Utilizing In Vitro and PBPK Tools to Link ADME Characteristics to Plasma Profiles: Case Example Nifedipine Immediate Release Formulation

Transcription

1 Utilizing In Vitro and PBPK Tools to Link ADME Characteristics to Plasma Profiles: Case Example Nifedipine Immediate Release Formulation CHRISTIAN WAGNER, 1 KIRSTIN THELEN, 2 STEFAN WILLMANN, 2 ARZU SELEN, 3 JENNIFER B. DRESSMAN 1 1 Institute of Pharmaceutical Technology, Goethe University, Frankfurt am Main, Germany 2 Technology Package Computational Systems Biology, Bayer Technology Services GmbH, Leverkusen, Germany 3 Office of New Drug Quality Assessment/OPS/CDER, United States Food and Drug Administration, Silver Spring, Maryland Received 4 March 2013; revised 22 April 2013; accepted 23 April 2013 Published online 20 May 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI /jps ABSTRACT: One of the most prominent food drug interactions is the inhibition of intestinal cytochrome P450 (CYP) 3A enzymes by grapefruit juice ingredients, and, as many drugs are metabolized via CYP 3A, this interaction can be of clinical importance. Calcium channelblocking agents of the dihydropyridine type, such as felodipine and nifedipine, are subject to extensive intestinal first pass metabolism via CYP 3A, thus resulting in significantly enhanced in vivo exposure of the drug when administered together with grapefruit juice. Physiologically based pharmacokinetic (PBPK) modeling was used to simulate pharmacokinetics of a nifedipine immediate release formulation following concomitant grapefruit juice ingestion, that is, after inhibition of small intestinal CYP 3A enzymes. For this purpose, detailed data about CYP 3A levels were collected from the literature and implemented into commercial PBPK software. As literature reports show that grapefruit juice (i) leads to a marked delay in gastric emptying, and (ii) rapidly lowers the levels of intestinal CYP 3A enzymes, inhibition of intestinal first pass metabolism following ingestion of grapefruit juice was simulated by altering the intestinal CYP 3A enzyme levels and simultaneously decelerating the gastric emptying rate. To estimate the in vivo dispersion and dissolution behavior of the formulation, dissolution tests in several media simulating both the fasted and fed state stomach and small intestine were conducted, and the results from the in vitro dissolution tests were used as input function to describe the in vivo dissolution of the drug. Plasma concentration time profiles of the nifedipine immediate release formulation both with and without simultaneous CYP 3A inhibition were simulated, and the results were compared with data gathered from the literature. Using this approach, nifedipine plasma profiles could be simulated well both with and without enzyme inhibition. A reduction in small intestinal CYP 3A levels by 60% was found to yield the best results, with simulated nifedipine concentration time profiles within 20% of the in vivo observed results. By additionally varying the dissolution input of the PBPK model, a link between the dissolution characteristics of the formulation and its in vivo performance could be established Wiley Periodicals, Inc. and the American Pharmacists Association J Pharm Sci 102: , 2013 Keywords: absorption; bioavailability; computational ADME; cytochrome P450; dissolution; drug metabolizing enzymes; first-pass metabolism; pharmacokinetics; quality by design (QbD) INTRODUCTION It is well known that the main factors that influence the bioavailability of a drug after oral administra- Correspondence to: Jennifer B. Dressman (Telephone: ; Fax: ; dressman@em.unifrankfurt.de) Journal of Pharmaceutical Sciences, Vol. 102, (2013) 2013 Wiley Periodicals, Inc. and the American Pharmacists Association tion are the solubility and dissolution of the drug at the site of absorption, degradation and complexation of the drug in the gut lumen, the ability of the drug to permeate the gut wall and enter into systemic circulation [including the possibility of the drug being a substrate for efflux transporters, e.g. P-glycoprotein (P-gp)], and metabolic degradation, that is first pass metabolism in the gut wall and liver. 1 JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER

2 3206 WAGNER ET AL. Besides solubility and permeability issues, first pass metabolism, including in the gut wall, can be an important limitation to the oral bioavailability of certain drugs. With a contribution of approximately 80%, the most prominent enzymes responsible for intestinal first pass metabolism belong to the cytochrome P450 (CYP) 3A subfamily. 2 In proximal regions of the small intestine, CYP 3A4 is the most prominent enzyme of the CYP 3A subfamily. Other enzyme subfamilies present in the small intestine belong, for example, to CYP 2C and 2D. Interestingly, the individual contributions of the various CYP subfamilies to intestinal metabolism are different compared with hepatic metabolism. 2,3 As the intestinal CYP enzymes are located in enterocytes, which constitute the surface cell monolayer of the intestinal mucosa, 4 drug molecules and food components can come in contact with these enzymes during the absorption process and thus be metabolized. Vice versa, intestinal CYP enzymes can be influenced by drug and food components, resulting in either increased (e.g., enzyme induction by increased enzyme expression) or decreased (e.g., enzyme inhibition) CYP activity. The interaction between grapefruit juice and dihydropyridine calcium channel blockers, such as felodipine and nifedipine, is one of the most well-known examples of enzyme inhibition caused by food components However, not only calcium channel blockers, but also other CYP 3A substrates show clinical interactions with grapefruit juice, for example terfenadine, 16,17 midazolam, 18 triazolam, 19 cyclosporine, 20,21 atorvastatin, 22 and saquinavir. 23 Grapefruit juice contains flavonoids such as naringin, naringenin, quercetin, and kaempferol, and furanocoumarins, for example bergamottin and dihydroxybergamottin, which are known to be potent and selective inhibitors of CYP 3A enzymes. By contrast, other intestinal CYP enzymes, for example CYP 2D or CYP 1A, are not inhibited by grapefruit juice components. 9,10,13,14,24,25 Ingestion of grapefruit juice leads to a pronounced and long-lasting inhibition of small intestinal CYP 3A enzymes, which, in turn, may lead to clinical implications for drugs that are metabolized via this enzyme subfamily. After ingestion of grapefruit juice, small intestinal CYP 3A protein activity decreases rapidly, 5,6,14 and the explanation for this decrease in CYP activity is an irreversible, concentrationdependent inhibition of the CYP enzymes because of protein degradation. 9,13 By contrast, short-term ingestion of grapefruit juice does not affect hepatic CYP 3A enzymes significantly, and inhibition of hepatic CYP 3A enzymes occurs only after long-term ingestion of grapefruit juice After administration of grapefruit juice, the time to reach the maximum nifedipine plasma concentration, t max, was found to be increased. After concomi- JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013 Figure 1. Structural formula of the calcium channelblocking agent nifedipine. tant ingestion of grapefruit juice and nifedipine, Odou et al. 11 reported a delay in gastric emptying of approximately 30 min because of the low ph and the caloric content of grapefruit juice, and this delay in gastric emptying is a reasonable explanation for the increase in t max. Similar observations have also been reported for felodipine pharmacokinetics. 7 Physiologically based pharmacokinetic (PBPK) modeling can be regarded as a mathematical, in silico (computer-assisted) approach to simulate and predict the absorption, distribution, metabolism, and excretion (ADME) properties of compounds, and this technique has gained increased importance in recent years. In terms of improving predictability of the simulations, biorelevant dissolution testing can be coupled with PBPK software to better reflect the in vivo dissolution, especially of poorly soluble drug compounds. In addition, the PBPK modeling approach allows the integration of information related to organ systems that may be influenced due to disease states and/or reflect values unique to special patient populations and various drug delivery profiles The purpose of this work was to implement a detailed description of small intestinal CYP 3A first pass metabolism into the generic PBPK software tool PK-Sim R (Bayer Technology Services, Leverkusen, Germany) to simulate the pharmacokinetics after intravenous and oral administration of the model compound nifedipine. In a subsequent step, CYP 3A enzyme inhibition caused by grapefruit juice ingestion and its impact on the intravenous and oral pharmacokinetics of the model compound was simulated. Nifedipine (Fig. 1) was chosen as model compound as the drug s pharmacokinetic properties are well described in literature and as it is subject to intestinal and hepatic first pass metabolism via CYP 3A enzymes. The drug is a calcium channel-blocking agent used in the therapy of angina pectoris, Raynaud s phenomenon, and hypertension. It inhibits the transmembrane calcium influx into smooth muscle cells and hence leads to a prompt and marked hypotensive effect. Immediate release (IR) nifedipine formulations may show adverse effects, such as headache, flushing, and palpitation, caused by a rapid and profound vasodilatation. For this reason, together with the short half-life of the drug, modified release nifedipine DOI /jps

3 UTILIZING IN VITRO AND PBPK TOOLS TO LINK ADME CHARACTERISTICS TO PLASMA PROFILES 3207 formulations have gained increased importance in the therapy of hypertension Following oral administration, nifedipine can be absorbed completely over the whole length of the small intestine and colon, and the fraction absorbed of the drug is 1. 40,49 51 However, the absolute bioavailability ranges from 45% to 90%, suggesting that the drug undergoes extensive presystemic metabolism via intestinal and hepatic CYP 3A, which also leads to high variability in nifedipine pharmacokinetics. Nifedipine is metabolized completely by oxidation and glucuronidation, and the metabolites are pharmacologically inactive. The drug s elimination half-life is approximately 2 4 h and thus rather short. 40,41,44,47,50 54 With its low aqueous solubility but high permeability, nifedipine is a typical class 2 drug substance according to the Biopharmaceutics Classification System (BCS). 55 As one can see from the pharmacokinetics of nifedipine, there are various factors that influence the drug s ADME properties. In the first part of this work, we focus on a quantitative description of nifedipine s first pass metabolism and the impact of grapefruit juice ingestion on the drug s pharmacokinetics. The second part of this study aims to establish a link between the dissolution characteristics of nifedipine and its bioperformance. By using PBPK modeling, we were also able to investigate the relative influences of dissolution, gastric emptying, and first pass metabolism on the plasma profile of nifedipine. MATERIALS AND METHODS Materials Nifedipine pure drug substance (batch N7634) was obtained from Sigma Aldrich Chemie GmbH (Steinheim, Germany). Adalat R 10 mg (Bayer Vital, Leverkusen, Germany), an IR soft gelatine capsule filled with 10 mg nifedipine, dissolved in polyethyleneglycol 400 and peppermint oil, was obtained from a local drug distributor (Phoenix Pharmahandel GmbH & Co KG, Hanau, Germany). Egg-phosphatidylcholine (Lipoid E PC R ) was kindly donated by Lipoid GmbH (Ludwigshafen, Germany). Long-life, heat treated 3.5% whole fat milk (Milfina, Thalfang, Germany) and grapefruit juice (REWE Handelsgruppe GmbH, Cologne, Germany) were purchased commercially. Glycerol monooleate was obtained from Danisco Specialities (Brabrand, Denmark), and pepsin (0.51 U/mg) was obtained from Fluka Chemie AG (Buchs, Switzerland). Sodium taurocholate was obtained from Prodotti Chimici E Alimentari S.P.A. (Basaluzzo, Italy). Sodium oleate was purchased from Riedel-de Haën (Seelze, Germany). Hydrochloric acid, sodium chloride, and sodium hydroxide were obtained from VWR International (Fontenay sous Bois, France). Maleic acid was obtained from Merck KGaA (Darmstadt, Germany). Dichloromethane (analytical grade), acetonitrile, and methanol (both HPLC grade) were obtained from Merck KGaA (Darmstadt, Germany). Ultrapurified water for high-performance liquid chromatography (HPLC) was produced freshly upon usage. Nifedipine Light Stability, Solubility, and Dissolution Light Stability of Nifedipine Nifedipine is known to be light sensitive and hence, all experiments were performed under protection from incandescent light in a windowless room. A yellow light bulb (Osram, Augsburg, Germany) was used as the exclusive light source. In addition, the dissolution tester was covered with aluminum foil. Potential nifedipine degradation was checked under dissolution test conditions (i.e. after subjecting samples to yellow light as the only light source for 4 h). The experiments were performed in triplicate (n = 3). As a control, a nifedipine sample was kept under total protection from all light (n = 1). Media Used for the Solubility and Dispersion/Dissolution Studies For the solubility studies, the biorelevant media Fasted State Simulated Gastric Fluid (FaSSGF ph 1.6) and an updated version of Fasted State Simulated Intestinal Fluid (FaSSIF v2 ph 6.5) were used to simulate preprandial conditions in the upper gastrointestinal tract, and Fed State Simulated Gastric Fluid (FeSSGF ph 5.0) and an updated version of Fed State Simulated Intestinal Fluid (FeSSIF v2 ph 5.8) were used to simulate postprandial conditions in the upper gastrointestinal tract. Nifedipine solubility was also tested in the corresponding blank buffer solutions of those media, that is without addition of bile salts, lecithin, or glycerol monooleate. The composition and preparation of these media is well described in literature. 56,57 Additional solubility experiments were performed in grapefruit juice and in two compendial media, that is in Simulated Gastric Fluid sine pepsin (SGF sp ph 1.2) and Simulated Intestinal Fluid sine pancreatin (SIF sp ph6.8). Dispersion tests of the Adalat R 10 mg IR formulation were conducted in FaSSGF, FeSSGF, FaSSIF v2, FeSSIF v2, SGF sp, and SIF sp. According to the study protocol of the in vivo study to be simulated, 12 the Adalat R formulation was administered in the fasted state together with 200 ml of double strength grapefruit juice 2 h before and again at the time of dosing. To reflect the in vivo dissolution of nifedipine as closely as possible, an additional dispersion DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013

4 3208 WAGNER ET AL. test in a dissolution medium containing grapefruit juice was conducted. With an assumed basal volume of gastric juice of 50 ml and an additional 200 ml of ingested grapefruit juice, 12 this medium consisted of 20% FaSSGF ph 1.0 and 80% grapefruit juice (ph 3.3). The final ph of the medium was 2.7. Solubility Study For the solubility study in all media except FeSSGF and grapefruit juice, the Uniprep TM method (Whatman Inc., Sanford, Maine), a syringeless filter device (0.45 :m pore size), was used. 58 Solubility of nifedipine in grapefruit juice and FeS- SGF, a medium that consists of 50% milk, was determined using the shake flask method. As neither milk nor grapefruit juice can be filtered through 0.45 :m filters, 2.7 :m glass microfiber filters (25 mm GD/X, Whatman Inc., Florham Park, New Jersey) were used instead. Although this filter pore size (2.7 :m) is relatively coarse, the particle size in the nifedipine batch used for the solubility studies was mostly in the : m range (unpublished results). However, it cannot be completely excluded that some particles <2.7 :m may have passed the filter. To remove proteins from the samples and thus enable HPLC analysis, 1.0 ml acetonitrile was added to 0.5 ml of the filtered sample and centrifuged for 8 min at 20,800 g. The clear supernatant was then analyzed using HPLC-UV (High-Performance Liquid Chromatography section). The dispersions of the drug in the corresponding media were kept on an orbital shaker for 24 h at 37 C and quantified using HPLC. All solubility measurements were conducted at least in triplicate. The results were compared using analysis of variance (ANOVA) and Holm Sidak test (SigmaPlot R 11.0, Systat Software Inc., San Jose, California), and differences were considered significant based on a significance level of Dispersion and Dissolution Studies The dispersion tests of the Adalat R 10 mg IR formulation and the dissolution tests of unformulated drug substance (filled into hard gelatine capsules) in FeSSGF, FaSSIF v2, FeSSIF v2, and SIF sp were performed using a USP apparatus 2 (paddle assembly; Erweka DT 700, Heusenstamm, Germany) with a volume of 500 ml medium per vessel. To better mimic conditions both in the fasted state stomach and after ingestion of 200 ml grapefruit juice, dispersion tests of Adalat R in FaSSGF and in the grapefruit juice-containing medium (Media Used for the Solubility and Dispersion/Dissolution Studies section) were conducted in a mini paddle assembly (Erweka DT 700, Heusenstamm, Germany), a scaled JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013 down version of the USP apparatus 2, 59 at a medium volume of 250 ml. For both set-ups, paddle revolution speed was 75 rpm, and the temperature was maintained at 37 C ± 0.5 C. Sampling times were 5, 10, 15, 20, 30, 45, 60, 90, 120, and 180 min for SGF sp, FaSSGF, and grapefruit juice and additionally 240 min for FeSSGF, FaS- SIF v2, FeSSIF v2, and SIF sp. At each sampling point, an aliquot of 5 ml was taken per vessel and filtered immediately through a 0.45 :m PTFE filter (Rezist R 30, Whatman GmbH, Dassel, Germany) or a 2.7 :m glass microfiber filter (FeSSGF and grapefruit juice), discarding the first 2 ml; the withdrawn volume was replaced with fresh, prewarmed medium. The dispersion and dissolution tests were performed either in triplicate (grapefruit juice, SIF sp, SGF sp, and unformulated drug) or with n = 6(all other media). High-Performance Liquid Chromatography (HPLC) Nifedipine was quantified using an isocratic HPLC- UV assay. The HPLC system consisted of a VWR Hitachi Organizer, an L-2400 UV detector, an L-2200 autosampler, and an L-2130 pump (VWR International GmbH, Darmstadt, Germany). The chromatograms were evaluated using the EZChrome R Elite software (v ; VWR Hitachi, Japan). The samples were quantified using a LiChroCART R Purospher R RP 18 e( mm, 5 :m; Merck KGaA, Darmstadt, Germany) HPLC column. The flow rate was 1.0 ml/ min, and nifedipine UV absorption was measured at a wavelength of 235 nm. The injection volume was 25 :L per run. The HPLC analysis was performed under ambient conditions. To separate unchanged nifedipine from its degradation product (Light Stability of Nifedipine section), a mobile phase comprising 50% ultra purified water, 25% acetonitrile, and 25% methanol was used for the stability test. 60 The mobile phase for the solubility and dispersion/dissolution studies consisted of 30% ultra purified water, 45% acetonitrile, and 25% methanol. The total run-time per HPLC run was 30 min (light stability testing) and 7 min (solubility and dissolution study), respectively. The limit of quantification (LoQ) for both methods was 20 ng/ml. Physiologically Based Pharmacokinetic Model Source of In Vivo Data The in vivo data utilized in this work were obtained from a clinical study conducted to evaluate the influence of grapefruit juice on the absolute bioavailability of nifedipine. 12 For this purpose, nifedipine was administered in a crossover study design to eight DOI /jps

5 UTILIZING IN VITRO AND PBPK TOOLS TO LINK ADME CHARACTERISTICS TO PLASMA PROFILES 3209 Table 1. Parameter Nifedipine Input Parameters Used for the PBPK Model Value Plasma protein binding 97% 50,52,69,70 Molecular weight g/mol Log P Intestinal permeability cm/s a pka Not ionized under physiological conditions b11 Intestinal solubility (FaSSIF v2) 17.1 :g/ml c a Calculated by the PBPK software from the physicochemical properties of nifedipine. b With a pka of approximately -1 (basic), nifedipine is not ionized under physiological conditions and can thus be considered as a neutral compound. 11 c Result from this study. healthy volunteers on four different occasions as follows: (i) 2.5 mg intravenously together with 200 ml water orally, (ii) 2.5 mg intravenously together with 200 ml double strength grapefruit juice orally both 2 h before and at the time of dosing, (iii) one Adalat R 10 mg IR capsule orally in the fasted state together with 200 ml water, and (iv) one Adalat R 10 mg IR capsule orally in the fasted state together with 200 ml double strength grapefruit juice both 2 h before and at the time of dosing. 12 PBPK Model Plasma profiles following intravenous and oral administration of nifedipine, both with and without simultaneous administration of grapefruit juice, were simulated using PBPK software tools (PK-Sim R v and MoBi R v , both from Bayer Technology Services, Leverkusen, Germany). The set-up of the PBPK structure and the source of the default parameters used by the software have been described in detail elsewhere However, in the version of the PK-Sim R software that was available at the time of the study, the input of enzyme contents directly into the intracellular compartment of the intestine needed for the detailed description of nifedipine gut wall metabolism was not possible because of the structure of the continuous-tube absorption model. 64 To circumvent this shortcoming, a PBPK model that describes nifedipine distribution after intravenous administration of nifedipine was developed in PK-Sim R. This model was then exported to the MoBi R software, where the new gastrointestinal transit and absorption model was added. 61,62 In the novel compartmental gastrointestinal model, detailed information about CYP 3A concentrations in the duodenum and proximal and distal jejunum and ileum, respectively, were introduced into the software to describe nifedipine gut wall metabolism. Table 1 summarizes important input parameters used for the simulations. The information about CYP 3A amounts in the liver and small intestine needed for model development Table 2. CYP 3A Characteristics Used for the Computer Simulations Compartment Regional CYP 3A Amounts (nmol) 72,73 CYP 3A Concentration (:M) Liver Duodenum Jejunum 38.4 Proximal: 1.05 Distal: 0.99 Ileum 22.4 Proximal: 0.84 Distal: 1.44 were collected from the literature. 72,73 As total regional CYP 3A enzyme levels in the small intestine and the liver were expressed in nmol, the amounts of CYP enzymes had to be converted into concentrations, using Eq. 1 C CYP3A = A CYP3A f cell V comp (1) where C CYP3A is the concentration of CYP 3A enzymes in the liver and each intestinal compartment, A CYP3A is the molar amount of CYP 3A enzymes in the liver and the mucosa of each intestinal compartment, f cell is the cellular fraction of the liver and the mucosa of each intestinal segment, and V comp is the organ volume of the liver and the mucosa of each intestinal segment. Regional small intestinal CYP 3A concentrations were calculated on a surface area-based approach. The values for f cell and V comp were the default values from the software. Table 2 summarizes the CYP 3A concentrations used for the simulations. Nifedipine metabolism was assumed to follow Michaelis Menten-kinetics (Eq. 2) v = v max C NIF K M + C NIF (2) where v is the rate of reaction, v max is the maximum rate of reaction, C NIF is the concentration of unmetabolized nifedipine in each compartment, and K M is the Michaelis Menten-constant. However, v max can be replaced by the term k cat C CYP3A, where k cat is the catalytic constant (turnover number) and C CYP3A is the CYP 3A concentration in the corresponding compartment, and Eq. 2 can thus be written as Eq. 3 v = k cat C CYP3A C NIF K M + C NIF (3) The value for K M is 10.1 :M and was taken from literature, 72 the time-dependent change in C NIF is calculated automatically by the software, and k cat was fitted to match the in vivo plasma profile after administration of 2.5 mg nifedipine intravenously. To optimize k cat and fit it to the intravenous DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013

6 3210 WAGNER ET AL. nifedipine data, the fminsearch algorithm (MATLAB R, The MathWorks TM Inc., Natick, Massachusetts) was used; the error function used was the root mean square function. The value for k cat was then used for all simulations, that is intravenous and oral, both with and without concomitant ingestion of grapefruit juice. Simultaneous oral administration of nifedipine and grapefruit juice leads to a significant increase in nifedipine bioavailability due to inhibition of small intestinal first pass metabolism. The most reasonable explanation for the CYP 3A enzyme inhibition is rapid protein degradation, 9,13 and it is reported that small intestinal CYP 3A concentrations decrease by approximately 45% 62% after grapefruit juice ingestion. 9,13 Other researchers report a decrease in CYP 3A activity between 10% and 100%, 24 32% and 56%, 13 or 73%, 25 depending on the nature and concentration of the active ingredients. According to the in vivo study performed by Rashid et al., ml double strength grapefruit juice was administered 4 h before and again at the time of dosing. However, no detailed information about the extent of CYP 3A protein degradation covering this situation was available, so the small intestinal CYP 3A concentration was fitted within a reasonable range to match the in vivo Area under the curve (AUC) observed after ingestion of two times 200 ml double strength grapefruit juice. 12 An acceptable match was defined as an f 1 factor 20% for the comparison of the simulated and the in vivo measured plasma profile (Plasma Profile Comparison section). As short-time grapefruit juice ingestion principally affects only intestinal first pass metabolism, 9,13 the hepatic CYP 3A concentration was held constant in all simulations. The default value of 15 min for 63% gastric emptying to occur was used for the simulations following oral administration of nifedipine without grapefruit juice. As administration of grapefruit juice leads to an approximately 30 min delay in gastric emptying time, 7,11 the gastric emptying time for the simulations following simultaneous oral administration of nifedipine and grapefruit juice was increased to 45 min. All other parameters of the software remained constant and were the default values of the software. To describe intralumenal nifedipine dissolution, a two-parametric Weibull function (Eq. 4) [ (t Tlag ) b ] m = 1 exp a where m is the cumulative fraction of drug dissolved, a is a time parameter that defines the scale of the process, b is a shape parameter, and T lag is the time until the dissolution process starts, was fitted to the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013 (4) in vitro dispersion profile of one Adalat R 10 mg IR soft gelatine capsule in FaSSIF v2 and implemented into the software. Sensitivity Analysis To estimate the factors that influence the bioperformance of the tested nifedipine IR formulation, a sensitivity analysis was performed by varying the small intestinal and hepatic CYP 3A concentrations, gastric emptying time, and dissolution characteristics. Intestinal and hepatic CYP 3A concentrations, 9,72,73 as well as gastric emptying, are known to be highly variable. Thus, a sensitivity analysis was performed to estimate the influence of variations in small intestinal and hepatic CYP 3A enzyme concentrations on the pharmacokinetics of nifedipine. For this purpose, a 10-fold variation in small intestinal CYP 3A concentration was simulated. 4 To estimate the influence of hepatic metabolism on nifedipine pharmacokinetics, hepatic CYP 3A concentrations were also varied by up to a factor of 10 from the mean hepatic CYP 3A concentration. 4,73 To estimate the influence of variation in the formulation s dissolution rate on the simulated plasma profile, the dissolution input into the model was varied over a wide range. Virtual dissolution profiles were constructed with 90% dissolution times of 5, 10, 15, 20, 25, 30, 45, 60, 90, 120, and 180 min, and these virtual dissolution profiles were implemented in the PBPK model. This approach was used to establish a dissolution window (design space) for nifedipine IR formulations. The simulated profiles were then compared with the in vivo profile 12 in terms of their AUC, c max,and f 1 factor (Plasma Profile Comparison section). Plasma Profile Comparison The difference factor f 1, which gives the percent difference between two plasma profiles, was calculated according to Eq. 5 f 1 = n t=1 R t T t n t=1 R t 100% (5) where n is the number of time points used for comparison, R t is the nifedipine plasma concentration observed in vivo at time t, andt t is the corresponding predicted nifedipine plasma concentration at time t. As the acceptance criterion, differences of 20% or less between the concentration time profiles (simulated vs. observed) were deemed acceptable (f 1 20%). 78 As additional criteria for equivalence of the simulated and the in vivo plasma profiles, the ratios (point estimates) of the simulated and the in vivo observed AUC and c max were calculated, and values between DOI /jps

7 UTILIZING IN VITRO AND PBPK TOOLS TO LINK ADME CHARACTERISTICS TO PLASMA PROFILES 3211 Figure 2. Nifedipine 24-h equilibrium solubility (±SD) in :g/ml. The results are presented as mean from at least three experiments (n 3). White bars represent blank buffers from the biorelevant media and water, black bars represent compendial media, and gray bars represent grapefruit juice and biorelevant media. 80% and 125% (difference of 20%) were deemed acceptable. Note that the f 1 factor and point estimate comparisons of the simulated and the observed profiles should be interpreted as a general indication of similarity and are not intended to have any regulatory connotation. RESULTS AND DISCUSSION Nifedipine Light Stability Study Under the influence of light, nifedipine decomposes rapidly to a nitroso- and a nitropyridine-derivate, and the rate of photodecomposition is dependent on the light intensity. However, no degradation was observed when the samples were kept under dissolution test conditions, that is over 4 h protected from actinic light and with a yellow light as the only light source. Nifedipine Solubility Study Figure 2 presents the 24 h equilibrium solubility results of nifedipine in different media. Under physiological conditions, nifedipine acts as a neutral molecule, and its solubility is independent of the ph of the medium. The solubility of the drug in plain buffers falls in the range between 9.2 :g/ ml (blank FaSSIF v2) and 11.9 :g/ml (blank FeSSIF v2). By contrast, the solubility of the drug in biorelevant media, except for FaSSGF, is significantly higher than in the corresponding blank buffers ( p < for FaSSIF v2, FeSSGF, and FeSSIF v2, respectively). The difference in the solubility between FaSSGF and the blank FaSSGF buffer is statistically not significant (p > 0.05). The solubility of nifedipine in grapefruit juice lies between the solubility of the drug in the blank buffers and FaSSIF v2. The difference in solubility between grapefruit juice and the blank buffers of FaSSGF and FaSSIF v2 on the one hand and FaSSGF and FaS- SIF v2 on the other hand is statistically significant ( p < 0.001). However, it is likely that, even though there is a statistically significant difference based on the magnitude of the difference, its clinical relevance is expected to be negligible because the overall solubility of the drug in those media is still poor. For a 10 mg dose of nifedipine, the drug s solubility has to exceed 40 :g/ml to be classified as highly soluble (dose/solubility ratio 250 ml). For all tested media except FeSSGF and FeSSIF v2, the drug s solubility is far below that solubility value. Nifedipine solubility in media reflecting the fed state gastrointestinal tract, that is FeSSGF and FeS- SIF v2, is significantly higher ( p < 0.001) than in media simulating the fasted state gastrointestinal tract, that is FaSSGF and FaSSIF v2, which might lead one to assume that nifedipine would exhibit a positive food effect. However, the extent of absorption is not altered when the Adalat R IR formulation is administered with or without food, 54,79 and this observation is consistent with the release of the drug from the capsule as a solution with no subsequent precipitation at the 10 mg dose. 35 Nifedipine Dispersion and Dissolution Study Figure 3 displays the dispersion test results for the Adalat R 10 mg IR formulation (a) and the dissolution test results for the unformulated drug substance (b) in various media. For the Adalat R 10 mg IR formulation, complete drug release and dispersion was observed in all media, and the release and dispersion rates of nifedipine were comparable in all tested media. By contrast, the dissolution of the pure drug substance is considerably slower and incomplete. As has been observed before, 35 the formulation generates a marked supersaturation of nifedipine, and the nifedipine concentration in the vessels (40 :g/ml for a medium volume of 250 ml, and 20 :g/ml for a medium volume of 500 ml) is higher than the theoretically attainable equilibrium solubility (Nifedipine Solubility Study section). In the tested media (except in FeSSGF and FeSSIF v2), the nifedipine concentration is higher than one would expect from the solubility measurements, and this supersaturation is stable for at least the duration of the dissolution test. Thus, precipitation of the drug upon its passage through the gastrointestinal tract appears unlikely. For a 20 mg dose of the Adalat R formulation, gastric precipitation was observed in vitro 35 and potentially in vivo, 51 whereas precipitation was not observed for either the 5 mg or the 10 mg dose strengths. Nifedipine precipitation thus seems to be DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013

8 3212 WAGNER ET AL. In the dispersion studies, biorelevant media representing the fasted and fed state small intestine (i.e. FaSSIF v2 and FeSSIF v2) did not show advantages compared to SIF sp for the selected formulation (Adalat R 10 mg) and test conditions (paddle assembly, 500 ml test volume, 75 rpm). Figure 3. Dispersion of the Adalat R 10 mg IR formulation (a) and dissolution of 10 mg pure drug substance (b) in 250 ml FaSSGF ( ), 500 ml FeSSGF ( ), 500 ml FaSSIF v2 ( ), 500 ml FeSSIF v2 ( ), and 500 ml SIF sp (x). For the Adalat R formulation, an additional dispersion test in 250 ml grapefruit juice ( ) and 500 ml SGF sp ( ) wasconducted. The results present the mean dispersion/dissolution profiles (n = 3orn = 6) ± SD. dose-dependent. In this study, no precipitation was observed for the 10 mg dose strength, and the result of a former study 35 was thus confirmed. As the drug is being released from the formulation as a solution and not as a suspension, these dispersion results may reasonably be valid for other nifedipine IR formulations or other drug compounds with similar dissolution and solubility characteristics. Onset of dispersion of nifedipine from the Adalat R 10 mg IR formulation in grapefruit juice was delayed for approximately 5 min in comparison with the other media, and, as the rate and extent of dissolution is comparable with the other media tested, it is reasonable that the delay in the in vitro dissolution derives from a delay in the capsule disintegration. However, the impact of this delay on the in vivo performance of the formulation is unclear. In the simulations, it is assumed that this delay does not alter the in vivo pharmacokinetics of the drug significantly, and this assumption has been confirmed with an additional simulation in which no substantial change was observed after simulating an additional disintegration lag-time of 5 min (results not shown). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013 PBPK Modeling As described by Rashid et al., 12 oral administration of 200 ml of double strength grapefruit juice twice before intravenous administration of 2.5 mg nifedipine did not alter intravenous nifedipine pharmacokinetics significantly, compared to administration without grapefruit juice. By contrast, ingestion of grapefruit juice before oral nifedipine leads to an increase in nifedipine AUC by approximately 51% and additionally to a delay in t max. 7,12 Figure 4 shows the simulated results in comparison with the results observed in vivo. As shown by the intravenous simulations, nifedipine distribution, metabolism, and elimination could be described accurately using the simulation software (Fig. 4a), and the percent difference between the observed plasma profiles in vivo and the simulated ones (f 1 factor) is 9.6% for both simulations (with and without concomitant ingestion of grapefruit juice). The k cat calculated from the intravenous data (PBPK Model section)was41s 1, irrespective of concomitant grapefruit juice ingestion. Even though nifedipine is distributed into intestinal tissue and is thus theoretically subject to metabolism in the gut wall after intravenous administration, the fraction of drug which reaches the intestinal mucosa via the systemic circulation is too low to enable detection of changes in intestinal metabolism and influence the drug s pharmacokinetics significantly (Fig. 4a). The intravenous model was subsequently used for simulating oral nifedipine plasma profiles after administration of one Adalat R 10 mg IR soft gelatine capsule with concomitant administration of either 200 ml of water or two times 200 ml of double strength grapefruit juice. 12 The simulations show that nifedipine plasma profiles following oral administration of the IR formulation without grapefruit juice slightly overestimate the in vivo observed c max and AUC; however, both values are still within the 80% 125% range (119.2% for c max and 109.2% for AUC, respectively; Fig. 4b). The f 1 factor for this comparison is 12.4% and thus 20%. For the simulations of nifedipine exposure after simultaneous administration of grapefruit juice, inhibition of hepatic CYP 3A was considered to be negligible. 5,9,10,12 As short-term ingestion of grapefruit juice ingestion solely affects intestinal first pass metabolism, the small intestinal and hepatic k cat DOI /jps

9 UTILIZING IN VITRO AND PBPK TOOLS TO LINK ADME CHARACTERISTICS TO PLASMA PROFILES 3213 Figure 4. Comparison between the in vivo ( : with grapefruit juice; : without grapefruit juice) and the simulated (pale gray lines) nifedipine plasma profiles. (a) Administration of 2.5 mg nifedipine intravenously. (b) Administration of one Adalat R 10 mg IR soft gelatine capsule orally. The in vivo data 12 is presented as mean ± SD. The inserts display the same plot on a semilogarithmic scale. values as well as the hepatic CYP 3A concentration were not changed, whereas the small intestinal CYP 3A concentrations following grapefruit juice ingestion were fitted to match the AUC observed in vivo. According to the in silico simulations, a reduction of small intestinal CYP 3A levels (which are provided in Table 2) of 60% was found to yield the best simulation result for a quantitative prediction of the in vivo performance of the drug, and this reduction is in line with data reported in the literature. 9,13,24,25 Schmiedlin-Ren and co-workers 13 reported a decrease in small intestinal CYP 3A concentration of between 45% and 56%, and other researchers have reported a decrease up to 62%, 9 73%, 25 and 100%, 24 respectively. In addition to small intestinal protein degradation, a delay in gastric emptying 5,11 of 30 min because of the ph and caloric content of grapefruit juice was assumed (Fig. 4b). All other parameters remained unchanged and were the default values of the software. The simulated profile accurately matches the in vivo plasma profile, and the f 1 factor for this comparison is 11.0%. The point estimates for the simulation are 102.2% (c max ) and 103.4% (AUC), respectively. Sensitivity Analysis and the Impact of Formulation Changes on the In Vivo Plasma Profile In vivo, intestinal and hepatic CYP 3A content and activity are known to be highly variable among different subjects. To estimate the extent of small intestinal first pass metabolism in subjects having high or low small intestinal CYP 3A concentrations, the impact of a 10-fold variation in small intestinal CYP 3A DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013

10 3214 WAGNER ET AL. Figure 6. Observed in vivo 12 ( ± SD) and simulated nifedipine plasma profiles with CYP inhibition extents of 25%, 50%, 60%, 75%, and 100%. Figure 5. Panel a shows the same data as Figure 4b. In addition, it shows the effect of variations in hepatic and small intestinal CYP 3A concentrations on the simulated profiles. Light gray lines indicate a 10-fold variation in hepatic (- -) or small intestinal ( ) CYP 3A concentrations, compared with mean hepatic and small intestinal CYP 3A concentrations. The simulated profiles from Figure 4b are plotted as dark gray lines. Panel b: Comparison of changes in gastric emptying alone ( ) and 60% CYP 3A inhibition alone (- -) on the simulated plasma profile of nifedipine. For comparison, in vivo plasma profiles both with ( ± SD) and without ( ± SD) concomitant ingestion of grapefruit juice 12 are also plotted in both panels. concentrations on the nifedipine plasma profile was simulated (see Fig. 5a, solid light gray lines). Likewise, to estimate the extent of hepatic metabolism in subjects having high or low small hepatic CYP 3A concentrations, the impact of a 10-fold variation in hepatic CYP 3A concentrations on the nifedipine plasma profile was simulated (see Fig. 5a, broken light gray lines). 4 Other physiological variations, such as variations in the intestinal surface area, were not included in the simulation. According to the simulations, variations in both small intestinal and hepatic CYP 3A enzyme concentrations have a large impact on the plasma profiles. As expected, if both the small intestinal and hepatic concentrations of CYP 3A enzymes are allowed to vary 10-fold, the resulting variability in the simulated plasma profiles is even bigger (data not shown). Note that following intravenous administration of nifedipine, variations in intestinal CYP 3A activity JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013 do not influence the simulated nifedipine pharmacokinetics significantly; by contrast, variation in hepatic metabolic activity has a strong influence on the simulated intravenous pharmacokinetics of nifedipine (data not shown). Figure 5b shows a stepwise parameter modification of gastric emptying and CYP 3A inhibition. Although a delay in gastric emptying alone only affects the shape of the simulated profile, CYP 3A inhibition alone only leads to an increase in the simulated AUC. The simulations thus show that the in vivo nifedipine plasma profile can only be simulated well if both parameters, gastric emptying and CYP 3A inhibition, are considered. It is reasonable that the extent of CYP inhibition depends, on the one hand, on the amount of active constituents in the grapefruit juice, and, on the other hand, on the intestinal CYP 3A concentrations. 2,25,73,80 To estimate the impact of variations in the extent on CYP 3A inhibition on the bioperformance of nifedipine, nifedipine plasma profiles with different extents of CYP 3A inhibition were simulated, using extents of CYP inhibition of 25%, 50%, 75%, and 100%. The results of these simulations are displayed in Figure 6. As shown by the simulations (Fig. 6), a 25%, 50%, 60%, and 75% reduction of small intestinal CYP 3A concentration leads to an approximately 7%, 26%, 51%, and 71% increase of nifedipine AUC, compared with administration without grapefruit juice. Complete inhibition of small intestinal first pass metabolism (100% reduction of small intestinal CYP 3A concentrations) leads to an increase in the simulated AUC by approximately 116%. Compared with the inherent variability in CYP 3A concentrations, enzyme inhibition caused by grapefruit juice ingestion has a more modest impact on the pharmacokinetics of nifedipine (Fig. 5a and Fig. 6). Of course, the relative effect of grapefruit juice would depend on the CYP 3A concentration in the individual. If there is a DOI /jps

11 UTILIZING IN VITRO AND PBPK TOOLS TO LINK ADME CHARACTERISTICS TO PLASMA PROFILES 3215 Figure 7. Simulated virtual nifedipine plasma profiles in comparison with the in vivo observed data 12 ( ± SD). The insert depicts the virtual dissolution profiles that were used to generate the simulated plasma profiles. The dashed line (- -) indicates a dissolution extent of 90%. higher individual CYP 3A concentration, the effect of grapefruit juice is likely to be greater than if there is a lower individual CYP 3A concentration. For felodipine, another calcium channel-blocking agent which is metabolized via CYP 3A enzymes, complete absorption in the gastrointestinal tract has been observed, but the drug s absolute bioavailability ranges from 4% to 36% because of the very high presystemic clearance. 6 It is reported that presystemic, intestinal first pass metabolism accounts for approximately 70% of felodipine s metabolism. Following grapefruit juice ingestion, the extent of small intestinal first pass metabolism decreases to only 10%, and the drug s AUC increases by a factor of 3 to 5. 6 In the case of nifedipine, small intestinal first pass extraction and the change in the drug s bioavailability, associated with grapefruit juice ingestion, is thus less pronounced than for felodipine. For nifedipine, grapefruit juice ingestion leads to an approximately 51% increase in nifedipine AUC, 12 and this result could be simulated well with an assumed small intestinal CYP 3A decrease of 60%, which is very close to the 62% reduction of small intestinal CYP 3A concentrations reported in the literature. 9 However, it should be mentioned that these results are only valid for the selected CYP 3A concentrations, 72,73 and the results may be quantitatively different in a population with other CYP 3A concentrations. Besides first pass metabolism, drug release from the formulation and subsequent drug dissolution is a key factor in the in vivo performance of nifedipine. The IR capsule tested in this study releases the drug as a solution, and the drug in solution is stable and does not tend to precipitate at the dose used in this study. Under the assumption that the capsule disintegration time is short, the formulation behaves like an oral solution in vivo. However, for formulations with markedly longer release times than the one tested in this study, dissolution could become a critical parameter in the in vivo performance. In addition, variations of critical process parameters, scale-up of the manufacturing process, or changes in the formulation may be critical for the in vivo performance of the drug and formulation. To estimate the influence of changes in the formulation s release and dissolution characteristics on the drug s pharmacokinetics, which might be evoked by changes in the formulation or in the manufacturing process, additional simulations of nifedipine s pharmacokinetics were performed. For this purpose, the in vivo dissolution characteristics of the formulation were varied, and virtual dissolution profiles with 90% dissolution times of 5, 10, 15, 20, 25, 30, 45, 60, 90, 120, and 180 min were used as input parameters. Figure 7 shows the virtual dissolution profiles with 90% dissolution times between 5 and 180 min as well as the corresponding simulated plasma profiles. In addition, Table 3 presents the statistical analysis of these simulations. The simulations show that the release rate of the drug in vivo is a critical parameter for the bioperformance of the nifedipine formulation, in addition to the physiological factors addressed above. The simulated profiles and f 1 calculations show that in vivo drug release and dissolution must take place within DOI /jps JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013

12 3216 WAGNER ET AL. Table 3. Statistical Comparison Between the In Vivo Observed 12 and the In Silico Simulated Nifedipine Plasma Profiles Using Dissolution Profiles with Various Dissolution Characteristics as Input Function. Statistical Characteristics a 90% Dissolution Time (min) b f 1 factor (%) Ratio AUC c (%) Ratio c max c (%) For the comparison, only the time points corresponding to the sampling times in vivo were used for the calculations. To conclude that the simulations are equivalent to the in vivo profile, the ratios for the simulated and the in vivo observed AUCs must be in the range of 80% and 125%, and the f 1 factor has to be 20%. a Bold numbers indicate that the according equivalence criterion is not met. b Virtual 90% dissolution times (except 8.9 min, which derives from the in vitro dissolution test experiments in FaSSIF v2). c Ratios: AUC (predicted)/auc (in vivo) andc max (predicted)/c max (in vivo). a very narrow time frame (approximately 10 min) to achieve the target pharmacokinetic profile. Point estimates of AUC and c max appear to be less sensitive than f 1 calculations for this purpose (Table 3). It should be mentioned that these results are specific to the tested nifedipine IR formulation (Adalat R 10 mg) and to the in vivo study performed by Rashid et al. 12 Other drug compounds and formulations might provide other results according to the dissolution window that is needed to generate equivalent dissolution profiles. Nevertheless, the results point the way forward for utilizing PBPK modeling in identifying formulation safe spaces during development of new drug products. CONCLUSIONS The nifedipine IR formulation tested in this work showed very robust dissolution behavior in all media, and, for predicting the in vivo behavior of the formulation, the biorelevant media did not provide any significant advantage compared with compendial media. The main focus of the work was, however, to implement a detailed description of nifedipine first pass metabolism into a PBPK model to describe oral nifedipine pharmacokinetics. In a subsequent step, inhibition of CYP 3A enzymes, induced by grapefruit juice, and its impact on oral nifedipine pharmacokinetics was simulated quantitatively, and the contribution of variations in the content (and thus activity) of intestinal and hepatic CYP 3A on overall nifedipine in vivo performance was evaluated. The simulation results show that variations in the extent of nifedipine intestinal first pass metabolism, caused by highly variable interindividual CYP 3A concentrations, are the most likely explanation for the high variability of the drug s pharmacokinetics. Grapefruit juice leads to a marked reduction in small intestinal CYP 3A concentrations, 5,6,14 and a reduction of small intestinal CYP 3A concentrations by 60% yielded the best modeling results. This reduction is in line with results from the literature. 9,13,24,25 In a further, exploratory in vitro in silico approach, a link between the in vitro dissolution characteristics and the in vivo performance of the formulation was created, and an in vitro dissolution design space was established. Under the assumption that changes in the formulation can be detected using in vitro dissolution experiments, this exploratory approach can be usefully applied in the context of the quality by design (QbD) paradigm to simulate the impact of changes in the formulation on the bioperformance of the drug and formulation, and an in vitro dissolution design space, for example for reformulation of existing products and perhaps even for development of generic drug products could be implemented. In the PBPK model developed for nifedipine, in vitro dissolution characteristics of the formulation, drug characteristics, and potential interactions such as the effect of grapefruit juice on the drug s pharmacokinetics, were incorporated to simulate in vivo drug concentration time profiles of the drug. Such PBPK models can be easily generated for simulations of other drug compounds, enzyme systems, and for co-administration of enzyme inhibitors and inducers. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 102, NO. 9, SEPTEMBER 2013 DOI /jps