9 (≤10%) (Lawrence et al., 2014). Recombinant CYP2C8 produced only hydroxydabrafenib, whereas CYP3A4 formed both hydroxydabrafenib and carboxydabrafenib. However, the in vivo importance of CYP2C8 in dabrafenib pharmacokinetics seems to be quite modest (section VI.C.5; Suttle et al., 2015). Imatinib, the first tyrosine kinase inhibitor approved for clinical use, is metabolized by several CYP enzymes in vitro, with CYP2C8 and CYP3A4 being the most important ones (Nebot et al., 2010; Filppula et al., 2013a). CYP3A4 is involved in several metabolic pathways of imatinib, whereas CYP2C8 only catalyzes the formation of the main metabolite, N-demethylimatinib (Table 4; Rochat et al., 2008; Filppula et al., 2013a). The relative roles of CYP2C8 and CYP3A4 in the in vivo pharmacokinetics of imatinib are complex (section VI.C.5; Filppula et al., 2013b). In vitro, the multitargeted tyrosine kinase inhibitor ponatinib is mainly metabolized by CYP3A4, followed by CYP2C8, CYP2D6, and CYP3A5. The contributions of CYP3A4 and CYP2C8 to the in vivo elimination of ponatinib are estimated to 34% and 19%, respectively (FDA, 2012e). Furthermore, CYP2C8 plays a major role in the N-demethylation of the aurora kinase inhibitor tozasertib (Table 4; Ballard et al., 2007). With the exception of dabrafenib and imatinib, the in vivo importance of CYP2C8 in the metabolism of protein kinase inhibitors seems to have been poorly studied.
In addition, CYP2C8 participates to various degree to the metabolism of several other anticancer agents, as listed in Table 1.
2. Antidiabetic Agents.
The nonsulfonylurea insulin secretagogue repaglinide is metabolized primarily by CYP2C8 (Table 4) and CYP3A4, but it also undergoes direct glucuronidation by uridine-5′-diphosphoglucuronosyltransferase (UGT) 1A1 (Bidstrup et al., 2003; Kajosaari et al., 2005a,b; Gan et al., 2010). In addition, there is in vitro data suggesting that aldehyde dehydrogenase is involved in its metabolism (Säll et al., 2012). Moreover, repaglinide is a substrate of the hepatic uptake transporter organic anion-transporting polypeptide (OATP) 1B1 (Niemi et al., 2005b, 2011). The formation of the main metabolites of repaglinide, an oxidized dicarboxylic acid (M2) and, in particular, 3′-hydroxyl repaglinide (M4), is largely dependent on CYP2C8, whereas the less important aromatic amine metabolite (M1) is primarily formed by CYP3A4 (Bidstrup et al., 2003; Kajosaari et al., 2005a,b).
Pioglitazone, a thiazolidinedione peroxisome proliferator activated receptor (PPAR) γ agonist, is primarily metabolized by CYP2C8 in vitro, with smaller contributions by CYP3A4 and the extrahepatic CYP1A1 (Jaakkola et al., 2006c; FDA, 2013a). In vitro, CYP2C8 forms the pharmacologically active hydroxypioglitazone (M-IV) and ketopioglitazone (M-III) (Jaakkola et al., 2006c; Tornio et al., 2008b), which are the main metabolites in human serum with concentrations equal to or greater than those of the parent drug (Eckland and Danhof, 2000). In vivo studies support the central role of CYP2C8 in pioglitazone metabolism observed in vitro (section VI.C.2).
Rosiglitazone, another PPAR-γ agonist, is also a substrate of CYP2C8. In vitro, it undergoes CYP2C8-mediated p-hydroxylation and N-demethylation (Table 4), followed by sulfate and glucuronic acid conjugation (Baldwin et al., 1999; Kaspera et al., 2011; FDA, 2014a). CYP2C9 also participates in its metabolism to a minor extent (Baldwin et al., 1999). Rosiglitazone p-hydroxylation is recommended by the Food and Drug Administration (FDA) as a marker reaction for in vitro CYP2C8 activity (http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/ucm093664.htm). In vitro, troglitazone is metabolized by CYP2C8 to its quinone metabolite (M3) (Table 4) at a two- to eightfold higher rate than by CYP2C19, CYP3A4, and CYP2C9 (Yamazaki et al., 1999b). M3 is a minor metabolite of troglitazone, but it has been suggested to be responsible for the drug-induced hepatotoxicity associated with troglitazone use (Yamazaki et al., 1999b; Smith, 2003). Furthermore, the thiazolidinedione R483 is primarily metabolized by CYP2C8 and CYP2C19 in vitro (Bogman et al., 2010). CYP2C8 catalyzes the formation of the weakly active M1 metabolite (Table 4) and its further metabolism to M4, which is the main metabolite of R483 in plasma. In turn, CYP2C19 forms M2, which shows similar pharmacological activity as parent R483 (Bogman et al., 2010).
Additionally, CYP2C8 is involved, to a minor extent, in the metabolism of the sulfonylureas gliclazide, glyburide, and tolbutamide, the dipeptidyl peptidase 4 inhibitor sitagliptin, and the PPARα agonist sipoglitazar (Table 1; Relling et al., 1990; Srivastava et al., 1991; Veronese et al., 1993; Elliot et al., 2007; Vincent et al., 2007; Zharikova et al., 2009; Nishihara et al., 2012). Tolbutamide p-methyl hydroxylation has been used as a marker reaction for CYP2C8 activity in several in vitro studies. However, its Michaelis-Menten constant (Km) for CYP2C8 is very high (>530 µM; Table 4), and it is effectively metabolized by CYP2C9 at lower concentrations.
3. Antimalarial Agents.
The 4-aminoquinoline derivative amodiaquine, widely used for treatment of malaria for more than 60 years, is a substrate of CYP2C8 (Li et al., 2002). It is also metabolized by the extrahepatic enzymes CYP1A1 and CYP1B1 to a minor extent (Li et al., 2002). Amodiaquine N-deethylation is a frequently used in vitro marker reaction for CYP2C8 because of its high affinity (Km typically around 2 µM) and high turnover rate (Table 4). N-desethylamodiaquine, which is the main metabolite of amodiaquine, is assumed to be the main entity responsible for the pharmacological response to amodiaquine (Churchill et al., 1985; Mount et al., 1986).
Chloroquine, also a 4-aminoquinoline, is mainly metabolized to its N-desethyl metabolite by CYP2C8 (Table 4) and CYP3A4, with a small contribution by CYP2D6 in vitro (Kim et al., 2003; Projean et al., 2003a). Furthermore, CYP2C8 also seems to play a small role in the in vitro metabolism of the antimalarial agents dapsone, halofantrine, and piperaquine (Table 1; Baune et al., 1999; Winter et al., 2000; Lee et al., 2012c).
4. Lipid-lowering Drugs.
CYP2C8 participates to a small extent in the metabolism of several HMG-CoA reductase inhibitors (statins), but it has a major role for the biotransformation of cerivastatin. Cerivastatin is extensively metabolized in humans (Boberg et al., 1997; Mück, 2000). Parent cerivastatin (acid) is metabolized by CYP2C8 and CYP3A4, whereas cerivastatin lactone is predominantly metabolized by CYP3A4 (Boberg et al., 1997; Wang et al., 2002; Fujino et al., 2004). The formation of the major metabolite of cerivastatin, 6-hydroxycerivastatin (M-23), is primarily mediated by CYP2C8, whereas both CYP2C8 and CYP3A4 produce demethylcerivastatin (M1) (Wang et al., 2002; Kaspera et al., 2010). The notorious in vivo interaction between gemfibrozil and cerivastatin is discussed in section VI.C.4.
The parent simvastatin lactone is either oxidized by CYP3A4/5 or hydrolyzed to its acid form, which is pharmacologically active (Prueksaritanont et al., 1997, 2003). In human liver microsomes (HLM), the metabolism of simvastatin acid was catalyzed primarily (≥80%) by CYP3A4/5, with a smaller contribution (≤20%) by CYP2C8 (Prueksaritanont et al., 2003). Recombinant CYP2C8 formed all three simvastatin acid metabolites (M1-M3) observed in HLM (Table 4; Prueksaritanont et al., 2003). In vitro, fluvastatin is mainly metabolized by CYP2C9 into three metabolites, but CYP1A1, CYP2C8, CYP2D6, and CYP3A4 form 5-hydroxyfluvastatin (Fischer et al., 1999). Both atorvastatin (acid) and its lactone are primarily metabolized by CYP3A4 to their hydroxylated metabolites in vitro, but CYP2C8 is involved in the formation of p-hydroxy atorvastatin acid to a small extent (Jacobsen et al., 2000b). Furthermore, pitavastatin acid is metabolized by CYP2C9 and CYP2C8 in vitro, whereas its lactone is metabolized by CYP3A4 and CYP2D6 (Fujino et al., 2004).
5. Other Drugs.
Early in vitro studies concluded that the leukotriene receptor antagonist montelukast is mainly metabolized by CYP2C9 and CYP3A4 (Chiba et al., 1997), whereas the role of CYP2C8 was not evaluated. However, in vitro studies performed more than a decade later demonstrated that CYP2C8 is the key enzyme involved in the oxidative metabolism of montelukast (Filppula et al., 2011; VandenBrink et al., 2011). CYP2C8 catalyzes the main metabolic pathway of montelukast; formation of the pharmacologically active 36-hydroxymontelukast (M6), and its subsequent metabolism to the secondary metabolite M4, a dicarboxylic acid (Table 4). In addition, CYP2C8 forms 25-hydroxymontelukast (M3) (Filppula et al., 2011). These in vitro findings are in agreement with X-ray crystallography data, demonstrating a ligand-protein binding interaction between montelukast and CYP2C8 (Schoch et al., 2008). The montelukast molecule was positioned with its benzyl ring in close proximity to the heme iron of CYP2C8. The montelukast metabolites M3, M4, and M6 formed by CYP2C8 in vitro, all result from the oxidation of the benzyl ring of montelukast.
The novel prolyl hydroxylase inhibitor daprodustat (GSK1278863), an antianemic agent, is primarily metabolized by CYP2C8, with a smaller contribution by CYP3A4 in vitro (Johnson et al., 2014). It seems to be more sensitive than repaglinide to CYP2C8 inhibition by gemfibrozil in vivo (section VI.C.8; Johnson et al., 2014).
The novel nonstructural 5B nonnucleoside polymerase inhibitor dasabuvir is extensively metabolized by CYP2C8, with a small contribution by CYP3A (FDA, 2014g). CYP2C8 also plays an intermediate/small role in the metabolism of the nonstructural protein 3/4 A protease inhibitor paritaprevir and nonstructural protein 5A inhibitor ombitasvir (FDA, 2014g; Menon et al., 2015). However, no in vitro metabolism data have yet been published for these compounds.
The prostacyclin analog treprostinil is primarily metabolized by CYP2C8, followed by CYP2C9 in vitro (FDA, 2009b). Incubation of treprostinil with recombinant CYP2C8 for 15 minutes resulted in a 95% depletion of treprostinil concentrations, whereas only 22% was consumed by recombinant CYP2C9. CYP2C8 seems to be of importance in the in vivo pharmacokinetics of treprostinil (FDA, 2009b).
The sedative agent zopiclone is metabolized by CYP2C8 and CYP3A4 in vitro (Becquemont et al., 1999). In HLM, CYP2C8 was the main enzyme catalyzing N-demethylation of zopiclone, followed by CYP2C9 and CYP3A4. CYP2C8 also participated in the formation of N-oxide zopiclone, together with CYP3A4 (major) and CYP2C9 (Becquemont et al., 1999). However, in another in vitro study, montelukast (CYP2C8 inhibitor) and gemfibrozil (CYP2C8 and CYP2C9 inhibitor) had no effect on the elimination of clinically relevant concentrations of zopiclone (Tornio et al., 2006), supporting in vivo data showing a lack of effect of gemfibrozil on zopiclone concentrations in healthy subjects (section VI.C.8).
Based on in vitro studies, CYP2C8 likely plays an intermediate role in the elimination of 9cUAB30, alitretionin, amiodarone, cisapride, fenretinide, fluoxetine, irosustat, isotretionin, loperamide, olanzapine, olodaterol, propanoic acid, dronedarone, tazarotenic acid, verapamil, and vidupiprant (AMG 853) (Table 1). For instance, CYP2C8 catalyzes dealkylation of both enantiomers of the calcium channel blocker verapamil and its metabolite norverapamil (Busse et al., 1995; Tracy et al., 1999). Tazarotenic acid, the active moiety of the antipsoriatic agent tazarotene, is mainly metabolized by CYP2C8 and flavin-containing monooxygenases in vitro (Attar et al., 2003). When tazarotenic acid was incubated with 10 individual recombinant CYP enzymes, only CYP2C8 markedly catalyzed sulfoxidation, which is the main metabolic pathway of tazarotenic acid.
There is a vast amount of in vitro data suggesting that CYP2C8 may be of relevance in the metabolism of a number of other drugs (Table 1). For the majority of these compounds, the role of CYP2C8 in their in vivo elimination is likely to be small or negligible. However, for some drugs, the in vivo contribution of CYP2C8 to their metabolism cannot be estimated based on available information, and may, in fact, exceed 20%. Alternatively, CYP2C8 may become a determinant in their metabolism after inhibition of other enzymes important for their elimination. For instance, CYP2C8 is involved in the metabolism of seratrodast, a thromboxane A2 receptor antagonist in vitro (Kumar et al., 1997). The main metabolic pathway of seratrodast, 5-methylhydroxylation, is primarily catalyzed by CYP3A and CYP2C9, but CYP2C8 contributes to a small degree. However, CYP2C8 is a major contributor to seratrodast 4′-hydroxylation, a minor metabolic route (Kumar et al., 1997).
6. Glucuronide Metabolites.
Several glucuronide metabolites have been reported to undergo metabolism by CYP2C8, including clopidogrel acyl 1-β-d-glucuronide, desloratadine glucuronide, diclofenac acyl glucuronide, estradiol-17β-glucuronide, gemfibrozil 1-O-β glucuronide, licofelone 1-O-acyl glucuronide, Lu AA34893 carbamoyl glucuronide, 2-[[5,7-dipropyl-3-(trifluoromethyl)-1,2-benzisoxazol-6-yl]oxy]-2-methylpropanoic acid (MRL-C) acyl glucuronide, and sipoglitazar β-1-O-acyl glucuronide (see Table 2 for references). Thus, CYP2C8 makes yet another exception to the old concept that drug metabolism is divided into sequential phase I and phase II reactions, i.e., functionalization and conjugation, respectively (Fig. 5).
Kumar et al. (2002) demonstrated the first example of CYP2C8-mediated metabolism of a glucuronide conjugate when they showed that the conversion of diclofenac acyl glucuronide to its 4′-hydroxy derivative is exclusively mediated by CYP2C8 in vitro. In 2005, it was reported that CYP2C8 is also able to directly catalyze the 2-hydroxylation of estradiol-17β-glucuronide in vitro (Delaforge et al., 2005). Docking of the glucuronide of estradiol into the crystal structure of CYP2C8 showed that the active site is large enough to inhabit the conjugate. Also the fetal CYP3A isoform CYP3A7 oxidized estradiol-17β-glucuronide, but CYP2C8 was five times more active than CYP3A7. However, CYP3A4 was not able to metabolize estradiol-17β-glucuronide (Delaforge et al., 2005). Moreover, the acyl glucuronide of the dual PPAR α/β agonist MRL-C was oxidized by CYP2C8 and to a minor extent by CYP3A4, but not by CYP2C9 (Kochansky et al., 2005). Furthermore, the main elimination pathway of licofelone, a dual inhibitor of cyclooxygenases 1 and 2 and 5-lipoxygenase, is glucuronidation of its carboxylic acid metabolite, followed by CYP2C8-catalyzed hydroxylation of the acyl glucuronide M1 to form the hydroxylated glucuronide M3 (Albrecht et al., 2008).
For two compounds, the formation of an unconjugated hydroxyl metabolite involves oxidation and subsequent deconjugation of a glucuronide metabolite. In vitro, the antidiabetic agent sipoglitazar was first glucuronidated to sipoglitazar β-1-O-acyl glucuronide (sipoglitazar-G1). Sipoglitazar-G1 was then metabolized to the main metabolite M-I by O-dealkylation by CYP2C8 and subsequent deconjugation (Nishihara et al., 2012). A similar finding was recently observed for desloratadine (Kazmi et al., 2015). The main metabolite of desloratadine, 3-hydroxydesloratadine, which is active, was formed via CYP2C8-mediated oxidation of desloratadine glucuronide and a deconjugation event (Kazmi et al., 2015). Thus, it seems that phase II metabolism occurs before phase I for these compounds (Fig. 5).
Also the glucuronide metabolites of clopidogrel, gemfibrozil, and Lu AA34893 are likely to be substrates of CYP2C8 (Ogilvie et al., 2006; Baer et al., 2009; Kazmi et al., 2010; Tornio et al., 2014). All three compounds are metabolism-dependent inhibitors of CYP2C8, as discussed in sections V.B and VI.B.
B. Endogenous and Natural Compounds
CYP2C8 metabolizes some endogenous and natural compounds (Table 3). CYP2C8 participates in the metabolism of arachidonic acid to biologically active epoxyeicosatrienoic acids (e.g., 11-, 13-, or 15-hydroxyeicosatrienoic acid), involved in the regulation of numerous physiologic processes, e.g., vascular function, blood pressure regulation, pancreatic peptide hormone secretion, and platelet aggregation (Daikh et al., 1994; Rifkind et al., 1995; Zeldin et al., 1995). The roles of CYP2C8 and other CYP enzymes in inflammation, cardiovascular disease, and cancer were recently reviewed by Chen and Wang (2015) and Fleming (2014).
CYP2C8 and CYP3A enzymes have generally been considered to be the primary CYPs involved in the metabolism of all-trans-retinoic acid, the active form of vitamin A (retinol) (Leo et al., 1989; Nadin and Murray, 1999; Marill et al., 2000). All-trans-retinoic acid is involved in gene transcription, cell division, and differentiation (Tzimas and Nau, 2001; Marill et al., 2003; Duester, 2008). According to more recent data, however, CYP26A1 and CYP3A4 are the primary determinants of all-trans-retinoic acid metabolism in humans, whereas the role of CYP2C8 is of smaller importance (Thatcher et al., 2010). In vitro, CYP2C8 also catalyzes the metabolism of other retinoids, including 9-cis-retinoic acid, 13-cis-retinoic acid, and 9cUAB30 (Marill et al., 2002; Gorman et al., 2007).
Some in vitro data suggest that CYP2C8 may contribute to the metabolism of the steroids 17β-estradiol, progesterone, and testosterone (Waxman et al., 1991; Spink et al., 1992, 1994). However, there seems to be no evidence for a role of CYP2C8 in the metabolism of androgens in vivo. Although CYP2C8 participates in the metabolism of several endogenous compounds, no cardiovascular or other potentially CYP2C8-related adverse effects were observed in the Helsinki Heart Study, in which over 2000 middle-aged men with primary dyslipidemia ingested the strong CYP2C8 inhibitor gemfibrozil 600 mg twice daily for several years (Frick et al., 1987).
Furthermore, some natural compounds have been reported to undergo metabolism by CYP2C8 in vitro (Table 3). CYP2C8 and CYP3A4 are the primary enzymes involved in the in vitro metabolism of 1-hydroxyl-2,3,5-trimethoxyxanthone, a constituent of the Tibetan medicinal plant Halenia elliptica (Feng et al., 2014). CYP2C8 is responsible for the main metabolic pathway of silybin, the active component of silymarin in vitro (Jancova et al., 2007). The CYP2C8 inhibitor quercetin inhibited silybin O-demethylation by 80% in HLM, and recombinant CYP2C8 was the major enzyme forming O-demethyl silybin, with a small contribution by CYP3A4. Furthermore, CYP2C8 is the major enzyme responsible for the in vitro metabolism of tanshinol borneol ester, a combination of the natural compounds danshensu and borneol (Liu et al., 2010a). Recombinant CYP2C8 generated all five tanshinol borneol metabolites (M1-M5) observed in HLM incubations, whereas recombinant CYP3A4 only produced the M4 metabolite.
Nearly 100 nonsynonymous single nucleotide variations (SNV) and short deletions, as well as essential splice site variants have been found in the CYP2C8 gene. The variants described in the literature, dbSNP database, or the 1000 Genomes project database are listed in Table 5, together with their continental frequencies and predicted or experimentally determined effects on protein function. The vast majority of the nonsynonymous variants are rare and occur at minor allele frequencies of 0.01 or less in all investigated populations.
A. Population Genetics
Three alleles, known as CYP2C8*2, *3, and *4, account for the majority of nonsynonymous variability of CYP2C8 in humans. Their frequencies differ significantly both between and within continental populations (Table 5, Fig. 6).
The CYP2C8*2 allele (c.805A>T, p.Cys266Phe) occurs mostly in individuals with a sub-Saharan African ancestry. In sub-Saharan African populations, its allele frequency ranges from about 0.10 in a Fulani population in Burkina Faso to 0.37 in a Mbuti pygmy population in Congo (Cavaco et al., 2005; Rower et al., 2005; Parikh et al., 2007; Kudzi et al., 2009; Speed et al., 2009; Paganotti et al., 2011, 2012; 1000 Genomes Project Consortium, 2012; Arnaldo et al., 2013; Marwa et al., 2014). In an African-American population in the New York area, the allele frequency of CYP2C8*2 was 0.10 (Martis et al., 2013). The allele is also relatively common in the mixed Brazilian population with a frequency of 0.06, New York area Hispanic population with a frequency of 0.02, and North and South Indian populations, with frequencies of 0.03 and 0.01, respectively (Suarez-Kurtz et al., 2012; Martis et al., 2013; Minhas et al., 2013; Arun Kumar et al., 2015). The CYP2C8*2 allele is very rare or absent in East Asian and European populations, with the exception of an allele frequency of 0.01 in a Portuguese European sample (Nakajima et al., 2003; Muthiah et al., 2005; Cavaco et al., 2006; Pechandova et al., 2012; Vargens et al., 2012; Martis et al., 2013; Wu et al., 2013).
The CYP2C8*3 allele is a haplotype consisting of two nonsynonymous variants (c.416G>A, p.Arg139Lys and c.1196A>G, p.Lys399Arg), which appear to be in a complete or nearly complete linkage disequilibrium in all investigated populations (1000 Genomes Project Consortium, 2012). The linkage disequilibrium extends also beyond the CYP2C8 gene, as evidenced by a strong correlation between the CYP2C8*3 and CYP2C9*2 (rs1799853; c.430C>T, p.Arg144Cys) alleles in the Swedish population (Yasar et al., 2002). The highest allele frequencies of CYP2C8*3 are seen in individuals with a European ancestry. In European populations, the allele frequency of CYP2C8*3 ranges from 0.069 in Faroe Islanders to 0.198 in a Portuguese population, with an apparent north-to-south cline from lower to higher frequencies (Fig. 6; Yasar et al., 2002; Halling et al., 2005; Cavaco et al., 2006; Speed et al., 2009; 1000 Genomes Project Consortium, 2012; Pechandova et al., 2012; Suarez-Kurtz et al., 2012). The allele is also quite common in European American and North American Hispanic populations, with frequencies of 0.09 and 0.08, respectively (Martis et al., 2013). In the mixed Brazilian and Ecuadorian populations, its frequency is 0.08 and 0.07, and in a Chilean mestizo population it is 0.06 (Roco et al., 2012; Suarez-Kurtz et al., 2012; Vicente et al., 2014). There is wide variability in the frequency of CYP2C8*3 in sub-Saharan African populations, even within a country (Cavaco et al., 2005; Rower et al., 2005; Parikh et al., 2007; Kudzi et al., 2009; Arnaldo et al., 2013; Staehli Hodel et al., 2013; Marwa et al., 2014; Paganotti et al., 2014). For example, the frequency of CYP2C8*3 was found to be 0.00 in individuals in central Tanzania and as high as 0.10 in the Mwanza region of Tanzania (Staehli Hodel et al., 2013; Marwa et al., 2014).
The CYP2C8*4 (c.792C>G, p.Ile264Met) allele has its highest frequencies in European populations, with the allele frequency ranging from 0.04 in a Spanish population to 0.07 in the Irish (Cavaco et al., 2006; Speed et al., 2009; 1000 Genomes Project Consortium, 2012; Pechandova et al., 2012). Its frequency was 0.03 in a European American population and a mixed Brazilian population (Suarez-Kurtz et al., 2012; Martis et al., 2013). In Peruvian, Colombian, Puerto Rican, and North American Hispanic populations, the frequency ranges from 0.01 to 0.02 (1000 Genomes Project Consortium, 2012; Martis et al., 2013). The CYP2C8*4 allele is found with a frequency of about 0.03–0.04 in Indian individuals and 0.01 in the Pakistani (1000 Genomes Project Consortium, 2012; Minhas et al., 2013). In East Asian populations, the frequency of CYP2C8*4 is generally 0.01 or less, but a frequency of 0.02 was seen in an Uighur Chinese population (Nakajima et al., 2003; Muthiah et al., 2005; Speed et al., 2009; 1000 Genomes Project Consortium, 2012; Staehli Hodel et al., 2013; Wu et al., 2013). The allele is rare in individuals with a sub-Saharan African ancestry, with a frequency of below 0.01 in all investigated sub-Saharan African populations and 0.01 in an African American population (Cavaco et al., 2005; Rower et al., 2005; Kudzi et al., 2009; 1000 Genomes Project Consortium, 2012; Arnaldo et al., 2013; Martis et al., 2013).
In addition to the common variants, rare nonsynonymous CYP2C8 variants exist in all continental populations (Table 5). A number of the rare CYP2C8 variants can be predicted to result in a loss-of-function because of premature termination of protein synthesis. The c.635G>A (p.Trp212Ter) and c.721C>T (p.Arg241Ter) variants have a combined allele frequency of 0.003 in sub-Saharan African populations, and the c.820G>T (p.Glu274Ter) and c.1198G>T (p.Glu400Ter) variants have a combined allele frequency of 0.004 in East Asians (1000 Genomes Project Consortium, 2012). Other predicted loss-of-function CYP2C8 variants were not found in the 1000 Genomes Project populations, and data are too scarce to estimate their population frequencies.
B. Functional Studies
The functional effects of CYP2C8 variants have been investigated using recombinantly expressed variant proteins and HLM with different CYP2C8 genotypes. Recombinant CYP2C8.2 has been quite consistently associated with an about 50% decrease in the intrinsic clearance for paclitaxel 6α-hydroxylation, compared with CYP2C8.1 (Dai et al., 2001; Gao et al., 2010; Yu et al., 2013b). In addition, the intrinsic clearance of amodiaquine has been reduced by 80–90% in CYP2C8.2 and that of repaglinide by 20% compared with CYP2C8.1 (Parikh et al., 2007; Yu et al., 2013b). Similarly, the intrinsic clearances of arachidonic acid and tanshinol borneol ester appeared to be lower by CYP2C8.2 than by CYP2C8.1, but the differences were not statistically significant (Dai et al., 2001; Liu et al., 2010a). On the other hand, the intrinsic clearances for cerivastatin M-23 and M-1 metabolite formation and R- and S-ibuprofen hydroxylations have been nonsignificantly higher in CYP2C8.2 than in CYP2C8.1 (Kaspera et al., 2010; Yu et al., 2013b). Both the SIFT or Polyphen in silico prediction algorithms suggest that the amino acid change in CYP2C8.2 is deleterious for CYP2C8 activity (Table 5).
In several studies, the intrinsic clearance for paclitaxel 6α-hydroxylation by recombinant CYP2C8.3 has been between 30 and 85% lower than by CYP2C8.1 (Dai et al., 2001; Soyama et al., 2001; Gao et al., 2010; Yu et al., 2013b). Other studies employing recombinant CYP2C8.3 have shown increased intrinsic clearance for repaglinide and cerivastatin but reduced intrinsic clearance for R- and S-ibuprofen and nearly abolished intrinsic clearance for amodiaquine (Kaspera et al., 2010; Parikh et al., 2007; Yu et al., 2013b). The intrinsic clearances of arachidonic acid to 11,12- and 14,15-epoxyeicosatrienoic acid and tanshinol borneol ester have also been significantly lower by CYP2C8.3 than by CYP2C8.1 (Dai et al., 2001; Liu et al., 2010a). However, in a recent study expressing CYP2C8.3 together with cytochrome P450 reductase and cytochrome b5, the intrinsic clearance for paclitaxel 6α-hydroxylation was about twofold higher, that for amodiaquine about twofold higher, that for rosiglitazone about 2.5-fold higher, and that for cerivastatin about 4.5-fold higher compared with CYP2C8.1 (Kaspera et al., 2011). A stronger binding affinity of ligands to CYP2C8.3 together with an increase in heme spin change during binding of ligands and redox partners were suggested to partly explain the increased catalytic activity (Kaspera et al., 2011). In one study, HLM heterozygous for the CYP2C8*3 allele showed lowered paclitaxel 6α-hydroxylase activity and in another study no change in amodiaquine N-deethylation compared with microsomes homozygous for CYP2C8*1 (Bahadur et al., 2002; Kaspera et al., 2011). Studies employing HLM heterozygous or homozygous for CYP2C8*3 have shown increased intrinsic clearance of pioglitazone and imatinib (Muschler et al., 2009; Khan et al., 2015). In silico predictions suggest that neither of the amino acid changes in CYP2C8.3 affect CYP2C8 activity (Table 5). Taken together, in vitro evidence concerning the functional effects of CYP2C8*3 suggests some degree of a substrate-specific effect but is discrepant for some substrates in that both decreased and increased activities have been reported.
In three studies, paclitaxel 6α-hydroxylation intrinsic clearance was reduced by about 70% by recombinant CYP2C8.4 compared with CYP2C8.1 (Singh et al., 2008; Gao et al., 2010; Yu et al., 2013b). Similarly, the intrinsic clearances of repaglinide and R- and S-ibuprofen have been 20, 50, and 53% lower by CYP2C8.4 than by CYP2C8.1, respectively (Yu et al., 2013b). On the other hand, the intrinsic clearances of cerivastatin to M-23 and M-1 were about 2- to 2.5-fold higher by CYP2C8.4 than by CYP2C8.1 (Kaspera et al., 2010). The intrinsic clearance of tanshinol borneol ester was not significantly different between CYP2C8.4 and CYP2C8.1 (Liu et al., 2010a). One study suggests that the amino acid change in CYP2C8.4 disrupts heme binding and results in an inactive protein (Singh et al., 2008). HLM heterozygous for CYP2C8*4 showed a nonsignificant tendency for lower paclitaxel 6α-hydroxylase activity (Bahadur et al., 2002). In silico predictions suggest that the amino acid change in CYP2C8.4 is deleterious for CYP2C8 activity (Table 5).
In vitro studies employing recombinant CYP2C8 have shown reduced paclitaxel 6α-hydroxylase activity in association with the p.Arg186Gln, p.Ala238Pro (*14), and p.Pro404Ala variants but no change in activity due to the p.Gly171Ser, p.Ile223Met (*13), p.Lys247Arg, and p.Lys383Asn variants (Soyama et al., 2001; Hichiya et al., 2005; Hanioka et al., 2010). In one study, the p.Ala238Pro and p.Ile223Met variants were associated with reduced amiodarone metabolism (Hanioka et al., 2011). One study demonstrated lack of CYP2C8 protein expression in association with the p.Glu274Ter (*11) nonsense variant (Yeo et al., 2011).
C. Effects on Drug Metabolism in Humans
In contrast to previous in vitro studies suggesting a reduced CYP2C8 activity in association with the CYP2C8*3 allele (Dai et al., 2001; Bahadur et al., 2002), the first pharmacokinetic study in humans showed that the CYP2C8*3 allele was associated with reduced plasma concentrations of repaglinide (Niemi et al., 2003c). In this and later studies, individuals with the CYP2C8*1/*3 genotype have had an approximately 40–50% lower AUC of a subtherapeutic dose of repaglinide than individuals with the CYP2C8*1/*1 genotype (Niemi et al., 2005b,c). However, this finding has not been fully replicated in studies with higher repaglinide doses (Bidstrup et al., 2006; Tomalik-Scharte et al., 2011), suggesting that the effect of CYP2C8*3 allele on repaglinide pharmacokinetics may be dose dependent.
Similarly to repaglinide, the CYP2C8*3 allele has been associated with apparently increased clearance of the thiazolidinediones rosiglitazone and pioglitazone (Kirchheiner et al., 2006; Aquilante et al., 2008, 2013a; Tornio et al., 2008b). The AUCs of rosiglitazone or pioglitazone have been about 20–40% lower in CYP2C8*3 carriers than in noncarriers, with an apparent gene-dose effect (Kirchheiner et al., 2006; Aquilante et al., 2008, 2013a; Tornio et al., 2008b). Furthermore, in a study in patients with type 2 diabetes mellitus, the CYP2C8*3 allele has been associated with significantly lower trough rosiglitazone concentrations and an impaired lowering of glycosylated hemoglobin (HbA1c) during rosiglitazone treatment (Stage et al., 2013). In one study in African American subjects, the CYP2C8*2 allele had no impact on parent pioglitazone pharmacokinetics but was associated with impaired metabolism of pioglitazone to its M3 metabolite (Aquilante et al., 2013c).
Although CYP2C8*2 has been associated with significantly impaired amodiaquine metabolism in vitro (Parikh et al., 2007), the allele has not been clearly associated with amodiaquine efficacy or toxicity (Adjei et al., 2008). However, more recent evidence suggests that CYP2C8 genetic variability can influence the occurrence of amodiaquine or chloroquine resistance in malaria parasites (Paganotti et al., 2011; Cavaco et al., 2013).
Studies in cancer patients have suggested that the CYP2C8*3 allele can slightly impair the clearance of paclitaxel (Henningsson et al., 2005; Bergmann et al., 2011). Some studies have also suggested that the CYP2C8*3 allele or other CYP2C8 variants may be risk factors for paclitaxel-induced neurotoxicity or myelosuppression and affect the benefit-to-risk ratio of paclitaxel therapy (Green et al., 2011; Leskelä et al., 2011; Hertz et al., 2012; Hertz et al., 2014; Lee et al., 2015). Further studies are required to clarify the role of CYP2C8 genetic variants in affecting paclitaxel response.
Some studies have reported significantly increased plasma concentrations and apparently reduced clearance of racemic ibuprofen and its enantiomers in association with the CYP2C8*3 allele (Garcia-Martin et al., 2004; Martinez et al., 2005; Karaźniewicz-Łada et al., 2009). On the other hand, one study reported enhanced clearance of R-ibuprofen in association with CYP2C8*3 (Lopez-Rodriguez et al., 2008). Because ibuprofen is a substrate of CYP2C9, it is likely that the discrepancies are due to the strong linkage disequilibrium between CYP2C8*3 and CYP2C9*2 and reduced ibuprofen clearance in CYP2C8*3 carriers is in fact due to the CYP2C9*2 allele.
Although CYP2C8 is not known to be involved in bisphosphonate pharmacokinetics, an intronic SNV in CYP2C8 (rs1934951) has been associated with zoledronic acid-induced osteonecrosis of the jaw in patients treated for multiple myeloma (Sarasquete et al., 2008). In a more recent study, this SNV was associated with the mandibular localization of bisphosphonate-induced osteonecrosis (Balla et al., 2012). However, there was no significant relationship between the variant and the development of bisphosphonate-induced osteonecrosis of the jaw in men with prostate cancer (English et al., 2010) or in patients with multiple myeloma (Such et al., 2011). A meta-analysis found no significantly increased susceptibility to bisphosphonate-induced osteonecrosis of the jaw in rs1934951 carriers when all cancer types were pooled, but suggested a significant association in multiple myeloma patients (Zhong et al., 2013).
V. In Vitro Inhibition and Induction of Cytochrome P450 2C8
A. Reversible Inhibition
1. Drugs That Act as Inhibitors of Cytochrome P450 2C8.
Several drugs, drug metabolites, and other compounds have been found to inhibit CYP2C8 activity reversibly in vitro (Tables 6 and 7). In an in vitro screening of 209 commonly used drugs, 48 compounds exhibited greater than 50% inhibition of recombinant CYP2C8 activity at an inhibitor concentration of 30 µM (Walsky et al., 2005a). Montelukast, candesartan cilexetil, zafirlukast, clotrimazole, felodipine, and mometasone furoate inhibited CYP2C8 with concentrations supporting half of the maximal inhibition (IC50) of ≤3 µM in recombinant CYP2C8 and HLM. In another study, the inhibition of CYP2C8 by montelukast was found to be competitive and selective, with reversible inhibition constants (Ki) ranging from 0.0092 to 0.15 µM, depending on the protein concentration used in the incubation (Walsky et al., 2005b). However, despite their strong inhibitory effect on CYP2C8 in vitro, neither montelukast nor zafirlukast affected the pharmacokinetics of CYP2C8 substrate drugs in vivo (Jaakkola et al., 2006b; Kajosaari et al., 2006b; Kim et al., 2007). The lack of in vivo effect is likely explained by their extensive plasma protein binding (>99%) (FDA, 1998; Dekhuijzen and Koopmans, 2002). Also the inhibition of CYP2C8 by candesartran cilexetil (prodrug of candesartan), clotrimazole, and mometasone furoate are probably not clinically relevant. The antifungal clotrimazole and anti-inflammatory mometasone furoate are topically applied and are therefore unlikely to cause interactions because of low systemic concentrations (Walsky et al., 2005a). In the systemic circulation, candesartan cilexetil is cleaved to candesartan, and, consequently, the likelihood of a drug interaction elicited by its prodrug is low. Furthermore, predictions suggested a relatively weak potential for drug-drug interactions due to CYP2C8 inhibition by felodipine. No drug interaction studies between felodipine and CYP2C8 substrates have been reported.
Trimethoprim, an antimicrobial agent, is a competitive inhibitor of CYP2C8 in vitro (Wen et al., 2002), with a Ki value typically around 10–30 µM in HLM (Table 6). The inhibition of CYP2C8 by trimethoprim seems to be rather selective, because it does not inhibit CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP2E1, or CYP3A4 at concentrations below 100 µM. In healthy subjects, trimethoprim has moderately increased the plasma exposure to several CYP2C8 substrate drugs (section VI.).
The flavonoid quercetin is one of the earliest in vitro inhibitors of CYP2C8 detected. In studies of paclitaxel metabolism, it was observed that quercetin, unlike CYP3A4 inhibitors, inhibited paclitaxel 6α-hydroxylation (Harris et al., 1994; Kumar et al., 1994). Because it was shown that the 6α-hydroxylation of paclitaxel is mediated by CYP2C8, it was evident that quercetin is an inhibitor of this enzyme (Rahman et al., 1994). Quercetin inhibits CYP2C8 competitively with a Ki of 0.03–20 µM (Table 7) and is classified as a "preferred" probe in vitro inhibitor of CYP2C8 by the FDA (http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/ucm093664.htm). However, quercetin is not selective for CYP2C8; it also inhibits CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 with IC50 values of 3.1–47 µM (Obach, 2000; Zou et al., 2002). In vivo, quercetin at steady state did not affect the pharmacokinetics of rosiglitazone (Kim et al., 2005a).
The lipid-lowering drug gemfibrozil is a moderate, direct competitive inhibitor of CYP2C8 in vitro (Wen et al., 2001; Prueksaritanont et al., 2002; Wang et al., 2002), with a Ki range between 9.3 and 270 µM (Table 6). Gemfibrozil also inhibits CYP2C9 and CYP2C19 with Ki values of 5.8 and 24 µM, respectively, and CYP1A2 with a Ki of 82 µM (Wen et al., 2001). Moreover, it inhibits several drug transporters in vitro, most notably OATP1B1 (lowest reported Ki = 4 µM) and organic anion transporter (OAT) 3 (Ki = 6.8 µM) (Schneck et al., 2004; Shitara et al., 2004; Nakagomi-Hagihara et al., 2007). The strong interactions between gemfibrozil and CYP2C8 substrate drugs observed in vivo are mainly due to its glucuronide metabolite (Ogilvie et al., 2006). Gemfibrozil 1-O-β glucuronide affects CYP2C8 by mechanism-based inhibition (sections V.B and VI.B).
In vitro, the thiazolidinedione drugs pioglitazone, rosiglitazone, and troglitazone are potent, competitive inhibitors of CYP2C8 with IC50 and Ki values of <40 µM (Table 6). However, e.g., pioglitazone does not affect CYP2C8 in vivo, likely because of its extensive protein binding (Kajosaari et al., 2006a).
The antiviral agents atazanavir and efavirenz inhibit CYP2C8 in vitro with Ki values of 2.1 and 4.8 µM, respectively [inhibitor concentration (I) to Ki ratios (I/Ki) = 3.7 and 6.3, respectively] (Table 6). In vivo, atazanavir has slightly affected the pharmacokinetics of rosiglitazone (FDA, 2015b). According to predictions, efavirenz may increase the area under the plasma concentration-time curve (AUC) of CYP2C8 substrates by more than fourfold at steady state, and such effects have been observed in vivo (German et al., 2007).
The immunosuppressant teriflunomide inhibits CYP2C8 with a very low Ki of 0.10–0.15 µM (FDA, 2012a). Thus, its estimated I/Ki ratio of 1.1 indicates that interactions between teriflunomide and CYP2C8 substrate drugs are likely, in agreement with in vivo findings (section IV.C.2).
Numerous protein kinase inhibitors inhibit CYP2C8 to various degrees in vitro (Table 6). However, for the most part, their in vivo inhibitory effects on CYP2C8 have not been studied. For those whose inhibition has been examined in a clinical setting, it seems to be rather small/moderate. For instance, axitinib inhibits CYP2C8 in vitro with a Ki of 0.2–0.5 µM (I/Ki = 0.3–0.9), but it did not alter paclitaxel plasma concentrations in patients (FDA, 2012f; Wang et al., 2014a). Similarly, cabozantinib is a noncompetitive inhibitor of CYP2C8 in vitro (Ki = 4.6 µM, I/Ki = 0.7), but the in vivo pharmacokinetics of rosiglitazone was not affected by cabozantinib (FDA, 2012c; Nguyen et al., 2015). The inhibition of CYP2C8 by pazopanib (Ki of 3.7 µM, I/Ki = 35) may be of clinical relevance (FDA, 2009d; Tan et al., 2014; Wang et al., 2014b). Nilotinib is a strong competitive CYP2C8 inhibitor in vitro (Ki = 0.1–0.9 µM, I/Ki = 4.8–43), but it also induces CYP2C8 (FDA, 2007c). Hence, a clinical interaction study with a CYP2C8 probe substrate has been recommended by the FDA to evaluate the in vivo effect on CYP2C8 activity by nilotinib. Similarly, an interaction study with a CYP2C8 substrate drug has also been recommended for regorafenib, which inhibits CYP2C8 with a Ki value of 0.6 µM in vitro (I/Ki = 13.5) (FDA, 2012i). In addition, sorafenib seems to be a strong CYP2C8 inhibitor in vitro with Ki values < 3 µM (I/Ki = 9–22) (Table 6), but the effect of sorafenib on CYP2C8 in vivo has not been evaluated.
Also several other anticancer agents exhibit inhibition of CYP2C8 in vitro (Table 6). For instance, the androgen receptor antagonist enzalutamide is both a substrate and inhibitor of CYP2C8 in vitro (Ki = 5.5 µM, I/Ki = 6.5) (FDA, 2012k). Vismodegib, an oral hedgehog pathway inhibitor, inhibits CYP2C8 in vitro with a Ki of 6.0 µM (I/Ki = 2.7), but vismodegib at steady state did not affect the pharmacokinetics of rosiglitazone (Wong et al., 2009; LoRusso et al., 2013).
The iron chelator deferasirox inhibits CYP2C8 with an IC50 of 100 µM (I/Ki <0.01) (FDA, 2005a), but it has increased repaglinide AUC by 2.3-fold in vivo (Skerjanec et al., 2010). Febuxostat, a xanthine oxidase inhibitor, inhibits CYP2C8 in vitro with a Ki of 20 µM, suggesting that the inhibition may be of clinical relevance (I/Ki = 0.8). However, febuxostat at steady state had no effect on the concentrations of a single dose of rosiglitazone in vivo (Naik et al., 2012). Similarly, rosiglitazone pharmacokinetics was not affected by the platelet aggregation inhibitor vorapaxar in vivo (Ki = 0.86 µM, I/Ki = 0.06) (Chen et al., 2014; FDA, 2014l).
Sulfaphenazole, ketoconazole, diethyldithiocarbamate, methoxsalen (8-methoxypsoralen), and tranylcypromine, commonly used as in vitro inhibitors of CYP2C9, CYP3A4, CYP2E1, CYP2A6, and CYP2C19, respectively, also inhibit CYP2C8 in vitro (Table 6). For instance, sulfaphenazole is a strong competitive inhibitor of CYP2C9 with a Ki of 0.3 µM, whereas its Ki for CYP2C8 inhibition is 0.4–63 µM (Mancy et al., 1996; Hamman et al., 1997).
2. Natural Compounds.
A range of natural compounds have been tested for CYP2C8 inhibition in vitro, and inhibition parameters have been determined for several of them (Table 7). In a CYP inhibition screening of 10 herbal products commercially available in Australia, horsetail (Equisetum arvense) affected CYP2C8 with an IC50 of 93.0 µg/ml (Sevior et al., 2010). The authors suggested that the inhibition of CYP2C8 by horsetail, which is used for treatment of urinary tract infections, cystitis, and prostate problems, may be clinically relevant (Sevior et al., 2010). In another in vitro study, six herbal supplements inhibited CYP2C8 to various degree, but the inhibition by cranberry powder (IC50 = 24.7 µg/ml) and saw palmetto (IC50 = 15.4 µg/ml) were suggested to potentially be of clinical significance (Albassam et al., 2015).
Among five CYP enzymes tested, CYP2C8 was most sensitive to inhibition by green tea extract in HLM (IC50 = 4.5 µg/ml) (Misaka et al., 2013). The major catechin in green tea, (−)-epigallocatechin-3-gallate, inhibited CYP2C8 with a Ki of 6.8 µM, indicating that green tea intake may affect CYP2C8 in vivo. It has been reported that 15% of Japanese older than 40 years of age consume more than 1.8 l of green tea daily, corresponding to a daily epigallocatechin-3-gallate intake of 540–720 mg (Misaka et al., 2013).
B. Metabolism-dependent Inhibition
Metabolism-dependent inhibitors are compounds that are metabolized to metabolites or reactive intermediates that cause time-dependent enzyme inhibition. Metabolism-dependent inhibition may be either direct, quasi-irreversible, or irreversible (mechanism-based inhibition). Mechanism-based inhibitors inactivate their victim enzymes permanently, and enzyme activity can only be regained by de novo synthesis of the enzyme (Lin and Lu, 1998). Interestingly, two glucuronide metabolites, gemfibrozil 1-O-β glucuronide and clopidogrel acyl 1-β-d-glucuronide, affect CYP2C8 by mechanism-based inhibition or quasi-irreversible inhibition, leading to clinically important drug-drug interactions (Figs. 2 and 5; Table 8; Ogilvie et al., 2006; Tornio et al., 2014). Very recently, also the acyl glucuronide of deleobuvir, an HCV protease inhibitor, was found to be a very potent mechanism-based inhibitor of CYP2C8 (Sane et al. 2015). In addition, there is in vitro evidence suggesting that the carbamoyl glucuronide metabolite of Lu AA34893 may affect CYP2C8 in a similar manner (Kazmi et al., 2010). Of interest, parent clopidogrel and gemfibrozil do not seem to be metabolized by CYP2C8. For example, clopidogrel is mainly eliminated by carboxylesterase 1, whereas its activation is dependent on CYP2C19 and CYP3A4 (Mega et al., 2009; Simon et al., 2009; Holmberg et al., 2014; Tarkiainen et al., 2015).
The inhibitory effect of gemfibrozil on CYP2C8 is based principally on its metabolite, gemfibrozil 1-O-β glucuronide, which is formed mainly by UGT2B7 in hepatocytes (Shitara et al., 2004; Ogilvie et al., 2006; Mano et al., 2007). The metabolite acts as a mechanism-based inhibitor of CYP2C8, with inhibitor concentration supporting half of the maximal rate of enzyme inactivation (KI) and maximal rate of inactivation (kinact) values of 20–52 µM and 0.21 1/min in vitro (Ogilvie et al., 2006; Baer et al., 2009). Similarly, clopidogrel acyl 1-β-d-glucuronide causes a metabolism-dependent inhibition of CYP2C8 with KI and kinact values of 9.9 µM and 0.047 1/min (Tornio et al., 2014). The in vivo consequences of the inhibitory effects of these metabolites are discussed in section VI. In an in vitro study by Jenkins et al. (2011), the acyl glucuronides of atorvastatin, dehydroketoprofen, diclofenac, ibuprofen, indomethacin, rac-ketoprofen, mefenamic acid, R- and S-naproxen, and simvastatin did not affect CYP2C8 by metabolism-dependent inhibition.
Several other metabolism-dependent inhibitors of CYP2C8 have been reported in the literature (Table 8). However, the clinical importance of their interaction potential is unknown.
Increased expression of CYP2C8 protein in hepatocytes due to enzyme-inducing drugs/xenobiotics is an important mechanism of drug-drug interactions that can lead to markedly increased clearance of CYP2C8 substrates, resulting in reduced efficacy and therapeutic failure. Several drug-responsive nuclear receptors, including CAR, PXR, VDR, and GR, can mediate the transcriptional activation of the CYP2C8 gene by recognizing the respective responsive elements within the 5′-flanking promoter region of the gene (Chen and Goldstein, 2009). After activation of nuclear receptors by their ligands/activators (in particular, enzyme inducing drugs), the nuclear receptors enter the nucleus, bind to their responsive elements in the DNA, recruit coactivators that affect chromatin structure, and increase the transcription of the target genes (Handschin and Meyer, 2003). Apart from this general mechanism, certain compounds, such as phenobarbital and CITCO ([6-(4-chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde-O-(3,4-dichlorobenzyl)oxime), seem to cause induction by increasing the nuclear translocation of CAR, which is constitutively active (Zelko et al., 2001). Other nuclear receptors and transcriptional factors, such as HNF4α, HNF3γ, C/EBPα, and RORs, can regulate the constitutive expression of CYP2C genes, but these factors are probably not directly involved in induction of CYP2C8 (Ferguson et al., 2005; Chen and Goldstein, 2009; Rana et al., 2010). Yet, at least HNF4α seems to be required for upregulation by the PXR agonist rifampin (Rana et al., 2010).
In experimental in vitro studies, several compounds and ligands of the different nuclear receptors have been able to induce CYP2C8 (Table 9). On the basis of studies in cultured human hepatocytes, CYP2C8 is the most inducible member of the CYP2C subfamily (Gerbal-Chaloin et al., 2001; Feidt et al., 2010). Regarding inducibility of CYP2C8, the PXR-receptor seems to be the most important nuclear receptor, because typical PXR ligands/activators strongly induce CYP2C8 in vitro (Ferguson et al., 2005; Chen and Goldstein, 2009) and can cause induction of CYP2C8 also in vivo, whereas ligands of the other nuclear receptors cause only moderate induction of CYP2C8 in vitro (Ferguson et al., 2005; Chen and Goldstein, 2009) and have not been shown to markedly induce CYP enzymes in vivo in humans. PXR activators, such as phenobarbital, hyperforin (an ingredient of St. John's wort), and rifampin, have increased CYP2C8 expression at mRNA, protein, and activity levels several-fold in vitro (Dussault et al., 2001; Gerbal-Chaloin et al., 2001; Rae et al., 2001; Nishimura et al., 2002; Raucy et al., 2002; Madan et al., 2003; Ferguson et al., 2005; Komoroski et al., 2005; Thomas et al., 2015). In addition, certain other compounds, including ritonavir, nelfinavir, cyclophosphamide, lithocholic acid, and paclitaxel can induce CYP2C8 presumably by a PXR-mediated mechanism in vitro (Chang et al., 1997; Dussault et al., 2001; Synold et al., 2001; Ferguson et al., 2005; Dixit et al., 2007). It should be noted that in one study, rifampin induced CYP2C8 mRNA in only three of the eight commercially available cryopreserved hepatocyte lots tested (Yajima et al., 2014), suggesting that cryopreserved hepatocytes may not be a reliable system for studying CYP2C8 induction.
Apart from PXR, induction of CYP2C8 can be experimentally achieved at least via CAR- and GR-mediated and possibly also via VDR- and PPAR-alpha-mediated mechanisms (Ferguson et al., 2005; Chen and Goldstein, 2009). The CAR-agonists phenytoin and CITCO have markedly induced CYP2C8 expression in human hepatocytes (Ferguson et al., 2005). In addition, dexamethasone (GR agonist) can modestly increase CYP2C8 expression in in vitro systems (Gerbal-Chaloin et al., 2001; Rae et al., 2001; Raucy et al., 2002; Madan et al., 2003; Ferguson et al., 2005). Moreover, VDR may be involved in the induction of CYP2C8 by lithocholic acid in HepG2 cells (Makishima et al., 2002; Yajima et al., 2014), and the PPAR-alpha-agonist WY14,643 (4-chloro-6-(2,3-xylidino)-2-pyrimidinylthioacetic acid) has induced CYP2C8 mRNA in HepaRG cells (Thomas et al., 2015). In addition to the above compounds, certain other drugs have weakly induced CYP2C8 in vitro with unknown induction mechanisms, including tasimelteon and crizotinib (FDA, 2011e,m; EMA, 2012c; TGA, 2014).
In humans in vivo, rifampin has markedly reduced the plasma exposure to several CYP2C8 substrates (Jaakkola et al., 2006a; Niemi et al., 2000; Niemi et al., 2004a; Park et al., 2004), and it is consequently the preferred CYP2C8 inducer drug for use in clinical studies (section VII). Yet, the strength of the CYP2C8 inducing effect of rifampin (or any other PXR ligands) has been difficult to estimate, because the strong CYP3A4-inducing effect of rifampin is likely to partially explain its effects on the clearance of CYP2C8 substrates. Of note, one of the strongest clinical inducers of CYP enzymes, carbamazepine, which is also a weak PXR activator, seems to be poorly characterized with regard to CYP2C8 both in vitro and in vivo.
VI. Clinical Drug Interactions Mediated via Cytochrome P450 2C8
A. General Aspects
The CYP2C8 enzyme is involved in many drug-drug interactions in humans, including interactions based on either inhibition or induction of CYP2C8. However, its exact role in interactions is difficult to determine, because there are no fully selective in vivo inhibitors or inducers of CYP2C8 and all known CYP2C8 substrates are metabolized, at least to a small degree, also by other enzymes. Furthermore, the activities of OATP1B1, P-glycoprotein or other membrane transporters can affect the pharmacokinetics of many CYP2C8 substrate drugs, and some inhibitors of CYP2C8 inhibit these transporters, too. Because drug metabolizing enzymes and transporters may influence drug metabolism in concert, the isolated role of CYP2C8 in many drug-drug interactions can be very difficult to dissect in vivo.
The clinical significance of pharmacokinetic drug-drug interactions depends both on the therapeutic index of victim drug and on the extent of pharmacokinetic changes, in addition to various patient-related clinical factors. Of the pharmacokinetic parameters, at least the plasma AUC, peak concentration (Cmax), time to maximum concentration (tmax), and elimination half-life (t1/2) values are generally required for the characterization of an interaction. Here, to be brief, we usually report only fold-changes of the mean AUC values caused by interactions. Of note, e.g., interindividual genetic variation in the activity of drug metabolizing enzymes and transporters can cause a considerable variability in the extent of drug interactions. In particular, it is important to note that in an individual patient the exposure to a victim drug can change much more than the generally reported mean change.
Recognition of the role of CYP2C8 as an important oxidative enzyme in drug metabolism has led to changes in product information of many drugs, resulting in better predictability of their interactions and improved safety. In some cases, CYP2C8-mediated drug interactions and the resultant adverse effects have forced the manufacturers to withdraw drugs from clinical use or to add contraindications or limitations to their use. In general, special attention is needed, if a drug with narrow safety margin is extensively metabolized by CYP2C8 or if a drug is a strong CYP2C8 inhibitor like gemfibrozil and clopidogrel. In addition, it should be recognized that rifampin and some other potent enzyme inducers can markedly reduce plasma concentrations and effects of CYP2C8 substrate drugs.
As gemfibrozil is a well-characterized inhibitor of CYP2C8 that has increased the plasma concentrations of several drugs (Fig. 7; Table 10), we first present a detailed description of it as an in vivo inhibitor in the following part. The other clinically relevant CYP2C8 inhibitors, including clopidogrel, trimethoprim, efavirenz, and teriflunomide (Niemi et al., 2004b; German et al., 2007; FDA, 2012a; Tornio et al., 2014), are dealt with in the next part, where we focus on the CYP2C8-inhibition-mediated drug interactions of different therapeutic CYP2C8 substrate drugs. Thereafter, we present CYP2C8 induction-mediated drug interactions and their clinical relevance.
B. Gemfibrozil as Prototypical Inhibitor
1. In Vitro Versus In Vivo.
In vitro, the parent gemfibrozil is a moderately potent competitive inhibitor of CYP2C9 (Ki value of 5.8 µM; Wen et al., 2001), but it is over 10 times less potent as inhibitor of CYP2C8 (Ki value of 75 µM, Wang et al., 2002; Ki 87 µM, Prueksaritanont et al., 2002; Table 6). Gemfibrozil in concentrations up to 1,000 µM has no effect on CYP3A4 activity (midazolam 1′-hydroxylation) (Backman et al., 2000), but it is rather potent as an inhibitor of the OATP1B1 transporter, with a Ki value ranging in different studies from 4 to 31.7 µM (Schneck et al., 2004; Yamazaki et al., 2005; Hirano et al., 2006; Nakagomi-Hagihara et al., 2007).
In healthy volunteers, gemfibrozil (600 mg twice daily) slightly (by 23%) increased the AUC of a CYP2C9 substrate drug glimepiride (Niemi et al., 2001) but did not increase the exposure to racemic warfarin (Lilja et al., 2005). Gemfibrozil even caused a small but statistically significant decrease (−11%) in the AUC of the CYP2C9 substrate S-warfarin. These results strongly suggest that gemfibrozil is not a meaningful inhibitor of CYP2C9 in vivo in humans.
In vivo gemfibrozil is glucuronidated by the UGT 2B7 enzyme to gemfibrozil 1-O-β-glucuronide. The benzylic oxidation of the glucuronide by CYP2C8 leads to haem alkylation and irreversible inactivation of CYP2C8 (Baer et al., 2009; Jenkins et al., 2011). In HLM, the kinact value of CYP2C8 by gemfibrozil 1-O-β-glucuronide has been 0.21 1/min and KI 20–52 µM (Ogilvie et al., 2006). The glucuronide metabolite is also a competitive inhibitor of OATP1B1 (OATP2) transporter (Ki 24 µM; Shitara et al., 2004). On the basis of clinical studies on the dose/time dependency of the effect of gemfibrozil on the pharmacokinetics of repaglinide and statistical models of enzyme and transporter inhibition, it has been estimated that the in vitro mechanism-based inhibition of CYP2C8 by gemfibrozil 1-O-β-glucuronide manifests into a strong and long-lasting inhibition of CYP2C8 at typical clinical doses of gemfibrozil (Backman et al., 2009; Honkalammi et al., 2011a,b). In addition, the OATP1B1 inhibitory effect of the glucuronide can lead to an up to ∼50% transient inhibition of OATP1B1 in vivo. These effects are the main explanation to the effects of gemfibrozil on CYP2C8 and OATP1B1 substrates (Honkalammi et al., 2012). Similar estimations have been obtained with physiologically based pharmacokinetic modeling in a recent publication (Varma et al., 2015).
2. Gemfibrozil Dose Versus CYP2C8 Inhibition.
Single oral doses of gemfibrozil, i.e., 30, 100, 300, or 900 mg ingested 1 hour before repaglinide, increased the AUC of repaglinide in a dose-dependent manner 1.8-, 4.5-, 6.7-, and 8.3-fold compared with placebo, respectively (Fig. 8; Honkalammi et al., 2011a). Also after multiple doses of gemfibrozil (30, 100, or 600 mg twice daily for 5 days), the exposure to repaglinide increased dose dependently, but the greatest AUC increase did not exceed that observed after the single 900 mg gemfibrozil dose (Honkalammi et al., 2012). Thus, the maximum inhibition of CYP2C8 can be achieved by a single 900-mg dose of gemfibrozil (Fig. 9). Gemfibrozil in doses of 100 mg twice daily at steady state causes an about 95% inhibition of CYP2C8 (Fig. 9), and in doses of 10 mg twice daily, it causes an about 50% inhibition (Honkalammi et al., 2011a). The fraction of a small 0.25-mg dose of repaglinide metabolized by CYP2C8 is about 80–90%. However, because repaglinide is metabolized to some extent also by CYP3A4, the relative role of CYP2C8 and CYP3A4 in the biotransformation of repaglinide can depend on its dose and plasma concentrations as well as on individual pharmacogenetic factors (Bidstrup et al., 2003; Kajosaari et al., 2005a; Säll et al., 2012).
3. Onset and Duration of CYP2C8 Inhibition by Gemfibrozil.
The onset and duration of CYP2C8 inhibition by gemfibrozil have been studied in healthy volunteers using the gemfibrozil-repaglinide interaction as a model. Single 600-mg doses of gemfibrozil ingested 0, 1, 3, or 6 hours before repaglinide (0.25 mg) increased the geometric mean AUC of repaglinide 5.0-, 6.3-, 6.6-, and 5.4-fold, respectively (Fig. 8). The Cmax of the CYP2C8-mediated repaglinide M4-metabolite was 1.0-, 0.10-, 0.06-, and 0.09-fold compared with control phase, respectively (Honkalammi et al., 2011b). These results indicate that the strong inactivation of CYP2C8 occurs rapidly, being evident already within 1 hour after oral dosing of gemfibrozil.
When repaglinide was ingested 1, 24, 48, or 96 hours after discontinuation of a gemfibrozil treatment (600 mg twice daily for 3 days), the AUC of repaglinide was 7.6-, 2.9-, 1.4-, and 1.0-fold compared with the control phase, respectively (Backman et al., 2009). These findings confirmed and extended the previous findings, which had shown that the inhibitory effect of gemfibrozil persists for at least 12 hours after its ingestion (Tornio et al., 2008a). As the half-lives of gemfibrozil and its glucuronide are very short (about 1–2 hours), these findings convincingly demonstrate that the effect of gemfibrozil on repaglinide pharmacokinetics is based on irreversible mechanism-based inhibition of CYP2C8. A several-fold increase in repaglinide AUC was evident even at very low plasma concentrations of gemfibrozil 1-O-β-glucuronide, which are less than 1% of its peak concentrations. The results also showed that full CYP2C8 activity recovers gradually within 3–4 days after cessation of the clinically used therapeutic doses of 600 mg twice daily gemfibrozil.
4. Quantification of CYP2C8-Mediated Drug Interactions in Humans.
Interactions caused by a combination of two or more drugs, which inhibit, in addition to CYP2C8, also some other crucial enzyme or transporter, can increase exposure to a CYP2C8 substrates much more than is the sum of their separate effects causing a classic potentiation phenomenon. Thus, e.g., exposure to repaglinide is increased only slightly by itraconazole alone (1.4-fold), greatly by gemfibrozil alone (8.1-fold), and drastically (19.4-fold) by their combination (Fig. 10; Niemi et al., 2003b). The quantitative rationalization of gemfibrozil-drug interactions and consideration of transporter-enzyme interplay have been dealt quite recently by Varma et al. (2015).
C. Inhibition-Mediated Drug Interactions and Their Clinical Significance
Interactions of the oral antidiabetic drug repaglinide have been studied extensively, and it is a recommended model substrate drug for CYP2C8 interaction studies (EMA, 2012b; http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/ucm093664.htm). In healthy volunteers, gemfibrozil (600 mg twice daily for 3 days) raised the AUC of repaglinide 8.1-fold, itraconazole raised it 1.4-fold, and their combination raised it 19.4-fold (Niemi et al., 2003b). Gemfibrozil alone and in combination with itraconazole considerably enhanced also the blood glucose lowering effect of repaglinide (Niemi et al., 2003b). This pioneering study clearly indicated that the extent of interaction caused by a combination of two drugs can greatly exceed the sum of their separate effects. On the basis of these results and clinical observations of serious hypoglycemic episodes in diabetic patients, The European Agency for the Evaluation of Medicinal Products gave “EMEA public statement on repaglinide contraindication of concomitant use of repaglinide and gemfibrozil” (21.05.2003). Also the U.S. Food and Drug Administration warned against repaglinide-gemfibrozil interaction. The effect of gemfibrozil on repaglinide exposure was later confirmed and characterized in several studies as described in previous paragraphs (sections VI.B.1–3). The gemfibrozil-repaglinide interaction is mainly mediated via inhibition of CYP2C8 and OATP1B1 by the gemfibrozil 1-O-β-glucuronide.
The antimicrobial drug trimethoprim (160 mg twice daily for 3 days) raised in healthy volunteers the AUC of repaglinide by 1.6-fold compared with placebo (Niemi et al., 2004b). Symptomatic hypoglycemia developed in a diabetic patient 5 days after addition of trimethoprim/sulfamethoxazole therapy to his previously well-tolerated repaglinide (1 mg three times daily) treatment (Roustit et al., 2010).
The 300-mg loading dose of clopidogrel raised the AUC of repaglinide by 5.1-fold, and the following daily 75-mg doses of clopidogrel raised the AUC by 3.9-fold in healthy volunteers (Tornio et al., 2014). The increase in repaglinide AUC caused by clopidogrel was highest in subjects with the CYP2C8*1/*4 genotype (Tornio et al., 2014). The clopidogrel-repaglinide interaction is mediated by formation of the clopidogrel acyl-β-d-glucuronide, which is a potent time-dependent inhibitor of CYP2C8. On the basis of this short-term study, it has been extrapolated that the daily treatment with 75 mg of clopidogrel causes a continuous, 60–85% inhibition of hepatic CYP2C8 under steady-state conditions during chronic clopidogrel use. The pharmacokinetic interaction of clopidogrel and repaglinide resulted in an enhanced blood glucose-lowering effect of repaglinide. The concomitant use of repaglinide and clopidogrel is now contraindicated, e.g., in Canada (http://healthycanadians.gc.ca/recall-alert-rappel-avis/hc-sc/2015/54454a-eng.php).
The immunosuppressant teriflunomide (70 mg once daily for 4 days, followed by 14 mg once daily for 8 days) increased the AUC of repaglinide by 2.3-fold in healthy male subjects (FDA, 2012a) compared with when repaglinide was given alone. In three subjects the AUC was raised by 3.2- to 3.6-fold. Teriflunomide inhibits both CYP2C8 and OATP1B1, and the contribution of each mechanism to the increase in repaglinide exposure has not been established (FDA, 2012a). Of note, teriflunomide is the active metabolite of leflunomide and its plasma concentrations following leflunomide administration are equal to those observed when it is given alone. Hence, leflunomide may also be a clinically relevant CYP2C8 inhibitor.
Care is warranted if inhibitors of CYP2C8 are combined with repaglinide. In particular, combination of strong CYP2C8 inhibitors, such as gemfibrozil and clopidogrel, with repaglinide should be avoided. Blood glucose levels and symptoms of hypoglycemia should be monitored closely and the doses modified as needed. The interactions with repaglinide are likely to be stronger in CYP2C8*3 carriers than in CYP2C8*1 homozygotes (Tornio et al., 2008a).
2. Other Oral Antidiabetic Drugs.
The European product information of Actos (pioglitazone) stated earlier (e.g. 2004) that metabolism of pioglitazone occurs predominantly via CYP3A4 and CYP2C9 (Jaakkola et al., 2005), whereas the U.S. label stated that the major CYP isoforms involved were CYP2C8 and CYP3A4 (FDA, 1999). The in vitro study of Jaakkola et al. (2006c) showed that pioglitazone is metabolized mainly by CYP2C8 and to lesser extent by CYP3A4, whereas CYP2C9 is not significantly involved in pioglitazone elimination. In healthy volunteers, gemfibrozil raised the mean AUC of pioglitazone 3.2-fold (range 2.3-fold to 6.5-fold) and its elimination half-life 2.7-fold, but itraconazole had no effect on pioglitazone and did not alter the effect of gemfibrozil on its pharmacokinetics (Jaakkola et al., 2005). In two other studies, gemfibrozil increased the mean AUC of pioglitazone 3.4-fold (Deng et al., 2005) and 4.3-fold (range 1.3-fold to 12.1-fold) (Aquilante et al., 2013a). CYP2C8 genotype influences the relative change in pioglitazone exposure after gemfibrozil administration. Thus, CYP2C8*3 carriers had a greater mean increase by gemfibrozil in pioglitazone AUC (5.2-fold) compared with CYP2C8*1 homozygotes (3.3-fold) (Aquilante et al., 2013a). Trimethoprim (160 mg twice daily) raised in healthy volunteers the AUC of pioglitazone by 1.4-fold and had opposite effects on pioglitazone pharmacokinetics compared with the effects of CYP2C8*3 allele during the placebo phase (Tornio et al., 2008b).
The AUC of rosiglitazone was raised in healthy volunteers by gemfibrozil by 2.3-fold (Niemi et al., 2003a). In another study, trimethoprim (160 mg twice daily) increased rosiglitazone AUC by 1.4-fold and reduced the formation of N-demethylrosiglitazone (Niemi et al., 2004a). The effect of trimethoprim (200 mg twice daily) on rosiglitazone pharmacokinetics was confirmed by Hruska et al. (2005), who also demonstrated the competitive inhibition of rosiglitazone p-hydroxylation by trimethoprim in vitro. Atazanavir (400 mg once daily) increased the AUC of a single dose of rosiglitazone by 1.4-fold (FDA, 2015b).
The AUC of nateglinide was increased only by 1.5-fold by 3 days' pretreatment with therapeutic doses of both gemfibrozil and itraconazole (Niemi et al., 2005a). Thus, neither CYP2C8 nor CYP3A4 has a substantial significance to the pharmacokinetics of nateglinide. Gemfibrozil increased also the AUC of the dipeptidyl peptidase inhibitor sitagliptin by 1.5-fold (Arun et al., 2012). However, the gemfibrozil-sitagliptin interaction seems to be mainly mediated by inhibition of the renal OAT3, with a minor contribution by CYP2C8.
If gemfibrozil, clopidogrel, or other inhibitors of CYP2C8 will be combined with pioglitazone or rosiglitazone, blood glucose levels, symptoms of hypoglycemia, and other potential adverse effects (e.g., fluid retention) should be monitored closely and the doses be modified as needed. As shown for the gemfibrozil-pioglitazone and gemfibrozil-repaglinide interactions (Tornio et al., 2008a; Aquilante et al., 2013a), interactions may be stronger in CYP2C8*3 carriers than in CYP2C8*1 homozygotes.
Although amodiaquine N-deethylation is a widely used marker reaction for CYP2C8 activity in vitro, the sensitivity of amodiaquine to CYP2C8 inhibition is poorly characterized in humans. Amodiaquine is rapidly and extensively metabolized by CYP2C8 to active desethylamodiaquine, which has a long half-life of 9–18 days. In healthy subjects, trimethoprim and efavirenz have been reported to increase the AUC of amodiaquine by 1.6- and 1.8-fold, respectively, and to reduce that of desethylamodiaquine by 12 and 26%, respectively (Soyinka et al., 2013; Akande et al., 2015). In an earlier study, efavirenz raised the AUC of amodiaquine in two healthy subjects by 2- to 4-fold and decreased the AUC of desethylamodiaquine by 24 and 8.5% (German et al., 2007). In both of these subjects, marked elevation of hepatic transaminase levels occurred several weeks after stopping the 3 days' combined use, forcing premature discontinuation of the interaction study. The dramatic, delayed hepatotoxicity warrants great care in combination of any CYP2C8 inhibitor with amodiaquine.
Cerivastatin was initially considered as a safe statin because of its dual biotransformation routes, mediated both via CYP3A4 and CYP2C8 (Mück, 1998; 2000). However, it soon became obvious that cerivastatin greatly increased the incidence of fatal rhabdomyolysis, particularly when taken along with gemfibrozil (Staffa et al., 2002). Consequently, cerivastatin was withdrawn from the market in 2001, only 3 years after its launch. Despite the “dual metabolic pathway” and supposed "low propensity for drug interactions" (Mück et al., 1998; Mück, 2000), the elimination of cerivastatin relied predominantly on CYP2C8. In healthy volunteers, gemfibrozil (600 mg twice daily) raised the AUC of the parent cerivastatin (acid) by 5.6-fold, the AUC of cerivastatin lactone by 4.4-fold, and that of the CYP3A4-dependent metabolite M-1 by 4.35-fold, whereas gemfibrozil decreased the AUC of the CYP2C8-dependent metabolite M-23 by 78% (Fig. 11; Backman et al., 2002). The increased exposure to cerivastatin, to its lactone, and to M-1 and the reduced formation of the CYP2C8-dependent metabolite revealed the strong CYP2C8 inhibitory effect of gemfibrozil. In addition to irreversible inhibition of CYP2C8 by gemfibrozil 1-O-β-glucuronide, inhibition of the hepatic OATP1B1 may contribute to the gemfibrozil-cerivastatin interaction (Ogilvie et al., 2006; Shitara et al., 2004; Tamraz et al., 2013).
Interestingly, about 10 years after the withdrawal of cerivastatin, it was found that in addition to gemfibrozil, also concomitant use of clopidogrel was strongly associated with cerivastatin-induced rhabdomyolysis, with an odds ratio of ∼30 (48 when gemfibrozil users were excluded) (Floyd et al., 2012). Recently, Tornio et al. (2014) showed that glucuronidation converts clopidogrel to a strong time-dependent inhibitor of CYP2C8, clopidogrel acyl-β-d-glucuronide. The formation of this metabolite leads to uninterrupted inhibition of CYP2C8 during clopidogrel treatment and explains the increased risk of rhabdomyolysis during concomitant use of cerivastatin and clopidogrel (Tornio et al., 2014). Also trimethoprim increased the AUC of cerivastatin (by 1.4-fold) and its lactone (1.5-fold) (Backman et al., 2003). Because cerivastatin has been withdrawn from the market, its interactions are no more of direct clinical relevance. However, they are examples of clinically important challenges in drug development and have been of paramount importance in understanding the significance of CYP2C8 in drug metabolism.
Interestingly, cerivastatin and repaglinide have pharmacokinetic similarities. Both drugs are substrates of CYP2C8, CYP3A4, and OATP1B1. The CYP3A4 inhibitor itraconazole has raised their AUC only slightly, i.e., by 1.4-fold (repaglinide), 1.15-fold (cerivastatin acid), and 1.8-fold (cerivastatin lactone), whereas gemfibrozil has raised their AUC values much more, i.e., by 8.1-fold (repaglinide), by 5.6-fold (cerivastatin), and by 4.4-fold (cerivastatin lactone) (Kantola et al., 1999; Backman et al., 2002; Niemi et al., 2003b). Because the combination of a CYP3A4 inhibitor and a CYP2C8 inhibitor caused a drastic increase in repaglinide AUC (by 19.4-fold; Niemi et al., 2003b), it is reasonable to assume that also the exposure to cerivastatin acid and to its more lipophilic lactone form have raised even more by gemfibrozil —or clopidogrel—if the patients had been using also CYP3A4 inhibiting drugs. However, there seems to be no studies on the effect of CYP2C8 and CYP3A4 inhibitor combinations on the plasma concentrations of cerivastatin.
Gemfibrozil raises the AUC of nearly all statin acids, including simvastatin acid, lovastatin acid, atorvastatin, pravastatin, rosuvastatin, and pitavastatin (Neuvonen et al., 2006). However, the role of CYP2C8 in some gemfibrozil-statin interactions seems to be limited or nonexistent. They are mainly mediated by inhibition of OATP1B1, OAT3, or other transporters (Shitara et al., 2004; Neuvonen, 2010; Niemi et al., 2011).
5. Anticancer Drugs.
Most anticancer drugs have a narrow therapeutic range. Although paclitaxel is a well-established CYP2C8 probe in vitro, its interactions with CYP2C8 inhibitors and inducers have not been widely studied in humans. Lapatinib and pazopanib are relatively strong inhibitors of CYP2C8, and they have raised the AUC of paclitaxel up to 1.8-fold (Tan et al., 2014). In a case report, the only clopidogrel user in a cohort of 93 ovarian carcinoma patients treated with paclitaxel had the second lowest clearance of unbound paclitaxel in the cohort. She was hospitalized three times because of severe paclitaxel toxicity (Bergmann et al., 2015).
Gemfibrozil has raised the AUC of the androgen receptor antagonist enzalutamide by 4.3-fold, and itraconazole raised it by 1.4-fold compared with control (Gibbons et al., 2015). These results agree well with the in vitro findings that CYP2C8 is the predominant enzyme in the elimination of enzalutamide. The composite exposure of enzalutamide and its active metabolite was raised by 2.2-fold by gemfibrozil and by 1.3-fold by itraconazole. A reduction of the enzalutamide dose by about 50% is recommended when gemfibrozil is used concomitantly. There are no published studies on the effect of clopidogrel on enzalutamide pharmacokinetics. However, a close follow up and reduction of enzalutamide dose can be recommended also in their possible coadministration. It should also be noted that combined inhibition of CYP2C8 and CYP3A4 can cause a greater increase in enzalutamide + metabolite AUC. Enzalutamide itself is an inhibitor of CYP2C8 and may moderately raise the exposure to its substrate drugs, e.g., pioglitazone AUC by 20% (Gibbons et al., 2015).
Gemfibrozil did not affect the AUC of imatinib after a single imatinib dose but reduced the AUC of N-demethylimatinib by 48%, indicating a significant participation of CYP2C8 in the metabolism of imatinib in humans (Filppula et al., 2013b). After a single dose, imatinib seems to be mainly metabolized by CYP3A4, but the fraction of imatinib metabolized by CYP3A4 decreases after its multiple doses because of autoinhibition of the CYP3A4-mediated metabolism of imatinib (Filppula et al., 2012, 2013a). This autoinhibition is likely to increase the relative role of CYP2C8 in imatinib elimination and its sensitivity to interactions caused by CYP2C8 inhibitors during long-term treatment. According to pharmacokinetic simulations, imatinib exposure may raise up to twofold at steady state if a strong CYP2C8 inhibitor is given concomitantly with imatinib (Filppula et al., 2013b).
In melanoma patients, gemfibrozil increased the AUC of the CYP3A4 and CYP2C8 substrate dabrafenib by 1.5-fold and ketoconazole increased it by 1.7-fold (Suttle et al., 2015). It is probable that a combined administration of CYP2C8 inhibitors and CYP3A4 inhibitors with dabrafenib can increase its exposure more than does either of these inhibitors alone. In addition to paclitaxel, dabrafenib, and imatinib, some other anticancer drugs are metabolized by CYP2C8 (Table 1). However, their interactions with CYP2C8 inhibitors have not been characterized in humans. Considering the narrow therapeutic range of many anticancer drugs, close follow up for possible adverse effects is warranted if gemfibrozil, clopidogrel, trimethoprim, or other inhibitors of CYP2C8 are used with paclitaxel or other anticancer drugs metabolized by CYP2C8.
6. Antiviral Drugs.
Gemfibrozil (600 mg twice daily) has increased the AUC of the antihepatitis C drug dasabuvir about 11-fold and their concomitant use is contraindicated (Menon et al., 2015). It is reasonable to assume that also other potent inhibitors of CYP2C8 such as clopidogrel increase greatly the exposure to dasabuvir, and their use together should be avoided or the dose of dasabuvir be reduced markedly. It can be speculated that savings could be achieved by using small amounts of expensive dasabuvir (about one-tenth of normal dose) with small doses (e.g., 100 mg) of gemfibrozil. This should lead to similar plasma concentrations of dasabuvir as those achieved by normal dasabuvir doses administered without inhibitor of CYP2C8. Also some other new antiviral drugs are partially metabolized by CYP2C8, but their susceptibility to interact with drugs affecting CYP2C8 activity in humans needs further studies.
7. Antiasthmatic Drugs.
In healthy volunteers. gemfibrozil raised the AUC of montelukast 4.5-fold and its elimination half-life 3.0-fold (Karonen et al., 2010). Gemfibrozil reduced the AUC of the secondary M4 metabolite of montelukast by more than 90%. In another study, gemfibrozil alone raised the AUC of montelukast 4.3-fold, itraconazole had no significant effects, and the effects of the gemfibrozil-itraconazole combination on montelukast pharmacokinetics did not differ from those of gemfibrozil alone (Karonen et al., 2012). These findings indicate that CYP2C8 but not CYP3A4 is important in the pharmacokinetics of montelukast. In contrast to the effect of gemfibrozil on montelukast, the pharmacokinetics of zafirlukast is not affected by gemfibrozil (Karonen et al., 2011), although both of these cysteinyl leukotriene receptor antagonists are potent in vitro inhibitors of CYP2C8 (Walsky et al., 2005a).
Montelukast has a relatively large safety margin, and the clinical significance of its interactions with CYP2C8 inhibitors seems to be limited. However, neuropsychiatric symptoms have developed in a woman with HIV infection when montelukast was added to her therapy containing the CYP2C8 inhibitor efavirenz (Ibarra-Barrueta et al., 2014). She had used efavirenz, emtricitabine, and tenofovir disoproxil fumarate for years with good tolerance until montelukast was started for asthma. Shortly thereafter unbearable symptoms appeared, consisting of disturbed sleep, vivid dreams and irritability, confusion, and concentration difficulties. After 2 months of concomitant use, montelukast was withdrawn and the psychiatric symptoms completely disappeared. This case report indicates that adverse effects can develop when these drugs are used together, although the mechanism of adverse effects is not fully clear.
8. Other Substrate or Inhibitor Drugs.
Gemfibrozil raised in healthy volunteers the AUC of loperamide 2.1-fold, itraconazole raised it 3.8-fold, and the gemfibrozil-itraconazole combination raised loperamide AUC 12.6-fold compared with placebo phase (Niemi et al., 2006). This finding strongly suggests that gemfibrozil can markedly increase the loperamide exposure in subjects who are using potent inhibitors of CYP3A4, i.e., when another important metabolic route is blocked. Administration of cotrimoxazole (trimethoprim + sulphamethoxazole) has increased the AUC of loperamide by 1.9-fold (Kamali and Huang, 1996). Also some other opioids, e.g., buprenorphine, are CYP2C8 substrates (Table 1). However, there seem to be no published studies on their possible interaction with gemfibrozil or other CYP2C8 inhibitors.
Gemfibrozil raised the AUC of the prolyl hydroxylase inhibitor agent daprodustat (GSK1278863) 18.6-fold (Johnson et al., 2014). This result together with in vitro studies indicates the crucial significance of CYP2C8 in its pharmacokinetics. CYP2C8 inhibitors should not be used with this erythropoiesis-stimulant agent or its dose needs to be reduced very greatly. On the other hand, at least theoretically, it could be possible to take advantage of this interaction in a product containing very small doses of daprodustat and gemfibrozil.
In healthy volunteers, gemfibrozil raised only slightly the AUC of R-ibuprofen, by 1.3-fold, after the ingestion of racemic ibuprofen (Tornio et al., 2007). In vitro CYP2C8 participates in the metabolism of zopiclone (Becquemont et al., 1999). In humans, however, gemfibrozil did not increase the AUC of the parent zopiclone but moderately (2-fold and 1.2-fold) increased the AUC of its potentially active metabolites (Tornio et al., 2006). Also many other drugs are substrates of CYP2C8 in vitro, but their concomitant administration with gemfibrozil has not appreciably increased their AUC, suggesting that the CYP2C8-mediated biotransformation is of limited significance to their total clearance (Table 1).
Many compounds are moderate inhibitors of CYP2C8 in vitro, but their concomitant ingestion with repaglinide or other CYP2C8 substrates does not raise exposure to these substrates in humans. The reason for the apparent discrepancy between the in vitro and in vivo results can be, for example, their low potency as CYP2C8 inhibitors or their high protein binding in vivo (e.g., montelukast).
Some parent drugs such as gemfibrozil and clopidogrel are relatively weak inhibitors of CYP2C8 in vitro, but they are metabolized in vivo to glucuronide metabolites, which are potent CYP2C8 inhibitors. In general, negative interaction results with gemfibrozil in vivo exclude a clinically meaningful interaction mediated by CYP2C8 inhibition. On the other hand, increased exposure to a victim drug by gemfibrozil does not yet indicate that CYP2C8 has a significant role in its metabolism because there may be other mechanisms mediating the observed interaction.
Patients often concomitantly use different drugs that together inhibit several CYP enzymes, e.g., CYP1A2, CYP2C8, CYP2C9, CYP2B6, CYP2D6, or CYP3A4. The combined inhibition of two or more of these enzymes often results in patients in a stronger interaction than is caused by inhibition of a single enzyme in healthy volunteer studies. This aspect together with other causes of interindividual variation should be taken into consideration when the results of experimental interaction studies in healthy volunteers are translated into the clinic.
D. Induction-Mediated Drug Interactions
Rifampin (rifampicin) can markedly increase the clearance of many CYP2C8 substrate drugs, decrease their AUC, and diminish their clinical efficacy. Both CYP2C8 and CYP3A4 are involved in the biotransformation of many drugs, which can also be substrates of various transporters (Table 1). Because both CYP2C8 and CYP3A4 enzymes and some transporters can be highly inducible, the importance of CYP2C8 in many rifampin interactions is difficult to determine exactly (Niemi et al., 2000). Apart from rifampin, there are very few clinical studies concerning the effects of other CYP enzyme inducers on the pharmacokinetics of CYP2C8 substrates.
1. Rifampin (Rifampicin).
Rifampin (600 mg/day), given for several days, has decreased the plasma exposure to repaglinide by 31–80% depending on the time interval from the last rifampin dose to repaglinide ingestion (Table 11; Niemi et al., 2000; Hatorp et al., 2003; Bidstrup et al., 2004). The time interval affects the extent of interaction because rifampin is also a competitive inhibitor of OATP1B1, CYP2C8 and CYP3A4 (Kajosaari et al., 2005a; Varma et al., 2013). Interestingly, intake of St John’s Wort for 14 days has had no significant effect of the pharmacokinetics of repaglinide (Fan et al., 2011).
Rifampin has also reduced the concentrations of the thiazolidinediones pioglitazone and rosiglitazone. Rifampin caused a substantial (54%) decrease in the AUC of pioglitazone and increased the ratios of metabolite M-IV to pioglitazone and of M-III to pioglitazone in urine by 98 and 95% (Jaakkola et al., 2006a). Similarly, rifampin reduced the mean AUC of rosiglitazone by 54% and increased the formation of N-demethylrosiglitazone (Niemi et al., 2004a). In Korean men, rifampin decreased rosiglitazone AUC by 65% (Park et al., 2004). Addition of tuberculosis treatment, containing rifampin, to treatment of a woman with type 2 diabetes caused her to lose glycemic control, demonstrating potential clinical significance of the rifampin-rosiglitazone interaction (Pimazoni, 2009).
VII. Points to Consider When Investigating Cytochrome P450 2C8-Mediated Drug Metabolism and Interactions
Studies focusing on drug metabolism and metabolic drug-drug interactions are an essential part of modern drug development, from early preclinical phases to the clinical development phase and beyond. By using specific and sensitive research methods, it is possible to get a detailed and accurate view of potential issues related to variability in drug metabolism already during the preclinical and early clinical phases of development. Methods to investigate CYP2C8 in vitro and in clinical studies have evolved markedly even during the last decade.
A. In Vitro
1. General Aspects.
Comprehensive in vitro studies to investigate the roles of different CYP enzymes in the metabolism of a (new) drug and to uncover its potential for causing inhibition or induction of drug metabolism are typically conducted already during the early preclinical phases of drug development. The results from these studies are then used for in vitro-in vivo extrapolations, to anticipate factors affecting the clearance of the drug as well as its potential to act as a perpetrator of pharmacokinetic drug interactions, i.e., to affect the clearance of other drugs. The prerequisite for accurate extrapolations is that in vitro investigations are conducted with care and are sufficiently comprehensive, avoiding the many pitfalls of in vitro studies, understanding the many limitations of the different approaches, and covering complex issues, such as the potential for autoinhibition or -induction. Yet it should be understood that accurate extrapolations are not possible without some clinical pharmacokinetic data at the relevant dose of the investigational drug.
The general aspects as well as the potential pitfalls of in vitro studies and extrapolations are well covered by many excellent review articles and guidelines (Houston and Galetin, 2008; Pelkonen et al., 2008; Grimm et al., 2009; EMA, 2012b; Pelkonen, 2015; http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DevelopmentResources/DrugInteractionsLabeling/ucm093664.htm). Therefore, this review focuses on issues that are directly related to CYP2C8, i.e., in vitro methods used for measurement of CYP2C8 activity (e.g., to test the potential of the investigational drug to inhibit CYP2C8 activity) and reaction phenotyping (does CYP2C8 metabolize the drug) and in vivo studies to characterize the drug interaction potential of the new drug (either as a perpetrator or victim drug).
2. Assessment of CYP2C8 Activity In Vitro.
Specific assessment of CYP2C8 activity is necessary, in particular when studying the potential of a drug to cause inhibition of CYP2C8 but also when using a panel of HLM for reaction phenotyping using the correlation approach. An ideal in vitro probe substrate is selective/specific, has a sufficient turnover, follows Michaelis-Menten kinetics, and is not sensitive to experimental conditions. There are several useful, selective probe substrates to study CYP2C8 in vitro, including paclitaxel, amodiaquine, montelukast, rosiglitazone, pioglitazone, and cerivastatin, each having its specific strengths and weaknesses (Tables 12 and 13).
Paclitaxel 6-α-hydroxylation is the prototypical marker reaction for CYP2C8 (Rahman et al., 1994; Sonnichsen et al., 1995). It is highly selective for CYP2C8, but the metabolic turnover is fairly low, often leading to relatively long incubation times, which may lead to significant inhibitor metabolism/depletion during the incubation (Table 13). This may partly explain why paclitaxel seems to be less sensitive to competitive CYP2C8 inhibitors than most other CYP2C8 marker substrates (VandenBrink et al., 2011). In particular, long incubation times should be avoided when studying the potential for time-dependent or mechanism-based inhibition in systems based on a preincubation step, because inactivation proceeding during the incubation may decrease the sensitivity of the experimental system to detect inactivation.
Amodiaquine metabolism to N-desethylamodiaquine is probably the second most used CYP2C8 marker reaction. It is well characterized and highly selective for CYP2C8 and has a high turnover (Li et al., 2002), allowing for short incubation times. Overall, it seems to have no major drawbacks in in vitro use.
Few years ago, montelukast, a selective competitive inhibitor of CYP2C8, was shown to be a potential CYP2C8 marker substrate, because its 36-hydroxylation (M6 formation) is mediated primarily by CYP2C8 with a minor contribution by CYP2C9 (Filppula et al., 2011). In a successive study, montelukast 36-hydroxylation proved to be a sensitive and useful reaction to investigate CYP2C8 inhibition in vitro (VandenBrink et al., 2011). One of the weaknesses of montelukast is that it is highly susceptible to microsomal protein binding, necessitating careful standardization of incubation conditions (Walsky et al., 2005b).
Of the other potential marker reactions, cerivastatin 6-hydroxylation (M-23 formation) seems to be highly specific for CYP2C8 (Wang et al., 2002; Shitara et al., 2004). In addition, the hydroxylations of rosiglitazone (p-hydroxylation) (Baldwin et al., 1999) and pioglitazone (M-IV formation; Jaakkola et al., 2006c) seem to be relatively, albeit not completely, selective for CYP2C8. Finally, the most used in vivo CYP2C8 probe drug repaglinide, although sometimes recommended as an in vitro probe (Kajosaari et al., 2005a; VandenBrink et al., 2011), is challenging to use in vitro, e.g., because of a need for extremely low substrate concentrations and lack of commercially available metabolite standards (Table 13).
3. In Vitro Methods to Estimate the Contribution of CYP2C8 in the Metabolism of a Drug.
The basic methods used for estimating the contributions of CYP enzymes to the metabolism of a drug, i.e., the so-called reaction phenotyping, are the use of diagnostic inhibitors in a complete natural system, such as HLM, and the use of recombinant expressed enzymes. In both approaches, knowledge of clinically relevant concentrations of the drug is a prerequisite for estimation of the contributions of the different CYP enzymes in vivo. The advantage of HLM is the natural composition of the system, allowing relatively straightforward estimation of the contributions. However, this approach requires human material collected according to high ethical standards and is entirely dependent on the strength and specificity of the inhibitors. On the other hand, although recombinant expressed enzymes can be regarded as a specific tool, in vivo extrapolations of recombinant enzyme results require the use of enzyme source and batch specific conversion factors (preferably based on enzyme activity), complicating the extrapolations.
Recombinant expressed human CYP2C8 is commercially available at least as bacterial cell- and insect cell-based products. During the last decade, both chemical inhibitors and inhibitory antibodies have become available that are both CYP2C8 specific and strong. In the following, we review the documentation regarding chemical CYP2C8 inhibitors.
One of the most widely used chemical CYP2C8 inhibitors is quercetin (Rahman et al., 1994). However, it is neither very selective for CYP2C8 nor very strong and therefore, it can barely be recommended as a diagnostic inhibitor. Today, there are several more selective alternatives available, including trimethoprim, montelukast, and gemfibrozil 1-O-β-glucuronide.
The IC50 of trimethoprim for CYP2C8 is approximately 50 µM, i.e., it is not a very strong inhibitor, but its IC50 for other CYP enzymes is at least one order of magnitude greater, making it a relatively selective inhibitor (Wen et al., 2002). Montelukast, on the other hand, is a potent and highly selective competitive inhibitor of CYP2C8, with an IC50 as low as 0.01 µM, when a low microsomal protein concentration is used, whereas its IC50 for other CYP enzymes is at least two orders of magnitude greater (Walsky et al., 2005b). The major drawback of montelukast seems to be its nonspecific microsomal protein binding, whereby increasing the microsomal protein concentration by 80-fold yields an about 100-fold decrease in its inhibition potency (Walsky et al., 2005b).
The mechanism-based CYP2C8 inactivator gemfibrozil 1-O-β-glucuronide is another appealing CYP2C8 inhibitor. With a 30-minute preincubation, its IC50 for CYP2C8 is about 2 µM, whereas its IC50 values for CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6, and CYP3A4 are more than 300 µM, suggesting an even better selectivity than that of montelukast (Ogilvie et al., 2006). Moreover, it is unlikely to be markedly affected by microsomal protein concentration. Whether clopidogrel acyl-β-d-glucuronide is a similarly selective CYP2C8 inactivator remains to be investigated (Tornio et al., 2014).
B. In Vivo
1. General Aspects.
Current guidelines recommend the conduct of clinical drug-drug interaction studies on the basis of in vitro studies on the CYP inhibitory effects of the drug and its main circulating metabolites (potential perpetrator), as well as on the basis of the results of the in vitro reaction phenotyping studies of the drug and its main metabolic pathways (victim). A rational selection for the first CYP-specific clinical studies is to focus on the enzyme that is inhibited most (lowest IC50/Ki) by the drug or its metabolite, preferably using the highest clinically used dose of the drug, and on the enzyme that is considered the most important in its own metabolism. For the first type of studies, a sensitive and selective in vivo probe substrate is used, and for the second type of studies, a strong and selective in vivo probe inhibitor is needed. For CYP2C8, there are several alternative probe substrates and a few inhibitors that can be used in clinical trials.
The contribution of CYP2C8 enzyme to the total clearance of its substrates varies greatly (Table 1). Most CYP2C8 substrates are partially metabolized also by other enzymes, are substrates of some membrane transporters, or are excreted in urine or feces in unchanged form. Thus, the significance of CYP2C8 in interactions cannot be calculated directly from changes in victim drug AUC. If the CYP2C8 substrate drug is also a substrate of transporters or other CYP enzymes, their contribution needs to be considered in the interaction, as exemplified in the dissection of the gemfibrozil-repaglinide interaction (Honkalammi et al., 2011a, 2012). For example, gemfibrozil is in vivo an inhibitor of CYP2C8 as well as of OATP1B1 and OAT3, and it can increase the AUC of certain drugs (e.g., pravastatin), which are not substrates of CYP2C8 but are substrates for OATP1B1 or OAT3 (Kyrklund et al., 2003). On the other hand, CYP2C8 inhibitors usually increase the AUC of CYP2C8 substrates less than they diminish the CYP2C8-specific metabolic routes, because the CYP2C8-independent elimination routes remain unaffected.
CYP2C8-mediated drug interactions are often studied in healthy volunteers in a randomized crossover manner by administering a potential substrate drug with and without a probe inhibitors of CYP2C8, such as the recommended probe inhibitor gemfibrozil (Table 12). To better simulate real clinical situations in which patients often are using several different drugs concomitantly, substrate drugs can be administered in multiple-phase studies, given alone, with an inhibitor of CYP2C8 (gemfibrozil), with an inhibitor of another relevant CYP enzyme (e.g., with CYP3A4 inhibitor itraconazole), and together with a combination of inhibitors. However, there are only a few studies in which the effects of multiple inhibitors (e.g., inhibitors of CYP2C8 and CYP3A4) on the pharmacokinetics of their substrate drugs have been studied both separately and together (Niemi et al., 2003b, 2006; Jaakkola et al., 2005; Karonen et al., 2012).
2. In Vivo Cytochrome P450 2C8 Probe Substrates.
The crucial characteristics of an in vivo probe substrate are its selectivity and sensitivity. In an ideal case, at least 80% of the substrate is metabolized by the enzyme of interest, allowing for an interpretation based on the AUC of the parent drug. In some cases, the use of an enzyme-specific metabolite to parent ratio may be used to increase sensitivity and specificity, but with an additional caveat because of potential variability in metabolite elimination. The feasibility of an in vivo probe substrate also depends heavily on its safety, in particular when large increases in its systemic concentrations can be anticipated. Moreover, the pharmacokinetic characteristics of the probe may affect its suitability. For example, a probe substrate with a significant first-pass metabolism and short elimination half-life may be able to catch even transient changes in enzyme activity, which may be necessary when studying inhibitors with a short half-life or time-dependent changes in enzyme activity.
The antidiabetic agent repaglinide is overall the most studied and best documented in vivo probe substrate of CYP2C8, and consequently both the European Medicines Agency (EMA) and FDA recommend its use as a CYP2C8 probe (Table 12). Although in vitro studies are not fully consistent with the major in vivo role of CYP2C8 in the total metabolism of repaglinide (Gan et al., 2010; Säll et al., 2012; Varma et al., 2013, 2015), repaglinide seems to be very sensitive to inhibitors of CYP2C8 activity, such as gemfibrozil, trimethoprim, and clopidogrel (Niemi et al., 2003b, 2004b; Tornio et al., 2014). On the basis of detailed mechanistic drug-drug interaction studies with the strong CYP2C8 inactivator gemfibrozil, it has been estimated that the contribution of CYP2C8 to repaglinide (0.25 mg) metabolism is about 85%, indicating that the AUC of repaglinide can be increased up to an average of sevenfold by strong CYP2C8 inhibition (Honkalammi et al., 2012). The half-life of repaglinide is also relatively short (1 hour), which allows for a full pharmacokinetic study within 1 day and can be beneficial when a measure of CYP2C8 activity within a narrow time frame is desired. The weakness of repaglinide is that it is partially metabolized by CYP3A4 (Bidstrup et al., 2003; Niemi et al., 2003b; Kajosaari, 2005a) and also a substrate of OATP1B1 (Niemi et al., 2005b). Thus, e.g., the effect of gemfibrozil on repaglinide pharmacokinetics is partially mediated by inhibition of OATP1B1, in addition to inhibition of CYP2C8 (Honkalammi et al., 2011a). Moreover, as it increases insulin secretion from pancreatic β cells, there is a risk of hypoglycemia, particularly when it is given to healthy subjects. Thus, the smallest possible dose (e.g., 0.25 mg) of repaglinide should be used, and meals, close follow up, and blood glucose monitoring be arranged when repaglinide is used.
Theoretically, the antimalarial agent amodiaquine and its N-deethylation could be useful in vivo CYP2C8 probes. However, there is very little clinical documentation for its use as a probe drug. Moreover, the safety of amodiaquine as an in vivo probe drug in drug-drug interactions studies seems to be questionable (German et al., 2007).
Of the other potential in vivo probe substrates of CYP2C8, the two thiazolidinediones pioglitazone and rosiglitazone are the best documented. As pointed out in the previous section, CYP2C8 is the main enzyme mediating their primary hydroxylation reactions in vitro (Baldwin et al., 1999; Jaakkola et al., 2006c). Accordingly, the typical dosing of gemfibrozil 600 mg twice daily, which has been estimated to cause over 95% inhibition of CYP2C8 (Fig. 9; Honkalammi et al., 2012), increased the AUC of rosiglitazone about 2.3-fold and that of pioglitazone 3.2-4.3-fold, simultaneously reducing the concentrations of their hydroxyl metabolites (Niemi et al., 2003a; Deng et al., 2005; Jaakkola et al., 2005; Aquilante et al., 2013a). Unlike repaglinide, they are insensitive to OATP1B1 function (Kalliokoski et al., 2008a). However, they have a long half-life, necessitating an up to 72-hour blood sampling period for a full pharmacokinetic analysis. On the basis of its better availability and sensitivity to CYP2C8 inhibition, pioglitazone is slightly preferable over rosiglitazone as a CYP2C8 probe.
The leukotriene receptor antagonist montelukast is another sensitive CYP2C8 substrate that could be used as a CYP2C8 probe. Gemfibrozil has increased its AUC almost fivefold and markedly reduced the formation of its 36-hydroxylated metabolite (Karonen et al., 2010). On the other hand, montelukast is also partially metabolized by CYP3A4 in vitro (Filppula et al., 2011; VandenBrink et al., 2011). However, the strong CYP3A4 inhibitor itraconazole has had no effect on montelukast concentrations (Karonen et al., 2012), indicating that the role of CYP3A4 in montelukast metabolism is minor. Moreover, montelukast is not known to be a substrate for OATP1B1.
In addition to the above substrates, there are some other CYP2C8 substrate drugs that could be used as in vivo markers on the basis of their sensitivity to interact with gemfibrozil. Such drugs include, for example, daprodustat and dasabuvir. However, the former is not yet on the market, and the second one is expensive, and more documentation is needed before they can be recommended as probe substrates.
3. In Vivo Cytochrome P450 2C8 Probe Inhibitors.
Probe inhibitors are needed for studying the contribution of CYP2C8 in the metabolism of a new drug, as well as for documenting the risk of drug-drug interactions affecting the drug in vivo. Among clinically used CYP2C8 inhibitors, gemfibrozil is the strongest known. Its CYP2C8 inhibitory effect is also highly selective due to the specific mechanism that is mediated via specific CYP2C8 inactivation by the glucuronide metabolite of gemfibrozil (Ogilvie et al., 2006). In vitro, parent gemfibrozil inhibits CYP2C9 activity with a fairly low Ki (about 6 µM), but its inhibitory effects on the other main CYP enzymes are much weaker (Backman et al., 2000; Wen et al., 2001; Wang et al., 2002). In clinical studies, gemfibrozil at a dose of 600 mg twice daily has not increased the concentrations of the CYP2C9 substrate warfarin (Lilja et al., 2005) or had any effect that could be due to inhibition of CYP3A4 on the concentrations of the parent lactone forms of simvastatin and lovastatin (Backman et al., 2000; Kyrklund et al., 2001). On the other hand, gemfibrozil has drastically, up to 18.6-fold, increased the AUCs of CYP2C8 substrate drugs (Fig. 7; Table 10), suggesting that with regard to CYP enzymes, the inhibitory effect of gemfibrozil is highly selective for CYP2C8.
The CYP2C8 inhibitory effect of gemfibrozil is strong, rapid, and long lasting. In studies using repaglinide as the CYP2C8 probe substrate, subtherapeutic doses of gemfibrozil have considerably elevated the concentrations of repaglinide (Honkalammi et al., 2011a, 2012), and it has been estimated that the clinically used gemfibrozil dosing (600 mg twice daily) inhibits CYP2C8 activity by about 99% and that one-tenth of this dose would already lead to more than 90% inhibition of CYP2C8 (Fig. 9). Although CYP2C8 inhibition by gemfibrozil is based on time-dependent inactivation of the enzyme by the primary glucuronide metabolite of gemfibrozil, CYP2C8 inhibition occurs rapidly after gemfibrozil dosing. When repaglinide was given 0, 1, 3, or 6 hours after a single 600 mg dose of gemfibrozil, the AUC of repaglinide was increased 5.0-, 6.3-, 6.6-, and 5.4-fold, respectively, indicating that strong inhibition of CYP2C8 can be achieved almost immediately after a single dose of gemfibrozil (Fig. 8, Honkalammi et al., 2011b). It has also been demonstrated that the CYP2C8 inhibitory effect of gemfibrozil persists virtually unchanged throughout the typical 12-hour dosing interval of gemfibrozil (Tornio et al., 2008a). Thus, gemfibrozil can have a strong effect on CYP2C8 substrates, irrespective of their half-life or time of daily dosing relative to gemfibrozil administration (Fig. 9; Table 10), making it an ideal in vivo probe inhibitor of CYP2C8. The only caveat with gemfibrozil is that it is also a moderate inhibitor of OATP1B1 and OAT3 and can therefore also increase the concentrations of some drugs that are not or only partially metabolized by CYP2C8 (Kyrklund et al., 2001, 2003; Backman et al., 2002; Schneck et al., 2004; Neuvonen et al., 2006; Whitfield et al., 2011).
Compared with gemfibrozil, all other clinically documented CYP2C8 inhibitors seem to be suboptimal. Trimethoprim is relatively selective for CYP2C8, but as expected from its in vitro inhibitory effects (Wen et al., 2002), it is only a weak CYP2C8 inhibitor at clinically feasible doses (Niemi et al., 2004a,b; Hruska et al., 2005; Tornio et al., 2008b), and therefore it can only be regarded as a confirmatory CYP2C8 inhibitor in vivo. Clopidogrel is the second strongest CYP2C8 inhibitor documented so far, increasing the AUC of repaglinide about fivefold (Tornio et al., 2014). Clopidogrel is obviously also a useful diagnostic CYP2C8 inhibitor, but it is not fully selective and has not been extensively documented so far. In addition to strongly inhibiting CYP2C8, clopidogrel is also a moderate inhibitor of CYP2B6 (Turpeinen et al., 2005). Furthermore, it has been suspected of causing CYP2C19 inhibition (Nishiya et al., 2009). However, it seems to have practically no effect on CYP3A4 or OATP1B1 activities in vivo (Tornio et al., 2014; Itkonen et al., 2015).
VIII. Conclusions and Future Prospects
CYP2C8 is one of the main oxidative drug metabolizing enzymes in the liver. Its expression and function have been studied in detail, and for example, it has been estimated that its in vivo turnover half-life is about 22 hours in humans. The CYP2C8 gene contains 9 exons and shares 74% sequence homology with CYP2C9. More than 100 nonsynonymous CYP2C8 SNVs are known to date, but only some of them are associated with functional variability. Interethnic and geographical differences exist in the frequency of variants. For example, the low-activity variant CYP2C8*2 (c.805A>T) is common in Africans but rare in Caucasians and Asians. CYP2C8*3 (c.416G>A) and CYP2C8*4 (c.792C>G), on the other hand, are common in Caucasians but rare or absent in Africans and Asians (Fig. 6). The interethnic characterization and functional activity of variants deserve further studies.
Studies on the role of CYP2C8 in drug metabolism have demonstrated that it is the most important enzyme for the elimination of several drugs, such as cerivastatin, montelukast, repaglinide, pioglitazone, and rosiglitazone, whose metabolism had been earlier thought to be attributed mainly to other enzymes. CYP2C8 is crucial also for the biotransformation of daprodustat (GSK12788693), enzalutamide, dasabuvir, and many other recently developed new drugs, and overall, it contributes to the elimination of more than 100 drugs. CYP2C8 has a large active site cavity, and it can accommodate and also metabolize certain acyl glucuronides, such as desloratadine, diclofenac, and sipoglitazar glucuronides.
In vitro, there are several marker reactions for the assessment of CYP2C8 activity, including paclitaxel 6-α-hydroxylation, amodiaquine deethylation, montelukast 36-hydroxylation, cerivastatin 6-hydroxylation (M-23 formation), rosiglitazone parahydroxylation, and pioglitazone M-IV formation. Each of these reactions has its strengths and weaknesses. The use of clinically relevant drug concentrations in vitro is a prerequisite for the estimation of the contribution of different CYP enzymes in vivo. Gemfibrozil 1-O-β-glucuronide is potent and selective as a diagnostic inhibitor of CYP2C8-mediated metabolism in vitro. Quercetin and trimethoprim are relatively weak and unselective as CYP2C8 inhibitors, whereas montelukast is potent and selective, but suffers from nonspecific protein binding. Also clopidogrel acyl 1-β-d-glucuronide may be a suitable in vitro inhibitor, but further documentation is needed.
Many drugs are CYP2C8 inhibitors or inducers. Gemfibrozil is in vivo, unlike in vitro, a potent, irreversible inhibitor of CYP2C8 via formation of gemfibrozil 1-O-β-glucuronide, and it is widely used as a probe inhibitor. Also clopidogrel acyl 1-β-d-glucuronide causes metabolism-dependent inactivation of CYP2C8, indicating that glucuronides may contribute as CYP2C8 inhibitors to drug-drug interactions. Also efavirenz, trimethoprim, and several protein kinase inhibitors are inhibitors of CYP2C8.
In vivo studies in humans on the role of CYP2C8 are challenged by the lack of suitable selective probe substrates, inhibitors, and inducers. Although repaglinide, pioglitazone, rosiglitazone, and montelukast are useful probe substrates, they all have their pros and cons, as discussed in the text before. With regard to CYP enzymes, gemfibrozil is a selective inhibitor of CYP2C8 in vivo. Already very small doses of gemfibrozil, i.e., about 10% of its usual therapeutic dose, rapidly cause a strong and long-lasting inactivation of CYP2C8. However, gemfibrozil is also an inhibitor of OATP1B1 and OAT3 transporters, which challenges interpretation of the interaction mechanisms if the CYP2C8 substrates are also substrates for these transporters. Rifampin is a very unselective albeit strong inducer of CYP2C8.
If a drug is significantly metabolized by CYP2C8 and CYP3A4, its concomitant administration with inhibitors of both of these enzymes, e.g., with gemfibrozil and itraconazole, can cause a much stronger interaction than is the sum of their separate effects. Thus, the drug-drug interaction studies performed by using inhibitors of one enzyme only may greatly underestimate the true risks as shown by the clinically very important CYP2C8-mediated interactions affecting repaglinide or cerivastatin. At least theoretically, small doses of gemfibrozil or other inhibitors of CYPC8 could be used as a booster to optimize the pharmacokinetics of CYP2C8 substrate drugs or to prevent formation of potentially toxic metabolites via CYP2C8-mediated reaction.
The authors thank Dr. Tommi Nyrönen for producing the docking simulations and three-dimensional artwork in Fig. 2.
Backman, Niemi, and Neuvonen have filed a patent application concerning use of gemfibrozil as a pharmacokinetic enhancer.
Some of the information in Tables 1, 3 and 6 is based on the UW Metabolism and Transport Drug Interaction Database (DIDB), Copyright University of Washington 1999–2015, as specified in the footnotes to the tables.
Participated in research design: Backman, Filppula, Niemi, and Neuvonen.
Wrote or contributed to the writing of the manuscript: Backman, Filppula, Niemi, and Neuvonen.
This work was supported by grants from the Academy of Finland [Grant decision 278123, 2014], the Helsinki University Central Hospital Research Fund, and the Sigrid Juselius Foundation (Helsinki, Finland).
- area under the plasma concentration-time curve
- CCAAT/enhancer-binding protein α
- constitutive androstane receptor
- intrinsic clearance
- peak concentration
- cytochrome P450
- European Medicines Agency
- Food and Drug Administration
- glucocorticoid receptor
- human liver microsomes
- 3-hydroxy-3-methylglutaryl-coenzyme A
- hepatic nuclear factor
- inhibitor concentration
- reversible inhibition constant
- inhibitor concentration supporting half of the maximal rate of enzyme inactivation
- maximal rate of inactivation
- Michaelis-Menten constant
- 2-[[5,7-dipropyl-3-(trifluoromethyl)-1,2-benzisoxazol-6-yl]oxy]-2-methylpropanoic acid
- messenger ribonucleic acid
- organic anion transporter
- organic anion-transporting polypeptide
- peroxisome proliferator activated receptor
- pregnane X receptor
- retinoic acid-related orphan receptors
- sorting intolerant from tolerant
- single nucleotide variation
- elimination half-life
- time to peak concentration
- vitamin D receptor
- Copyright © 2015 by The American Society for Pharmacology and Experimental Therapeutics