Supplemental material to this article can be found at: http://dmd.aspetjournals.org/content/suppl/2015/12/18/dmd.115.066985.DC1

1521-009X/44/3/466–475$25.00 http://dx.doi.org/10.1124/dmd.115.066985 DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 44:466–475, March 2016 Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics Contribution of Major Metabolites toward Complex Drug-Drug Interactions of Deleobuvir: In Vitro Predictions and In Vivo Outcomes s

Rucha S. Sane, Diane Ramsden, John P. Sabo, Curtis Cooper, Lois Rowland, Naitee Ting, Andrea Whitcher-Johnstone, and Donald J. Tweedie

Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Connecticut (R.S.S, D.R., J.P.S., L.R., N.T., A.W.J, D.J.T) and Department of Medicine, University of Ottawa, Ottawa, Ontario, Canada (C.C.)

Received September 4, 2015; accepted December 17, 2015

ABSTRACT The drug-drug interaction (DDI) potential of deleobuvir, an hepatitis note was more potent CYP2C8 inactivation by deleobuvir-AG than C virus (HCV) polymerase inhibitor, and its two major metabolites, deleobuvir and P450 induction by CD 6168 but not by deleobuvir. Downloaded from CD 6168 (formed via reduction by gut bacteria) and deleobuvir-acyl The predicted net AUC ratios for probe substrates were 2.92 glucuronide (AG), was assessed in vitro. Area-under-the-curve (CYP1A2), 0.45 (CYP2C9), and 0.97 (CYP3A4) compared with (AUC) ratios (AUCi/AUC) were predicted using a static model and clinically observed ratios of 1.64 (CYP1A2), 0.86 (CYP2C9), and compared with actual AUC ratios for probe substrates in a P450 1.23 (CYP3A4). Predictions of DDI using deleobuvir alone would cocktail of caffeine (CYP1A2), tolbutamide (CYP2C9), and midazo- have significantly over-predicted the DDI potential for CYP3A4 lam (CYP3A4), administered before and after 8 days of deleobuvir inhibition (AUC ratio of 6.15). Including metabolite data brought dmd.aspetjournals.org administration to HCV-infected patients. In vitro studies assessed the predicted net effect close to the observed DDI. However, inhibition,inactivationandinductionofP450s.Inductionwas the static model over-predicted the induction of CYP2C9 and assessed in a short-incubation (10 hours) hepatocyte assay, inhibition/inactivation of CYP1A2. This multiple-perpetrator DDI validated using positive controls, to circumvent cytotoxicity seen scenario highlights the application of the static model for with deleobuvir and its metabolites. Overall, P450 isoforms were predicting complex DDI for CYP3A4 and exemplifies the impor- differentially affected by deleobuvir and its two metabolites. Of tance of including key metabolites in an overall DDI assessment. at ASPET Journals on September 27, 2021

Introduction Agency (EMA, 2012), contribution of major metabolites should be Deleobuvir (BI 207127) is an inhibitor of virus (HCV) considered in assessing overall drug-drug interaction (DDI) liability . NS5B RNA polymerase and, in combination with and of the parent compound if a metabolite concentration is 25% of the . , achieved high, sustained viral-load reduction in treatment- parent and/or if a metabolite concentration is 10% of total drug- naïve and treatment-experienced patients with chronic genotype 1b related material in circulation. Higher cut-off values have been (Zeuzem et al., 2013). In a disposition study in which 14C- proposed in the literature using the logic that polar metabolites deleobuvir was administered to healthy male human volunteers, two lacking structural alerts are typically less potent P450 inhibitors and major metabolites of deleobuvir were identified, an alkene reduction inactivators than the parent drug (Callegari et al., 2013; Yu and product (CD 6168) formed by gut bacteria and deleobuvir-acyl Tweedie, 2013; Yu et al., 2015). These approaches have largely glucuronide (deleobuvir-AG) (Fig. 1) (Chen et al., 2015; McCabe focused on DDI as a result of inhibition and inactivation. et al., 2015). Following a single oral dose of 800 mg deleobuvir to Recent evaluations to estimate clinical drug-drug interaction risk involving a combination of inhibition, inactivation, and induction, healthy volunteers, CD 6168 and deleobuvir-AG exposure (AUC0–‘) in the systemic circulation was 27 and 43%, respectively, of the using dynamic physiologically based pharmacokinetic models, as parent deleobuvir. Both CD 6168 and deleobuvir-AG are pharma- well as simpler static models, have shown promising results for cologically active against HCV replication, albeit 10- and 3-fold less CYP3A4 (Fahmi et al., 2008a,b; Zhao et al., 2011; Einolf et al., potent, respectively, than deleobuvir (data on file at Boehringer 2014). In this study, in vitro DDI assessment was performed for Ingelheim). parent drug deleobuvir and both of its major metabolites. In vitro, According to the drug-drug interaction guidance documents from the US deleobuvir, CD 6168, and/or deleobuvir-AG affected activities of FoodandDrugAdministration (FDA, 2012) and the European Medicines several P450 isoforms by competitive inhibition, inactivation, and/or induction. Dynamic models incorporate temporal changes in drug and enzyme levels in a sophisticated manner, thereby minimizing dx.doi.org/10.1124/dmd.115.066985. overprediction of interaction. Static models usually incorporate s This article has supplemental material available at dmd.aspetjournals.org. only the maximal plasma concentrations or the portal vein inlet

ABBREVIATIONS: AUC, area under the concentration time curve; Cmax, maximum plasma concentration; CAR, constitutive androstane receptor; DDI, drug-drug interaction; deleobuvir-AG, deleobuvir-acyl glucuronide; GI, gastrointestinal; HCV, ; HLM, human liver microsomes; LC-MS/MS, high-performance liquid chromatography–tandem mass spectrometry; LDH, lactate dehydrogenase; MTT, 3-(4,5-dimethylthiazol-2-yl)- 2,5-diphenyl-tetrazolium bromide; OATP, organic anion-transporting polypeptide; P450, cytochromes P450; PEG, polyethylene glycol; PXR, pregnane X receptor; QD, every day (or daily); r, recombinant.

466 P450 DDI of Deleobuvir and Its Metabolites 467

for managing comedication use in larger scale clinical studies with deleobuvir are also discussed.

Materials and Methods 13 13 Materials. Deleobuvir, CD 6168, deleobuvir-AG, C6-deleobuvir, C6-CD 6168 were synthesized at Pharmaceuticals Inc. (Ridgefield, CT). Cryoplateable human hepatocytes were purchased from Invitrogen/ ThermoFisher Scientific (Grand Island, NY); donor information can be found in (Supplemental Table 1). Pooled human liver microsomes (HLM; (Supple- mental Table 2) and recombinant human cytochromes P450 (rP450; produced in baculovirus-infected insect cells) and control insect cell microsomes, rat type 1 collagen, L-glutamine, penicillin/streptomycin, ITS+ premix, and fetal bovine serum were purchased from Corning (Woburn, MA). NADPH, omeprazole (OME), phenobarbital (PB), rifampicin (RIF), trypan blue, ECM, William’s E media, LDH, and MTT cytotoxicity assay kits and dexamethasone were purchased from Sigma-Aldrich (St. Louis, MO). The mRNA catcher kits, reverse transcription polymerase chain reaction (RT-PCR),

and TaqMan primer probes were purchased from Invitrogen/ThermoFisher Downloaded from Scientific. All other reagents and solvents were of analytical grade or higher purity and were obtained from commercial suppliers. Human liver microsomes and rP450s were stored at –80C until used. Fig. 1. Structures of deleobuvir, CD 6168, and deleobuvir-AG. Cytochrome P450 Inhibition. Each isoform-specific cytochrome P450 (P450) probe substrate (CYP1A2, phenacetin; CYP2B6, bupropion; CYP2C8, amodia- quine; CYP2C9, diclofenac; CYP2C19, (S)-mephenytoin; CYP2D6, dextromethor- dmd.aspetjournals.org concentrations, which is not a realistic assumption throughout the phan; CYP3A4, midazolam) was incubated in the presence of various concentrations of deleobuvir, CD 6168, or deleobuvir-AG with pooled HLM time course of co-administration of interacting drugs. In the case of in potassium phosphate buffer (50 mM, pH 7.4) and respective isoform-specific deleobuvir, the complexity of incorporating three perpetrator com- metabolites (acetaminophen, hydroxybupropion, N-desmethylamodiaquine, 49- pounds, with mixed effects, made attempts at dynamic modeling very hydroxydiclofenac, 49 -hydroxymephenytoin, dextrorphan, 19-hydroxymidazolam) challenging, and consequently a simpler static model, modified to were monitored. HLM, inhibitor, and substrate were preincubated at 37Cfor incorporate multiple perpetrators, was used to evaluate the accuracy of 5 minutes. Incubations with isoform-selective P450 inhibitors (CYP1A2, prediction of the effect of deleobuvir on CYP1A2, CYP2C9, and a-naphthoflavone; CYP2B6, ticlopidine; CYP2C8, montelukast; CYP2C9, CYP3A4activityinvivo,whichthenmadefeasibleapredictionofthe sulfaphenazole; CYP2C19, S-(+)-N-3-benzylnirvanol; CYP2D6, quinidine; at ASPET Journals on September 27, 2021 effects on other P450 isoforms. CYP3A4, ketoconazole) were included as positive controls and showed An in vivo study was conducted in HCV-infected patients to assess the expected results (data not shown). All experiments were performed under net effect of deleobuvir (and generated metabolites) on the pharmacoki- linear conditions with respect to time and protein concentration and were conducted at substrate concentrations equivalent to their K values determined netics of caffeine, tolbutamide, and midazolam, used in combination as an m in HLM [phenacetin (40 mM), bupropion (90 mM), amodiaquine (1.5 mM), in vivo probe substrate cocktail for CYP1A2, CYP2C9, and CYP3A4 diclofenac (4 mM), (S)-mephenytoin (30 mM), dextromethorphan (7 mM), activities, respectively (Fuhr et al., 2007; Sabo et al., 2015). The patients in midazolam (3 mM), and testosterone (60 mM)]. Reactions were initiated by the this study also received polyethylene glycol (PEG)– and addition of NADPH (2 mM). A 50-ml aliquot of the incubation mixture was ribavirin as part of the standard of care. On the basis of information in removed at desired time points, and reactions were quenched with 40% the Copegus drug label, ribavirin would not have been expected to affect acetonitrile containing internal standard (nevirapine or stable labeled probe P450 enzymes. However, there had been reports of PEG-interferon a-2b metabolites) and 0.1% acetic acid. For CYP2C8, the reactions were quenched inhibiting CYP1A2, albeit very mildly. These studies had assessed the with 40% methanol with 0.3% formic acid. All quenched samples were filtered effect of PEG-interferon a-2bafterapproximately1monthoftreatment and analyzed for appropriate P450 reaction-specific probe metabolites by high- – (Gupta et al., 2011; Brennan et al., 2013), which is much longer than the performance liquid chromatography tandem mass spectrometry (LC-MS/MS). Concentrations that achieve a 50% inhibitory effect (IC50) were calculated using duration reported for our study. Further, in a parallel arm to our study, Â a Graph Pad Prism and Ki values were calculated as IC50 0.5, assuming PEG-interferon -2b and ribavirin were administered in an identical competitive inhibition. regimen in combination with faldaprevir and had no effect on CYP1A2 P450 Inactivation. Pooled HLM (0.4 mg/ml) and recombinant CYP3A4 and CYP2C9 activities (Sabo et al., 2015). As such, PEG-interferon (50 pmol/ml, ;0.6 mg/ml) were incubated with various concentrations of deleobuvir, a-2b was not expected to have an impact on CYP1A2, and the effects CD 6168, or deleobuvir-AG in potassium phosphate buffer (50 mM, pH 7.4) at 37C seen on the probe substrates are considered attributable to treatment for 5 minutes. Reactions were initiated by the addition of cofactor NADPH (2 mM). with deleobuvir and its metabolites. The in vitro DDI predictions for Control samples were included in which deleobuvir, CD 6168, or deleobuvir-AG CYP1A2, CYP2C9, and CYP3A4 for deleobuvir and its generated were incubated at the highest test concentration employed in the respective assay, in metabolites were compared with the clinical DDI results for the probe the absence of NADPH. At specific time points up to 35 minutes, an aliquot of this substrates. primary incubation was removed and diluted 20-fold by adding to a secondary incubation in which the enzyme activity remaining was measured using isoform- Some of the complexities of evaluating multiple perpetrator com- specific probe substrates outlined above. Concentration of the probe substrates used pounds, together with differential effects of parent drug versus were approximately 5- to10-fold their respective Km values. After incubation with metabolites, are highlighted. Additionally, an abbreviated incubation of NADPH (2 mM) at 37C for 6 minutes, the reactions were quenched with 40% 10 hours was validated in the hepatocyte induction assay to circumvent acetonitrile containing internal standard and 0.1% acetic acid. Samples were filtered toxicity observed with deleobuvir and its metabolites; this abbreviated and analyzed by LC-MS/MS to monitor the appropriate P450 reaction–specific probe incubation can be applied to other compounds. Data-driven strategies metabolites described above. 468 Sane et al.

Rate constants for loss of P450 activity, normalized to the corresponding controls significant difference between drug- and solvent-treated wells using a two- without the test compound, were used in eq. (1) to obtain inactivation parameters: tailed unpaired student t test in Microsoft Excel. Where concentration-

dependent induction was observed, the data were fit to determine EC50 values using eq. (2): where:

kobs is the observed rate constant for inactivation, kinact is the maximal inactivation rate constant, where X is the logarithm of concentration. [I] is the concentration of inactivator in the primary incubation, and Protein Binding Estimates in Plasma, HLM, and Induction Culture KI is the concentration of inactivator at which the rate of inactivation is Media. Teflon dialysis cells (Spectrum, Rancho Dominguez, CA) and dialysis half maximal. membranes (Spectra/Por, Spectrum) with 12,000–14,000 molecular weight cut- 14 Samples were analyzed for metabolite production on a 4000 QTRAP (AB off were used for equilibrium dialysis to determine binding of [ C]deleobuvir, [14C]CD 6168, and [14C]deleobuvir-AG in pooled human plasma, HLM Sciex, Thornhill, Ontario, Canada) attached to either a CTC PAL autosampler (0.05–0.5 mg/ml), and induction culture medium (bovine serum albumin, (Leap Technologies, Carrboro, NC) or a Waters Acquity UPLC system (Milford, 1.25 mg/ml). Concentration ranges of [14C]deleobuvir, [14C]CD 6168, and MA). The aqueous mobile phase (A) and organic mobile phase (B) consisted of 14 – m 95:5 (v/v) water/acetonitrile and 95:5 (v/v) acetonitrile/water, respectively. Both [ C]deleobuvir-AG tested were 0.05 100 M for microsomal binding. For pooled human plasma binding [14C]deleobuvir was tested at 0.15–59 mM, mobile phases contained 0.1% acetic acid. For the CYP2C8 assays, the mobile 14 14 [ C]CD 6168 was tested at 0.52–44 mM, and [ C]deleobuvir-AG was tested Downloaded from phase (A) consisted of 5 mM ammonium formate, 0.3% (v/v) formic acid in – m 14 water, and mobile phase (B) consisted of 0.3% (v/v) formic acid in methanol. at 0.5 70 M. Binding to induction medium was evaluated for [ C]CD 6168 m Samples were eluted through Waters YMC-Pack C4 (50 Â 3 mm, 5-mm, at concentrations between 0.1 and 30 M. For the protein binding as- sessment, five individual dialysis cells were prepared and rotated at 20 rpm CYP2C8) or a Phenomenex Synergi Max RP (150 Â 2mm, 4-mm) column, using using a Spectrum dialysis cell rotator in a water bath maintained at 37C validated probe substrate analysis methods. The multiple reaction monitoring 13 for 4 hours. At the end of the incubation period, the contents of each side transitions used were 152.0→ 110.0 (acetaminophen); 155.0→ 110.0 ([ C2, 15 were transferred to scintillation vials and processed for liquid scintillation N]-acetaminophen); 256.1→ 139.1 (hydroxybupropion); 262.1→ 139.1 counting. dmd.aspetjournals.org (hydroxybupropion-d ); 328.0→ 283.0 (N-desethylamodiaquine); 333.0→ 6 DDI Prediction. Prediction of drug interactions for all enzymes was 283.0 (N-desethylamodiaquine-d5); 312.0→230.0 (49-hydroxydiclofenac); 318.0→236.0 (49 hydroxydiclofenac-[13C ]); 258.0→157.0 (dextrorphan); conducted as described in the regulatory guidances by EMA (2012) and FDA 6 (2012). For determination of DDI potential for CYP1A2, CYP2B6, CYP2C8, 261.0→157.0 (dextrorphan-d3); 342.0→324.0 (19-hydroxymidazolam); 13 CYP2C9, and CYP3A4, the static model [eqs. (3) and (4)] was used for 347.0→327.0 (19-hydroxymidazolam-[ C ]); 305.0→269.0 (6b-hydroxy- 3 predicting area-under-the-curve (AUC) fold change of a sensitive substrate testosterone); 312.0→276.0 (6b-hydroxytestosterone-d7); 267.2→226.1 (neviripine); in positive ion mode. since deleobuvir and/or its metabolites were inhibitors, inactivators, and/or inducers of these P450 isoforms. Substrate and physiologic parameters and P450 Induction and Cytotoxicity. To assess induction of CYP1A2, explanation of the terms used in the calculations are displayed in (Supple- at ASPET Journals on September 27, 2021 CYP2B6, and CYP3A4 mRNA, three separate lots of cryoplateable human hepatocytes, precharacterized for prototypical induction response (Hu1419, mental Tables 4 and (5). Hu1424, and Hu8123), were used. Cytotoxicity was measured by lactate dehydrogenase (LDH) leakage and the MTT cell-proliferation assay (mea- AUCR = AUCi/AUC = suring the reduction of a tetrazolium component to an insoluble formazan product by mitochondria of viable cells), performed in the same donors at 8, Here, A , B , C refer to inhibition, inactivation and induction effects in the liver, 10, and 12 hours. Cytotoxicity for deleobuvir and CD 6168 was assessed at 24 h h h respectively, and the A , B , and C refer to the inhibition, inactivation, and and 48 hours during a separate study. The cells were plated per vendor g g g induction effects in the gastrointestinal (GI) track, respectively. The terms fm and recommendations and incubated at 37C with 5% CO2 for 4–6 hours to enable attachment of the cells. After adequate attachment, cells were overlaid with Fg refer to fraction metabolized by the pathway under consideration and fraction escaping gut metabolism, respectively. Details regarding each parameter in the extracellular matrix (ECM). Twenty four hours after preparation of the equation can be found in the EMA and FDA (draft) guidances and relevant input sandwich cultures, the medium was aspirated and replenished. Between 46 and 48 hours after recovery and plating, the cells were treated with either parameters for deleobuvir DDI predictions are provided in (Supplemental solvent [0.5% (v/v) equivalent parts acetonitrile and methanol], deleobuvir, Table 5). This equation was modified to incorporate multiple perpetrators in an additive manner as depicted in eq. (4) (Lutz et al., 2013; Rowland and Yeo CD 6168, deleobuvir-AG (0.1, 0.3, 1, 3, and 10 mM for all three), or et al., 2010). a prototypical P450 inducer [omeprazole (0, 0.3, 1, 3, 10, 30, 100, and 300 mM), phenobarbital (0, 3, 10, 30, 100, 300, 1000, and 3000 mM), or rifampicin (0, 0.3, 1, 3, 10, 30, 100, and 300 mM)] for 10 hours. A shorter incubation time (10 hours) was used to avoid cytotoxicity observed upon incubation of TABLE 1 hepatocytes with deleobuvir and its metabolites (at .1 mM) for the typical 48– a Ki (mM) values for competitive inhibition of P450 isoforms by deleobuvir, CD 72 hours employed for induction studies. For the precharacterization of the 6168, and deleobuvir-AG 10-hour time point and donor selection, cells from four donors were treated with 25 mM rifampicin or solvent for 0, 4, 6, 8, 12, 24, and 48 hours. Cells Deleobuvir CD 6168 Deleobuvir-AG were replenished with fresh media containing rifampicin again at 24 hours. P450 Isoform Total Freeb Total Free Total Free After the incubation, cell lysates were treated with RNA-later and stored at –20C until isolation of RNA. RNA samples were isolated from these CYP1A2 NA NA NA NA 12.5 6.13 frozen hepatocytes using an mRNA Catcher PLUS kit (cat. no. K1570-03; CYP2B6 17.5 11.3 NA NA 20.4 9.99 – CYP2C8 0.13 0.087 0.27 0.21 0.022 0.0108 ThermoFisher Scientific, Grand Island, NY) and stored at 80 C. The mRNA CYP2C9 4.7 3.0 18 13.5 14.4 7.06 expression for specific gene targets was determined by TaqMan Real-Time CYP2C19 NA NA NA NA 26 12.7 RT-PCR using a 7900 Real Time PCR System (Applied Biosystems/ThermoFisher CYP2D6 NA NA NA NA 50 24.5 Scientific) and available primer probe sets (Supplemental Table 3). For each CYP3A4 7.4 4.4 23 17 12 5.88 study, cells treated from three separate wells were analyzed in duplicate NA, no inhibition up to the highest concentration tested (50 or 75 mM) and data for each gene of interest was normalized using housekeeping genes, a Calculated as IC50 Â 0.5, assuming competitive inhibition. GAPDH, or b-actin. Statistical analysis was performed to determine any bAdjusted for protein binding in HLM. P450 DDI of Deleobuvir and Its Metabolites 469

compounds. Ex vivo, 40% loss of deleobuvir-AG was observed from spiked plasma; however, deleobuvir-AG was highly stable under acidified conditions. Hence, plasma samples were mixed with citric acid at collection to stabilize the acyl glucuronide metabolites from spontaneous chemical degradation. Deleobuvir and its metabolites were measured using validated bioanalytical assays and authentic standards as described previously (Chen et al., 2015). After day 9, the dosing continued with deleobuvir and faldaprevir in combination with pegylated-interferon and ribavirin until the end of treatment (24 weeks) (Sabo et al., 2015). The pharmacokinetic and pharmacodynamics endpoints of the complete study are beyond the scope of this discussion. The statistical model used for the analysis of the in vivo data was an analysis of variance (ANOVA) model on the logarithmic scale. This model included effects accounting for the following sources of variation: “subject” and “treatment.” The effect “subject” was considered as random, whereas the other effect was considered as fixed. Clinical Drug-Drug Interaction Study. The effect of deleobuvir on CYP1A2, CYP2C9, and CYP3A4 was evaluated in the clinic in the initial phase of a larger open label Phase II study. Caffeine, tolbutamide, and Results midazolam were used as probe substrates for CYP1A2, CYP2C9, and CYP3A4, P450 Inhibition and Inactivation. Of the seven isoforms evaluated, Downloaded from respectively. This multicenter study (seven sites in Canada) was conducted in deleobuvir, CD 6168, and/or deleobuvir-AG inhibited CYP2B6, HCV-infected patients. The study was conducted in accordance with the CYP2C8, CYP2C9, and CYP3A4 and the respective Ki values for International Conference on Harmonization guideline for Good Clinical Practice inhibition are presented in Table 1. CYP2C8 was inhibited most and the principles of the Declaration of Helsinki, and was approved by the local potently by deleobuvir-AG with a K value of 0.022 mM. Deleobuvir ethics committee and is registered at ClinicalTrials.gov under registration i number NCT01525628. and CD 6168 also inhibited CYP2C8 with Ki values of 0.13 and 0.27 m dmd.aspetjournals.org Patients (11 males and 8 females) were administered single oral doses of M, respectively. Deleobuvir-AG also inhibited CYP1A2, CYP2C19, caffeine (200 mg), tolbutamide (500 mg), and midazolam (2 mg) prior to and CYP2D6 (Table 1), whereas no inhibition of CYP1A2, CYP2C19, initiation of HCV therapy. Blood samples for pharmacokinetic analysis of drug and CYP2D6 was observed for deleobuvir and CD 6168 up to 50 mM. concentrations were collected over 24 hours. Therapy was then initiated with Out of the seven isoforms evaluated, deleobuvir mildly inactivated deleobuvir (600 mg TID; 6-hour/6-hour/12-hour dosing schedule) in combi- CYP1A2 and CYP3A4 and potently inactivated CYP2C8 (Table 2). nation with weight-based QD ribavirin (Copegus) and weekly pegylated- CD 6168 also inactivated CYP2C8 more potently than CYP3A4. a interferon -2a (Pegasys). The probe substrates were administered after 8 days Deleobuvir-AG inactivated CYP2C8 more potently than both dele- of dosing with deleobuvir, ribavirin, and pegylated-interferon a-2a, and blood obuvir and CD 6168. Figure 2, A–C shows relative loss of CYP2C8 at ASPET Journals on September 27, 2021 samples were again collected over 24 hours. Patients that required any activity by deleobuvir, CD 6168, and deleobuvir-AG at roughly that are moderate or potent inhibitors or inducers of CYP1A2, CYP2C9, or CYP3A were excluded from the study from up to 5 days prior to equimolar concentrations, and Fig. 2D shows the fitting of kobs and day 1 through to day 19 of the treatment phase. Plasma concentrations of kinact for CYP2C8 inactivation by these three perpetrators. Deleobuvir caffeine, tolbutamide, and midazolam and metabolites of interest were and deleobuvir-AG were also mild inactivators of CYP1A2 (Table 2). measured by LC-MS/MS (Tandem Laboratories, Salt Lake City, UT). Samples Time Course of mRNA Induction by Rifampicin. The time containing analyte and dueterated internal standard were extracted and course for maximal CYP3A4 mRNA induction was determined to analyzed in an API 4000 or API 5000 mass spectrometer (AB Sciex, validate the approach of using a shortened time course for generating Framingham, MA). Quantification was performed using linear (midazolam, induction parameters (EC50 and Emax) from human hepatocytes. At the 1-OH midazolam) or quadratic (caffeine, tolbutamide, 4-OH tolbutamide) prototypical incubation times of 48 or 72 hours, the E determination  max weighted regression analysis (1/ 2) of peak area ratios of analytes and was limited by cytotoxicity of deleobuvir and CD 6168 (data not internal standards. Quantification ranges (ng/ml) were: caffeine, 10.0–10,000; shown). The magnitude of mRNA induction for three lots of human midazolam and 1-OH midazolam, 0.2–10.0; tolbutamide and 4-OH tolbutamide, hepatocytes treated with rifampicin (25 mM), a prototypical CYP3A4 5.0–5,000. Pharmacokinetic parameters [maximum plasma concentration (Cmax), inducer, after 2-, 4-, 8-, 12-, 24-, and 48-hour incubation is displayed in AUC0–tz,orAUC0–‘] were calculated for the probe substrates using WinNonLin version 5.2. Fig. 3. In general, the donors reached maximal level of induction Plasma concentrations and pharmacokinetics of deleobuvir, deleobuvir-AG, between 8 and 12 hours. The magnitude of CYP3A4 mRNA increases and CD 6168 were monitored to ensure exposure to the study at these time points was consistent with the magnitude observed at

TABLE 2 21 Inactivation parameters KI (mM) and kinact (min ) for time-dependent P450 inactivation by deleobuvir, CD 6168, and deleobuvir-AG

Deleobuvir CD 6168 Deleobuvir-AG

P450 Isoform KI KI KI kinact kinact kinact Total Freea Total Free Total Free

mM mMmin–1 mM mMmin–1 mM mMmin–1 CYP1A2 28.1 8.99 0.0256 NA NA NA 54.6 8.74 0.044 CYP2B6 NA NA NA NA NA NA NA NA NA CYP2C8 0.107 0.066 0.0268 0.206 0.045 0.022 0.0521 0.0083 0.0521 CYP3A4 103 13.4 0.0485 201 46.6 0.0875 NA NA NA

NA, no inactivation observed. aAdjusted for protein binding in HLM. 470 Sane et al.

from 48- to 72-hour incubations (calculations not shown) (Fahmi et al., 2008a,b; Shou et al., 2008). Cytotoxicity Results. No notable changes in LDH or MTT assay (,15% of control values) were seen at 8, 10, and 12 hours at concentrations up to 100 mM of deleobuvir, CD 6168, or deleobuvir- AG. At later time points (24 and 48 hours) morphologic changes, increased LDH leakage (.50% of control values), and decreased expression of housekeeping gene b-actin were noted for deleobuvir and its metabolites at concentrations above 3 mM. Specifically, at 24 hours, 100 mMCD6168 caused an 80% reduction in mitochondrial activity and at 48 hours a .30% reduction was observed at concentrations $3 mM. mRNA Induction by Deleobuvir, CD 6168, and Deleobuvir-AG. The induction potential of various concentrations of deleobuvir, CD 6168, and deleobuvir-AG toward CYP1A2, CYP2B6, CYP2C8, CYP2C9, and CYP3A4 mRNA was investigated after incubation for 10 hours. Deleobuvir did not cause an increase in mRNA for any of the P450 isoforms tested, with the possible exception of CYP3A4, although Downloaded from increases were mild, inconsistent, not concentration-dependent, and not statistically significant and so are not considered clinically relevant. CD 6168 treatment resulted in concentration-dependent increases in CYP2B6, CYP2C8, CYP2C9, and CYP3A4 mRNA in one, two, or all three of the hepatocyte donors evaluated, as shown in Fig. 5, A–D. The increase was statistically significant (a = 0.05) at concentrations $1 mM for CYP2B6 and CYP2C8 in hepatocyte donors who showed dmd.aspetjournals.org induction. For CYP2C9, statistical significance was achieved at concentrations $0.3 mM for donor 1424 and at $ 3mM for donor 8123. For CYP3A4, all donors exhibited statistically significant induction at concentrations $0.3 mM. There was no induction observed for CYP1A2 by CD 6168. Deleobuvir-AG resulted in a statistically significant but mild increase in CYP1A2 mRNA in hepatocytes from

one out of three donors. Maximal induction response in this donor at ASPET Journals on September 27, 2021 compared with the prototypical inducer (omeprazole, 50 mM) was 30%. Table 3 summarizes Emax and EC50 parameters for deleobuvir metabolites as well as the prototypical inducers in hepatocyte donors where data fitting was possible. Induction parameters were derived using the unbound concentrations of the compounds in induction culture media. Protein Binding Determinations in HLM, Human Intestinal Microsomes, Hepatocyte Medium and Human Plasma. Deleobuvir and its major metabolites demonstrated significant plasma protein binding that was not saturable up to the concentrations evaluated. Average human plasma protein binding was 99.3% for deleobuvir and CD 6168 and 99.6% for deleobuvir-AG. Deleobuvir and its metabolites exhibited similar binding in HLM, which ranged from 25 to 36% at 0.05 mg/ml to 67 to 78% at 0.5 mg/ml of microsomal protein. At the concentrations tested, CD 6168 was ;85% bound in the induction assay medium. Bound concentration of deleobuvir-AG in the induction

Fig. 2. Time- and concentration-dependent inhibition of CYP2C8 by deleobuvir (A), CD 6168 (B), and deleobuvir-AG (C); kinetics of CYP2C8 inactivation by deleobuvir, CD 6168, and deleobuvir-AG at various concentrations (D) (one representative experiment from n = 2).

48 hours. To further validate this approach, induction parameters and EC50 and Emax values were determined for prototypical inducers [omeprazole (0, 0.3, 1, 3, 10, 30, 100, and 300 mM), phenobarbital (0, 3, 10, 30, 100, 300, 1000, and 3000 mM), and rifampicin (0, 0.3, 1, 3, 10, 30, 100, and 300 mM)] after 10-hour incubation (Fig. 4), and DDI for a CYP3A4 substrate was predicted by the net effect (static) model. The predicted AUCi/ AUC ratio was similar to that reported using induction parameters Fig. 3. Time course of mRNA induction after rifampicin (25 mM) treatment. P450 DDI of Deleobuvir and Its Metabolites 471

Fig. 4. Concentration-dependent increase in mRNA by prototypical P450 inducers after 10-hour incubation (squares, donor Hu8123; diamonds, donor Hu1424; circles, donor 1419). assay medium was calculated to be ;89% on the basis of the plasma Following the 8-day DDI assessment, faldaprevir was added to the protein binding data and interpolation to albumin content in the induction treatment, with treatment continuing over 24 weeks. Two patients

assay medium. This approach was only used for deleobuvir-AG after interrupted dosing prematurely owing to adverse events (febrile Downloaded from evaluation with CD 6168 and confirmation that such extrapolated neutropenic dermatosis at dosing day 9; grade 2 hyperbilirubinemia protein binding values were close to the experimentally determined and abdominal discomfort at dosing day 24). Both participants fully values. Owing to the low magnitude of induction observed with recovered. Nausea (36.8–37.5%), headache (15.8–18.8%), fatigue deleobuvir-AG, this was considered to be an acceptable approximation (12.5–15.8%), and photosensitivity reaction (10.5–31.1%) were the of protein binding for in vivo DDI prediction. Inhibition, inactivation, most commonly observed adverse events over the 24-week duration

and induction parameters were corrected for protein binding using the of dosing. As recognized in previous studies, there is considerable dmd.aspetjournals.org values for the appropriate matrix. overlap in side effect profile between faldaprevir and deleobuvir Safety Profile of the Clinical Study. Deleobuvir plus ribavirin and (SOUND C-2; Zeuzem et al., 2013). Furthermore, it is challenging to two doses of pegylated-interferon a-2a within the first 8 days of dosing disentangle the contribution of ribavirin and early pegylated in- were relatively well tolerated with no deaths or serious adverse events. terferon exposure to the observed side effect profile. at ASPET Journals on September 27, 2021

Fig. 5. Concentration-dependent increase in mRNA by CD 6168 after 10-hour incubation. 472 Sane et al.

TABLE 3

Median induction parameters EC50 (mM) and Emax from mRNA levels after treatment with CD 6168 in three human hepatocyte donors

CD 6168 or Deleobuvir-AGa Prototypical Inducerc

P450 Isoform EC50 (mM) Emax Fold Emax/EC50 EC50 (mM) Emax Fold Emax/EC50 Total Freeb CYP1A2a 6.2 0.71 6.7 9.4 11.9 20.4 1.7 CYP2B6 1.1 0.18 3.8 21 595 10.2 0.017 CYP2C8 2.7 0.43 5.9 14 NC 5.1 NC CYP2C9 0.25 0.040 2.1 53 NC 2.5 NC CYP3A4 0.39 0.058 12.4 214 0.813 26.8 33

NC, not calculable, the dynamic range was insufficient to generate non-ambiguous fitting parameters therefore the Emax reported is the maximal fold change observed over the rifampicin incubation concentrations. aCYP1A2 induction parameters are from deleobuvir-AG incubations; for other isoforms parameters are from CD 6168 incubations. bAdjusted for protein binding in induction media which contains bovine serum albumin in the insulin, transferrin and selenium ITS supplement. cPrototypical inducers were omeprazole (CYP1A2), phenobarbital (CYP2B6), and rifampicin (CYP2C8, CYP2C9, and CYP3A4). Downloaded from In Vivo DDI with Caffeine, Tolbutamide, and Midazolam and gastrointestinal tract by gut bacteria, it was regarded as administration Plasma Exposure of Deleobuvir, CD 6168, and Deleobuvir-AG. of a second parent drug (Chen et al., 2015; McCabe et al., 2015). The When caffeine was coadministered with deleobuvir on day 9, caffeine fraction absorbed for deleobuvir was considered to be 0.5, as an Cmax increased approximately 39% and AUC0–‘ increased approxi- absorption, distribution, metabolism, and excretion (ADME) study mately 64%, compared with day 1 (P450 probes alone) (Table 4). When showed ;50% of the radioactivity in feces was accounted for by CD

tolbutamide was coadministered with deleobuvir, tolbutamide Cmax and 6168 and its metabolites (Chen et al., 2015). Since deleobuvir-AG is dmd.aspetjournals.org AUC0–‘ decreased approximately 8 and 14%, respectively, compared predominantly formed in the liver and any amount eliminated in the bile with day 1. There was a small increase in 4-OH-tolbutamide/ is hydrolyzed back to deleobuvir in the GI tract, the contribution of tolbutamide AUC ratio compared with day 1 (Table 4). When deleobuvir-AG to DDI in the GI tract was deemed to be negligible. The midazolam was coadministered with deleobuvir, midazolam Cmax and in vitro inhibition, inactivation, and induction parameters used were AUC0–‘ increased approximately 24 and 23%, respectively, compared corrected for protein binding. The resulting projections and correspond- with day 1. There was an ;10% decrease in 1-OH-midazolam/ ing in vivo results are displayed in Table 6. There was an overprediction midazolam AUC ratio compared with day 1 (Table 4). of AUC changes for both CYP1A2 and CYP2C9 when considering at ASPET Journals on September 27, 2021 Net Effect Modeling and Comparison with Clinical Outcome. deleobuvir, CD 6168, and deleobuvir-AG together. For CYP3A4, the The static model described in the EMA (2012) and FDA (2012) prediction using parameters for deleobuvir alone indicated a significant guidances on DDI was used to predict the clinical outcome of increase in AUC (6.15-fold), whereas including CD 6168 and deleobuvir treatment on sensitive substrates of CYP1A2, CYP2C9, deleobuvir-AG predicted no change in AUC (0.97-fold), which agree and CYP3A4 and the predictions were retrospectively compared with well with the actual clinical data (1.23-fold). the in vivo results from the clinical DDI study described above. The maximal plasma concentrations obtained in this study (Table 5) were used to calculate the inlet portal concentrations for deleobuvir and CD Discussion 6168, which were then used in the net effect prediction. Maximal Regulatory guidances from the FDA and EMA have proposed that plasma concentrations for deleobuvir-AG were directly used in the metabolites present at greater than 25% of the parent molecule or predictions. This distinction was made because the glucuronide greater than 10% of drug-related material (EMA) should be investigated in metabolite is formed in the hepatocytes and thus portal vein vitro for DDI potential (EMA, 2012; FDA, 2012). In the current clinical concentrations were not expected to be higher than maximal plasma study, in HCV-infected patients, following 8 days of BID dosing at 600 mg concentrations. Since CD 6168 is formed presystemically in the of deleobuvir, CD 6168, a reduction product formed by gut bacteria

TABLE 4 Adjusted geometric means (%gCV) for pharmacokinetic parameters and relative bioavailability of caffeine, tolbutamide, 4-OH tolbutamide, midazolam, and 4-OH midazolam in HCV-infected patients before (day 1) and after (day 9) treatment of 8 days with 600 mg TID deleobuvir

Substrate PK parameter Day 9 (with DLV) Day 1 (CYP Probes) gMean Ratio 90% CI for Ratio

N gMean N gMean [%] [%] Caffeine (CYP1A2 probe) Cmax [mg/ml] 17 7.43 19 5.34 139 122 159 AUC0–‘ [mg∙h per milliliter] 15 127 19 77.5 164 130 208 Tolbutamide (CYP2C9 probe) Cmax [mmol/l] 17 157 19 170 92.2 86.7 98.1 AUC0–‘ [mmol∙h per milliliter] 17 1920 19 2220 86.4 77.6 96.1 rAUC0–‘, M/P 4-OH 16 0.0317 (110) 17 0.0209 (24.5) tolbutamide/tolbutamide Midazolam (CYP3A4 probe) Cmax [nmol/l] 17 29.6 19 23.8 124 108 143 AUC0–‘ [nmol∙h per milliliter] 17 131 19 107 123 99.1 152 rAUC0–‘, M/P 1-OH 17 0.219 (39.4) 19 0.243 (52.8) Midazolam/midazolam P450 DDI of Deleobuvir and Its Metabolites 473

TABLE 5 impact on CYP1A2 activity in vivo was driven by inactivation by Pharmacokinetic parameters for deleobuvir, CD 6168, and deleobuvir-AG on day 9 deleobuvir, rather than inhibition by deleobuvir-AG (Table 6).

N = 17 patients. AUCt,ss and Cmax,ss were derived from concentration-time data for the first Alternative perspectives to regulatory guidances on the potential of dosing interval of the day, where t = 6 hours a metabolite to inhibit or inactivate P450s have been discussed Analyte PK Parameter gMean gCV [%] (Callegari et al., 2013; Yu and Tweedie, 2013; Yu et al., 2015). However, these do not address induction of P450s by metabolites. Deleobuvir AUCt,ss (nmol∙h per liter) 41100 93.7 Structurally similar metabolites have been shown to possess similar or Cmax,ss (nmol/l) 10900 85.3 lower potency of induction (Petzer et al., 2003; Medina-Diaz et al., Deleobuvir-AG 2009). One recent report highlighted carboxymefloquine as being

AUCt,ss (nmol∙h per liter) 24300 131 a pregnane X receptor (PXR) ligand, whereas the parent mefloquine Cmax,ss (nmol/l) 5620 119 was not (Piedade et al., 2015). With deleobuvir, reduction of an alkene Metabolite/Parent ratio 0.590 51.8 to form CD 6168 resulted in a potent inducer. In an hepatocyte CD 6168 induction assay, under standard experimental conditions, metabolites AUCt,ss (nmol∙h per liter) 13300 123 Cmax,ss (nmol/l) 3040 115 can be formed from parent drug, and as such the induction liability of Metabolite: Parent ratio 0.322 54.5 the metabolite is also being considered, with the obvious caveat of ss, steadystate. differential levels in vitro and in vivo. However, CD 6168 is formed

presystemically by gut bacteria and is not generated in the liver Downloaded from (McCabe et al., 2015) and would therefore not be represented by adding (McCabe et al., 2015), and an acyl glucuronide of deleobuvir (deleobuvir- deleobuvir alone in these induction studies. Deleobuvir-AG would be AG) (Chen et al., 2015) were circulating at peak exposures of 59 and 32% of formed during incubation of hepatocytes with deleobuvir itself, but the the parent deleobuvir, respectively. CD 6168 exposure is higher upon extent may be lower than that observed in vivo (Chen et al., 2015). multiple dosing despite a short half-life (Chen et al., 2015), probably owing to the unique site of formation in the GI tract. In vitro studies were conducted Therefore, it was rational to conduct induction studies in which to determine the inhibition, inactivation, and induction of P450s by deleobuvir, deleobuvir-AG, and CD 6168 were individually assessed. dmd.aspetjournals.org deleobuvir and its two major metabolites. A number of contrasting The generally accepted standard methodologies for evaluation of outcomes were found. First, several P450 isoforms were affected in induction in human hepatocytes employ incubation times of 48 or opposing directions, i.e., increases in enzyme activity by induction or 72 hours (Lin, 2006; Sinz et al., 2008; Chu et al., 2009). Deleobuvir and decreases in activity by inhibition or inactivation. Second, deleobuvir its metabolites were cytotoxic to sandwich cultured hepatocytes at m metabolites were more potent than deleobuvir, especially in the case of concentrations above 1 M upon incubation for 48 hours. Elevation of CYP2C8 inactivation (deleobuvir-AG) and P450 induction (CD 6168). mRNA through upregulation by nuclear receptors is the initial event leading to increases in protein and can be detected within 4–6 hours

Deleobuvir and CD 6168 (reduction of an alkene), which are structurally at ASPET Journals on September 27, 2021 very similar, inhibited and inactivated P450 enzymes with similar (Zhang et al., 2010). Validation of shorter incubation times was done potencies (Tables 1 and 2). Deleobuvir-AG inactivated CYP2C8 more using a two-step approach. First, the optimal time course was potently than parent, providing another example of an acyl glucuronide established by monitoring CYP3A4 mRNA increases in hepatocytes that is a potent inactivator of CYP2C8, as has been shown with by rifampicin at various time points after treatment (Fig. 3). Second, at gemfibrozil (Ogilvie et al., 2006) and clopidogrel (Tornio et al., 2014). the optimal time point selected (10 hours), induction parameters for The exact mechanism of the inactivation with deleobuvir-AG is not prototypical inducers (omeprazole for CYP1A2, phenobarbital for known at this time. A comparison of the kinact/KI ratios (shown in CYP2B6, and rifampicin for CYP3A4) were found to be comparable to parentheses) indicates that deleobuvir-AG (1.00 minute21 × mM21)is those reported in the literature (Fahmi et al., 2008a,b, 2009, 2010; Shou 2 2 a more potent inactivator than gemfibrozil (0.0105 minutes 1 × mM 1) et al., 2008; Zhang et al., 2010). Under these validated experimental and clopidogrel (0.0047 minutes21 × mM21) in vitro. However, additional conditions, deleobuvir and deleobuvir-AG did not induce P450 factors have to be considered to translate this to their potential for in vivo enzymes. Strikingly, CD 6168, which is structurally very similar to inactivation, including relevance of individual parameters of the ratio, fm deleobuvir, was a potent inducer of CYP3A4 (Table 3). CD 6168 also of the comedication and perpetrator concentration, and fm for the relevant induced other PXR/constitutive androstane receptor (CAR) target pathway. genes, namely CYP2B6, CYP2C8, and CYP2C9 (Fig. 5) with a rank Additionally, deleobuvir-AG also inhibited CYP1A2 and CYP2D6 order of induction similar to that observed with the PXR agonist more potently than deleobuvir and inhibited CYP2B6 with a similar rifampicin, suggesting that induction via CD 6168 may also be via PXR potency as parent (Table 1). The competitive inhibition of CYP1A2 by (and possibly CAR). Additional studies are required to fully define the deleobuvir-AG was somewhat surprising since CYP1A2 is considered differential interactions of deleobuvir and CD 6168 with upstream to have an active site suitable for planar molecules (Zhou et al., 2009). effectors such as PXR and CAR. The dearth of reports of induction by However, on the basis of the static model for DDI prediction, the overall metabolites, but not the parent drug, such as observed with CD 6168

TABLE 6 Predicted exposure changes using the net effect model and observed exposure changes for CYP1A2, CYP2C9, and CYP3A4 probe substrates

Predicted Change Predicted Change In Vivo AUCi/AUC in AUCi/AUC P450 Isoform Probe Substrate in AUCi/AUC of the Probe Substrate by Deleobuvir by Deleobuvir Alone at Day 9 (%gCV) and Its Metabolites CYP1A2 Caffeine 3.14 2.92 1.64 (38.1) CYP2C9 Tolbutamide 1.07 0.45 0.86 (18.1) CYP3A4 Midazolam 6.15 0.97 1.23 (37.2) 474 Sane et al. and carboxymefloquine, suggests the need for additional consideration and CYP2C8. Further consideration is being given to the possibility of metabolite DDI via induction, beyond those proposed by pharma- that deleobuvir and/or its metabolites could be substrates and inhibitors ceutical researchers (Yu et al., 2015) and regulators. of OATP1B1 and OATP1B3. If deleobuvir and its metabolites show Thus, the overall DDI prediction of deleobuvir required an evaluation a potential for an OATP-based interaction, an even higher interaction of the DDI potential of the metabolites. On the basis of in vitro data for is possible with dual substrates of CYP2C8 and OATPs, such as deleobuvir alone, the inhibition potential would have been markedly repaglinide and simvastatin, than that predicted by the static model, higher than the actual DDI observed in the clinic for CYP3A4 (Table 6). which only considers interaction via CYP2C8. Inactivation of CYP3A4 clearly dominated the net effect prediction for Owing to ethical considerations, the current study was conducted in deleobuvir since it is not an inducer. Induction of CYP3A4 by CD 6168 HCV-infected patients, which imposed inherent design limitations. As counteracted the inactivation effect of deleobuvir. As a result, the net such, the effect of a single dose of deleobuvir on P450 probe substrates effect at steady state was predicted to be a lack of an overall effect on was not evaluated. Consequently, the impact of only competitive CYP3A4. The situation was somewhat different for CYP2C9, as inhibition was not determined. This limited the validation of the static deleobuvir and its metabolites do not inactivate CYP2C9. While the model for predictability of inhibition, which would have been de- tolbutamide exposure tended to be lower after deleobuvir dosing, the termined following one or two doses, since deleobuvir has a relatively effect was not statistically significant. Thus, the model overpredicted short half-life of 2.84 hours (Chen et al., 2015). Induction and the induction potential of CYP2C9. For CYP1A2, there was a mild inactivation require steady state to be reached as well as optimization overprediction of the inhibitory effects, but the DDI prediction was not of biologic processes (de novo generation of protein for induction) for Downloaded from very different when deleobuvir was considered with or without a net steady-state effect, which was evaluated in this study after 8 days metabolites. This is not surprising as induction of CYP1A2 was not of dosing with deleobuvir. observed with CD 6168 and only a weak induction was observed with In conclusion, there are two interesting aspects of deleobuvir in deleobuvir-AG. Thus, the overall DDI prediction for CYP3A4, when which metabolites significantly contribute to the DDI liability of the all perpetrators were considered, was close to that observed in vivo, parent drug. First, deleobuvir-AG provides another example of an acyl whereas the model overpredicted the induction effect for CYP2C9 and glucuronide that inactivates CYP2C8 more potently than the parent. the inhibitory effect for CYP1A2. The net effect model, developed Second, CD 6168 represents a structurally similar presystemic dmd.aspetjournals.org primarily using in vitro and in vivo data for CYP3A4 (Fahmi et al., metabolite that induced CYP2C9 and CYP3A4, whereas the parent 2008b), may need further refinement for other P450 isoforms. deleobuvir exhibited no induction. Combining the DDI liabilities of Alternative approaches using systemic concentrations for inhibition deleobuvir, such as inactivation of CYP3A4 and inhibition of CYP2C9, and inactivation, or more realistic fm and Fg values, may improve the together with the DDI liabilities of the metabolites was essential for prediction for CYP2C9 and CYP1A2. While the use of a static model determining the net effect on these P450 isoforms. has not been frequently reported for highly complex DDI predictions involving multiple perpetrators, particularly with opposing effects on at ASPET Journals on September 27, 2021 the enzymes, the model appeared to be highly valuable for deleobuvir. Acknowledgments Additional examples will further increase the confidence in this The authors thank Monica Keith-Luzzi and Kathy Phelan for conducting approach. Deleobuvir and its metabolites exhibited a tendency to protein binding studies, Dr. Timothy Tracy for scientific advice, Karen Kimura accumulate in cultured hepatocytes by up to 20-fold (Chen et al., 2015), for operational aspects of the clinical study, investigating physicians and last but suggesting a potential role for organic anion-transporting polypeptide not the least, the patients who participated in this study. (OATP)–mediated hepatic uptake. While FDA DDI guidance only Authorship Contributions recommends the use of unbound concentrations in the portal vein for Participated in research design: Sane, Ramsden, Sabo. DDI predictions, EMA DDI guidance recommends that the liver Conducted experiments: Ramsden, Rowland, Whitcher, Cooper. accumulation be taken into consideration for the predictions. Assuming Performed data analysis: Ting, Ramsden. intracellular protein binding is the same as plasma protein binding when Wrote or contributed to the writing of the manuscript: Sane, Ramsden, Sabo, ;20-fold higher perpetrator concentrations were used in the predic- Tweedie. tions, inhibition and inactivation of caffeine (CYP1A2) and midazolam (CYP3A) were significantly overpredicted, but tolbutamide (CYP2C9) References and CYP2C8 substrates repaglinide and cerivastatin had only minor Backman JT, Honkalammi J, Neuvonen M, Kurkinen KJ, Tornio A, Niemi M, and Neuvonen PJ differences (calculations not shown). This difference may be attribut- (2009) CYP2C8 activity recovers within 96 hours after gemfibrozil dosing: estimation of able primarily to somewhat lower fm values of tolbutamide, repaglinide, CYP2C8 half-life using repaglinide as an in vivo probe. Drug Metab Dispos 37:2359–2366. and cerivastatin compared with caffeine and midazolam (fm values are Brennan BJ, Xu ZX, and Grippo JF (2013) Effect of peginterferon alfa-2a (40KD) on cytochrome P450 isoenzyme activity. Br J Clin Pharmacol 75:497–506. referenced in (Supplemental Table 5). Owing to the mild nature of the Brown HS, Ito K, Galetin A, and Houston JB (2005) Prediction of in vivo drug-drug interactions DDI observed in the clinical study, comedications that are CYP1A2, from in vitro data: impact of incorporating parallel pathways of drug elimination and inhibitor absorption rate constant. Br J Clin Pharmacol 60:508–518. CYP2C9, and CYP3A4 substrates were not restricted from use in the Callegari E, Kalgutkar AS, Leung L, Obach RS, Plowchalk DR, and Tse S (2013) Drug Phase 3 studies. Although the prediction for CYP1A2 and CYP2C9 was metabolites as cytochrome p450 inhibitors: a retrospective analysis and proposed algorithm for evaluation of the pharmacokinetic interaction potential of metabolites in drug discovery and less accurate than for CYP3A4, the predictions for the former isoforms development. Drug Metab Dispos 41:2047–2055. were not unacceptable. Also, using this approach, a moderate-to-high Chen LZ, Sabo JP, Philip E, Rowland L, Mao Y, Latli B, Ramsden D, Mandarino DA, and Sane RS (2015) Mass balance, metabolite profile, and in vitro-in vivo comparison of clearance pathways increase in AUC of 2.2-fold and 5.5-fold, respectively, was predicted of deleobuvir, a hepatitis C virus polymerase inhibitor. Antimicrob Agents Chemother 59:25–37. for CYP2C8 substrates repaglinide [fm,CYP2C8 ;0.51 (Säll et al., 2012)] Chu V, Einolf HJ, Evers R, Kumar G, Moore D, Ripp S, Silva J, Sinha V, Sinz M, and Skerjanec ; – A (2009) In vitro and in vivo induction of cytochrome p450: a survey of the current practices and cerivastatin [fm CYP2C8 0.82; considered as 1 fm,CYP3A4 (Ohno and recommendations: a pharmaceutical research and manufacturers of america perspective. et al., 2007)], whereas a net induction was predicted for CYP2B6 Drug Metab Dispos 37:1339–1354. Einolf HJ, Chen L, Fahmi OA, Gibson CR, Obach RS, Shebley M, Silva J, Sinz MW, Unadkat substrate efavirenz, a known autoinducer. The effect on these enzymes JD, and Zhang L, et al. (2014) Evaluation of various static and dynamic modeling methods was not evaluated clinically because relatively few drugs are metabo- to predict clinical CYP3A induction using in vitro CYP3A4 mRNA induction data. Clin Pharmacol Ther 95:179–188. lized by them. Appropriate restrictions were implemented during larger EMA (2012) European Medicines Agency; Committee for Human Medicinal Products (CHMP); patient studies for comedications that are highly cleared by CYP2B6 Guideline on the investigation of drug interactions. 6-21-2012. 9-13-0013. Online source P450 DDI of Deleobuvir and Its Metabolites 475

Faber MS and Fuhr U (2004) Time response of cytochrome P450 1A2 activity on cessation of competitive and time-dependent enzyme inhibition by parent compound and its metabolite in heavy smoking. Clin Pharmacol Ther 76:178–184. both liver and gut - the effect of diltiazem on the time-course of exposure to triazolam. Eur J Fahmi OA, Boldt S, Kish M, Obach RS, and Tremaine LM (2008a) Prediction of drug-drug Pharm Sci 39:298–309. interactions from in vitro induction data: application of the relative induction score approach Sabo JP, Kort J, Ballow C, Kashuba AD, Haschke M, Battegay M, Girlich B, Ting N, Lang B, using cryopreserved human hepatocytes. Drug Metab Dispos 36:1971–1974. and Zhang W, et al. (2015) Interactions of the hepatitis C virus protease inhibitor faldaprevir Fahmi OA, Hurst S, Plowchalk D, Cook J, Guo F, Youdim K, Dickins M, Phipps A, Darekar A, with cytochrome P450 enzymes: in vitro and in vivo correlation. J Clin Pharmacol 55:467–477. and Hyland R, et al. (2009) Comparison of different algorithms for predicting clinical drug- Säll C, Houston JB, and Galetin A (2012) A comprehensive assessment of repaglinide metabolic drug interactions, based on the use of CYP3A4 in vitro data: predictions of compounds as pathways: impact of choice of in vitro system and relative enzyme contribution to in vitro precipitants of interaction. Drug Metab Dispos 37:1658–1666. clearance. Drug Metab Dispos 40:1279–1289. Fahmi OA, Kish M, Boldt S, and Obach RS (2010) Cytochrome P450 3A4 mRNA is a more Shou M, Hayashi M, Pan Y, Xu Y, Morrissey K, Xu L, and Skiles GL (2008) Modeling, reliable marker than CYP3A4 activity for detecting pregnane X receptor-activated induction of prediction, and in vitro in vivo correlation of CYP3A4 induction. Drug Metab Dispos 36:2355–2370. drug-metabolizing enzymes. Drug Metab Dispos 38:1605–1611. Sinz M, Wallace G, and Sahi J (2008) Current industrial practices in assessing CYP450 enzyme Fahmi OA, Maurer TS, Kish M, Cardenas E, Boldt S, and Nettleton D (2008b) A combined induction: preclinical and clinical. AAPS J 10:391–400. model for predicting CYP3A4 clinical net drug-drug interaction based on CYP3A4 inhibition, Tornio A, Filppula AM, Kailari O, Neuvonen M, Nyrönen TH, Tapaninen T, Neuvonen PJ, inactivation, and induction determined in vitro. Drug Metab Dispos 36:1698–1708. Niemi M, and Backman JT (2014) Glucuronidation converts clopidogrel to a strong time- FDA (2012) Guidance for Industry: Drug Interaction Studies-Study Design, Data Analysis, dependent inhibitor of CYP2C8: a phase II metabolite as a perpetrator of drug-drug inter- Implications for Dosing and Labeling Recommendations (Draft). 9-13-0013. actions. Clin Pharmacol Ther 96:498–507. Fuhr U, Jetter A, and Kirchheiner J (2007) Appropriate phenotyping procedures for drug me- Varma MV, Lai Y, Kimoto E, Goosen TC, El-Kattan AF, and Kumar V (2013a) Mechanistic tabolizing enzymes and transporters in humans and their simultaneous use in the “cocktail” modeling to predict the transporter- and enzyme-mediated drug-drug interactions of repagli- approach. Clin Pharmacol Ther 81:270–283. nide. Pharm Res 30:1188–1199. Galetin A, Gertz M, and Houston JB (2008) Potential role of intestinal first-pass metabolism in Varma MV, Lin J, Bi YA, Rotter CJ, Fahmi OA, Lam JL, El-Kattan AF, Goosen TC, and Lai Y the prediction of drug-drug interactions. Expert Opin Drug Metab Toxicol 4:909–922. (2013b) Quantitative prediction of repaglinide-rifampicin complex drug interactions using Gupta SK, Kolz K, and Cutler DL (2011) Effects of multiple-dose alfa-2b on dynamic and static mechanistic models: delineating differential CYP3A4 induction and the activity of drug-metabolizing enzymes in persons with chronic hepatitis C. Eur J Clin OATP1B1 inhibition potential of rifampicin. Drug Metab Dispos 41:966–974. Pharmacol 67:591–599. Venkatakrishnan K and Obach RS (2007) Drug-drug interactions via mechanism-based cyto- Downloaded from Hinton LK, Galetin A, and Houston JB (2008) Multiple inhibition mechanisms and prediction of chrome P450 inactivation: points to consider for risk assessment from in vitro data and clinical drug-drug interactions: status of metabolism and transporter models as exemplified by pharmacologic evaluation. Curr Drug Metab 8:449–462. gemfibrozil-drug interactions. Pharm Res 25:1063–1074. Venkatakrishnan K, Obach RS, and Rostami-Hodjegan A (2007) Mechanism-based inactivation Lin JH (2006) CYP induction-mediated drug interactions: in vitro assessment and clinical of human cytochrome P450 enzymes: strategies for diagnosis and drug-drug interaction risk implications. Pharm Res 23:1089–1116. assessment. Xenobiotica 37:1225–1256. Lutz JD, VandenBrink BM, Babu KN, Nelson WL, Kunze KL, and Isoherranen N (2013) Wang JS, Neuvonen M, Wen X, Backman JT, and Neuvonen PJ (2002) Gemfibrozil inhibits Stereoselective inhibition of CYP2C19 and CYP3A4 by fluoxetine and its metabolite: impli- CYP2C8-mediated cerivastatin metabolism in human liver microsomes. Drug Metab Dispos cations for risk assessment of multiple time-dependent inhibitor systems. Drug Metab Dispos 30:1352–1356. 41:2056–2065. Yu H and Tweedie D (2013) A perspective on the contribution of metabolites to drug-drug dmd.aspetjournals.org McCabe M, Sane RS, Keith-Luzzi M, Xu J, King I, Whitcher-Johnstone A, Johnstone N, Tweedie interaction potential: the need to consider both circulating levels and inhibition potency. Drug DJ, and Li Y (2015) Defining the role of gut bacteria in the metabolism of deleobuvir: in vitro Metab Dispos 41:536–540. and in vivo studies. Drug Metab Dispos 43:1612–1618 10.1124/dmd.115.064477. Yu H, Balani SK, Chen W, Cui D, He L, Humphreys WG, Mao J, Lai WG, Lee AJ, and Lim HK, Medina-Díaz IM, Estrada-Muñiz E, Reyes-Hernández OD, Ramírez P, Vega L, and Elizondo G et al. (2015) Contribution of metabolites to P450 inhibition-based drug-drug interactions: (2009) Arsenite and its metabolites, MMA(III) and DMA(III), modify CYP3A4, PXR and scholarship from the drug metabolism leadership group of the innovation and quality con- RXR alpha expression in the small intestine of CYP3A4 transgenic mice. Toxicol Appl sortium metabolite group. Drug Metab Dispos 43:620–630. Pharmacol 239:162–168. Zeuzem S, Asselah T, Angus P, Zarski JP, Larrey D, Müllhaupt B, Gane E, Schuchmann M, Obach RS, Walsky RL, and Venkatakrishnan K (2007) Mechanism-based inactivation of human Lohse AW, and Pol S, et al. (2013) Faldaprevir (BI 201335), deleobuvir (BI 207127) and cytochrome p450 enzymes and the prediction of drug-drug interactions. Drug Metab Dispos ribavirin oral therapy for treatment-naive HCV genotype 1: SOUND-C1 final results. Antivir 35:246–255. Ther 18:1015–1019.

Ogburn ET, Jones DR, Masters AR, Xu C, Guo Y, and Desta Z (2010) Efavirenz primary and Zeuzem S, Soriano V, Asselah T, Bronowicki JP, Lohse AW, Müllhaupt B, Schuchmann M, at ASPET Journals on September 27, 2021 secondary metabolism in vitro and in vivo: identification of novel metabolic pathways and Bourlière M, Buti M, Roberts SK, Gane EJ, Stern JO, Vinisko R, Kukolj G, Gallivan JP, cytochrome P450 2A6 as the principal catalyst of efavirenz 7-hydroxylation. Drug Metab Böcher WO, and Mensa FJ. (2013) Faldaprevir and deleobuvir for HCV genotype 1 infection. Dispos 38:1218–1229. N Engl J Med 369(7):630–9. Ogilvie BW, Zhang D, Li W, Rodrigues AD, Gipson AE, Holsapple J, Toren P, and Parkinson A Zhang JG, Ho T, Callendrello AL, Crespi CL, and Stresser DM (2010) A multi-endpoint eval- (2006) Glucuronidation converts gemfibrozil to a potent, metabolism-dependent inhibitor of uation of cytochrome P450 1A2, 2B6 and 3A4 induction response in human hepatocyte cul- CYP2C8: implications for drug-drug interactions. Drug Metab Dispos 34:191–197. tures after treatment with b-naphthoflavone, phenobarbital and rifampicin. Drug Metab Lett 4: Ohno Y, Hisaka A, and Suzuki H (2007) General framework for the quantitative prediction of 185–194. CYP3A4-mediated oral drug interactions based on the AUC increase by coadministration of Zhao P, Zhang L, Grillo JA, Liu Q, Bullock JM, Moon YJ, Song P, Brar SS, Madabushi R, standard drugs. Clin Pharmacokinet 46:681–696. and Wu TC, et al. (2011) Applications of physiologically based pharmacokinetic Petzer JP, Navamal M, Johnson JK, Kwak MK, Kensler TW, and Fishbein JC (2003) Phase 2 (PBPK) modeling and simulation during regulatory review. Clin Pharmacol Ther 89: enzyme induction by the major metabolite of oltipraz. Chem Res Toxicol 16:1463–1469. 259–267. Piedade R, Traub S, Bitter A, Nüssler AK, Gil JP, Schwab M, and Burk O (2015) Carboxyme- Zhou SF, Chan E, Zhou ZW, Xue CC, Lai X, and Duan W (2009) Insights into the structure, floquine, the major metabolite of the antimalarial drug mefloquine, induces drug-metabolizing function, and regulation of human cytochrome P450 1A2. Curr Drug Metab 10:713–729. enzyme and transporter expression by activation of pregnane X receptor. Antimicrob Agents Chemother 59:96–104. Renwick AB, Watts PS, Edwards RJ, Barton PT, Guyonnet I, Price RJ, Tredger JM, Pelkonen O, Address correspondence to: Dr. Rucha S. Sane, Translational Medicine and Boobis AR, and Lake BG (2000) Differential maintenance of cytochrome P450 enzymes in Clinical Pharmacology, Boehringer Ingelheim Pharmaceuticals, Inc., 900 Ridge- cultured precision-cut human liver slices. Drug Metab Dispos 28:1202–1209. Rowland Yeo K, Jamei M, Yang J, Tucker GT, and Rostami-Hodjegan A (2010) Physiologically bury Rd., Ridgefield, CT 06877. E-mail: [email protected] based mechanistic modelling to predict complex drug-drug interactions involving simultaneous