Human in Vivo Regional Intestinal Permeability - ACS Publications

Apr 28, 2015 - specific in vivo intestinal effective permeability (Peff) in humans. Plasma concentration−time profiles from 15 clinical studies that...
7 downloads 0 Views 869KB Size
Download PDF
Article pubs.acs.org/molecularpharmaceutics

Human in Vivo Regional Intestinal Permeability: Quantitation Using Site-Specific Drug Absorption Data Erik Sjögren,* David Dahlgren, Carl Roos, and Hans Lennernas̈ Department of Pharmacy, Biopharmaceutic Research Group, Uppsala University, SE-751 23 Uppsala, Sweden S Supporting Information *

ABSTRACT: Application of information on regional intestinal permeability has been identified as a key aspect of successful pharmaceutical product development. This study presents the results and evaluation of an approach for the indirect estimation of sitespecific in vivo intestinal effective permeability (Peff) in humans. Plasma concentration−time profiles from 15 clinical studies that administered drug solutions to specific intestinal regions were collected and analyzed. The intestinal absorption rate for each drug was acquired by deconvolution, using historical intravenous data as reference, and used with the intestinal surface area and the dose remaining in the lumen to estimate the Peff. Forty-three new Peff values were estimated (15 from the proximal small intestine, 11 from the distal small intestine, and 17 from the large intestine) for 14 active pharmaceutical ingredients representing a wide range of biopharmaceutical properties. A good correlation (r2 = 0.96, slope = 1.24, intercept = 0.030) was established between these indirect jejunal Peff estimates and jejunal Peff measurements determined directly using the single-pass perfusion double balloon technique. On average, Peff estimates from the distal small intestine and large intestine were 90% and 40%, respectively, of those from the proximal small intestine. These results support the use of the evaluated deconvolution method for indirectly estimating regional intestinal Peff in humans. This study presents the first comprehensive data set of estimated human regional intestinal permeability values for a range of drugs. These biopharmaceutical data can be used to improve the accuracy of gastrointestinal absorption predictions used in drug development decision-making. KEYWORDS: human in vivo intestinal permeability, colon, site-specific permeability, regional permeability

■

INTRODUCTION The systemic bioavailability (F) of the active pharmaceutical ingredient (API) in an orally administered drug product is generally required to be sufficiently high to produce a plasma concentration−time profile that will result in the desired pharmacological response. To accomplish this, it is necessary for the API to be adequately absorbed in the intestinal tract. The rate and extent of intestinal absorption are directly related to the biopharmaceutical property intestinal effective permeability (Peff).1 Regional differences in drug absorption along the gastrointestinal (GI) tract, especially regional intestinal permeability, have direct implications on the development of modified-release (MR) drug dosage forms.2−9 To date, the only substantial in vivo Peff information available is for the human jejunum; there remains a shortage of in vivo determinations for the distal parts of the human intestinal tract, especially for the colon.5,10,11 A better understanding of differences in the physiology and Peff along the intestine would improve predictions of total intestinal drug absorption. This is particularly important for drugs that are incompletely absorbed in the proximal small intestine (P-SI), such as APIs with low permeability [Biopharmaceutics Classification System (BCS) III and IV] or those with poor solubility/dissolution character© 2015 American Chemical Society

istics (BCS II and IV), and drugs intended for MR dosage forms. Improved information on the regional intestinal Peff for these drugs would be useful for assessing their site-specific absorption in relation to intestinal transit and the implications for product development. Several ongoing collaborative efforts among academia, the industry, and the regulatory agencies aim to facilitate the drug development and regulatory processes.12 One such interdisciplinary collaboration is the OrBiTo project, which addresses the need to improve the accuracy of predictive in vitro and in silico GI absorption tools and models.13,14 Improvements to existing models and the development of novel tools require additional accurate human intestinal Peff data, in particular from the distal small intestine (D-SI) and the large intestine (LI).15,16 The epithelial area/luminal volume ratio is higher in the small intestine (SI) than in the colon.17,18 Intraluminal transport rates (diffusion and convection) are also faster in the SI, as a result of the narrower intraluminal space, stronger motility, and higher Received: Revised: Accepted: Published: 2026

December 13, 2014 April 16, 2015 April 28, 2015 April 28, 2015 DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics

first pass extraction (Ffirstpass) was described as Ffirstpass= (1 − EG)(1 − EH). Deconvolution Procedure and Calculation of Peff. The rate of intestinal absorption (amount time−1) was acquired by deconvolution of plasma concentration−time data after regional administration of a single dose of the drug and use of the model fit to the IV PK data as unit impulse response. The deconvolution procedure has been used for several decades in the pharmaceutical field for estimation of in vivo drug absorption and drug release.26,27 On the basis of linear system analysis and the superposition principle, an input function (e.g., drug release or absorption) can be deconvoluted from the physiological response data (e.g., plasma concentration−time data) using the units impulse response.26,27 The deconvolution process in this study was carried out using WinNonlin software (Phoenix WinNonlin v6.3, Certara, L.P., St. Louis, MO, USA). Plasma concentration data obtained for up to 6 h after dose administration were analyzed using the automatic smoothing function included in the software and allowing for absorption at time 0, with an output of 101 data points. Only output data up to 1 h (=17 data points) were used in further calculations relating to the site-specific intestinal Peff in order to minimize the influence of potential changes in the system that would affect the intestinal absorption rate, such as luminal degradation or dispersion of the yet unabsorbed dose. Dispersion of the dose refers in this study to spreading of the fluid, along the intestine, in which the drug is present. The deconvoluted input rate was corrected for Ffirstpass to acquire the final intestinal absorption rate (eq 3).

level of mixing. These regional differences in physiology mean that, with regard to the digestion of a meal, the SI is the main region for absorbing nutrients while the colon mainly acts as a reservoir. Other potential regional intestinal differences of relevance to the Peff are the expression and activity of various absorptive and efflux carrier proteins and the membrane composition, which mainly affects transcellular passive diffusion.9 The aim of this study was to explore the potential for determining the human intestinal Peff for an array of drugs from the results of clinical biopharmaceutical studies in which the drugs were administered as a single dose at different intestinal sites. A deconvolution approach was applied to the reported plasma concentration−time data, and the intestinal Peff was calculated from the acquired results. This study was part of an overall objective to increase general knowledge of segmental intestinal permeability and to improve the accuracy of future GI absorption predictions.

■

MATERIALS AND METHODS Bioavailability, Absorption, and Permeability. F is the fraction of an administered dose that reaches the systemic circulation in an unchanged form.19 This pharmacokinetic (PK) parameter is determined by three key processes: the fraction of the dose absorbed (fabs) from the lumen, the gut wall extraction (EG), and the liver extraction (EH) (eq 1). F = fabs (1 − EG)(1 − E H)

(1)

GI absorption requires the API to be released from the formulation and dissolved as free API monomers in the GI lumen. The monomers can then cross the intestinal membrane by several mass transport mechanisms, such as passive transcellular transport, passive paracellular transport, and/or carrier-mediated (CM) transport in one or both directions.20,21 It is the net transport of the API from the intestinal lumen across the apical intestinal epithelial membrane barrier that is reflected by the intestinal Peff.9,22,23 Pharmacokinetics. Disposition PK parameters (presented with references in Table 2) were estimated from the reported historical concentration−time profiles after intravenous (IV) drug administration, using conventional first-order compartmental kinetics. Reported PK parameters were used if plasma profiles were not presented. There were no human IV data available for three APIs (bevirimat, fenofibric acid, and ipsapirone). This is clearly indicated in the compound section, together with information on the strategy for adjusting the calculations of the unit impulse response. Hepatic plasma clearance (CLH) was set equal to total plasma clearance (CL) when information on renal plasma clearance (CLR) was missing. CL was compensated for CLR for APIs with reported renal elimination to acquire CLH, assuming negligible influence from other elimination pathways. EH was then estimated from CLH and hepatic blood flow (QH, 1.62 L min−1)24 using eq 2, assuming that the concentration of drug in the blood equates to the drug concentration in plasma.25 EH =

CL H QH

absorption rate =

input rate Ffirstpass

(3)

This correction of the absorption rate also increases the total fabs to an equivalent degree. However, fabs was never allowed to exceed 1 as a consequence of the Ffirstpass correction. If that was to be the case, Ffirstpass was adjusted, so that a fabs equal to 1 was attained after correction. Output data after an estimated fabs of >0.90 were discarded to avoid division by very small numbers in further calculations. The amount of unabsorbed API remaining in the intestinal lumen was calculated according to eq 4. API in lumen = dose − Σ absorbed API

(4)

The accumulated amount absorbed was obtained from the deconvolution of plasma concentration−time data following correction for Ffirstpass. The luminal API concentration (Clumen) was then calculated by dividing the amount of API remaining in the lumen with the total volume of liquid that was administered at the dosing event (Vdose), i.e., the volume of the dose and the subsequent flushes, assuming instantaneous and homogeneous distribution of API in the Vdose at all times (eq 5). C lumen =

API in lumen Vdose

(5)

The corresponding intestinal absorption clearance rate (CLabs) was determined by dividing the local intestinal absorption rate by Clumen (eq 6).

(2)

CLabs =

EG was considered for compounds where the data could be attained (for details the reader is referred to the Investigated Compounds section). EG in the LI was assumed to be negligible if not specifically stated otherwise. The fraction escaping total

absorption rate C lumen

(6)

Finally, the site-specific Peff was calculated by dividing CLabs by the absorptive surface area (eq 7). 2027

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics

Table 1. Physicochemical and Biopharmaceutical Properties of the Investigated Active Pharmaceutical Ingredients (API)a f uncharged API bevirimat budesonide cyclosporin fenofibric acid fexofenadine ipsapirone lisdexamfetamine lumiracoxib metoprolol nifedipine ranitidine rivastigmine sumatriptan theophylline

MW

pKa

clogP

CYP3A4

P-gp

Vdose (mL)

585 430 1190 319 538 401 263 294 267 346 314 250 295 180

b

7.2 2.3 4.3 5.0 4.8 1.6 1.1 4.2 1.8 2.2 0.75 2.3 1.1 −0.21

N Y Y N N Y N N N Y N N N N

N Y Y

1 1 16.5d/11.5e 1 7.5 0.7 0.75 0.75 3.5 0.7 14f/0.6g 20 14.2 0.81d/0.90e

4.2 N N 2.9b 7.8c, 4.2b 7.1c 8.4c 4.7b 9.7c N 8.4c 8.9c 9.5c, 11b 8.4b

Y N N N Y N

BCS/BDDCS I/I II/II I/I III/III

II/ I/I II/II III/III I/I III/I I/I

pH 5.5

pH 6.5

pH 7

0.048 1 1 0.0025 0.95 0.022 0.001 0.14 0.00021 1 0.0013 0.00040 0.000093 1

0.0050 1 1 0.00025 0.95 0.19 0.01 0.016 0.0021 1 0.012 0.0040 0.00091 0.95

0.0016 1 1 0.000079 0.87 0.42 0.04 0.0050 0.0066 1 0.038 0.012 0.0029 0.86

a

BCS/BDDCS = Classification according to the Biopharmaceutics Classification System/Biopharmaceutics Drug Disposition Classification System;5,100 clogP = calculated n-octanol−water coefficient; CYP3A4 = substrate for cytochrome P450 3A4; f uncharged = fraction of uncharged API at specified pH; MW = molecular weight; P-gp = substrate for P-glycoprotein; pKa = dissociation constant; Vdose = volume of administered dose. bAcid pKa. cBase pKa. dSmall-intestinal dose. eLarge-intestinal dose. fIntubation study. gCapsule study.

Table 2. Pharmacokinetic (PK) Parameters, Hepatic Clearance (CLH), Renal Clearance (CLR), and Fraction Escaping First Pass Extraction for the Investigated Active Pharmaceutical Ingredients (API) Applied as Unit Impulse Response for the Deconvolution and Calculations of Absorption Ratesa fraction escaping first pass extraction API

A/dose (ng/mL/mg)

α (1/h)

bevirimatb,31 budesonide32 cyclosporin35 fenofibric acidb,37 fexofenadine39 ipsapironeb,41,42 lisdexamfetamine43−45 lumiracoxib47,48 metoprolol7,50 nifedipine52 ranitidine56 rivastigmine58,59 sumatriptan61 theophylline64

0.14 29 34 56 58 1100 69 20 12 62 38 26 70 69

0.043 13 1.3 2.5 3.1 6.9 18 0.44 9.7 8.4 6.5 12 19 6.0

B/dose (ng/mL/mg)

β (1/h)

4.0 3.1 67 16 150 16 3.3 3.4 8.6 7.8 6.9 9.0 42

0.31 0.11 0.24 0.49 0.39 2.2 0.11 0.20 0.55 0.40 0.58 3.2 0.10

C/dose (ng/mL/mg)

γ (1/h)

34 3.0

0.04 0.10

0.39

0.31

2.4

0.35

CLH (l/h)

CLR (l/h)

67 13

0 0

9.8

6.2

7.7 48 43 7.6

0 5.4 0 31

59 2.4

15 0

P-SI

D-SI

LI

1 0.31 0.40 0.88 0.90 0.10 0.062 0.90 0.50 0.56 0.92 0.44 0.24 0.98

1 0.31 0.40 0.88 0.90 0.10 0.082 0.99 0.50 0.56 0.92 0.49 0.24 0.98

1 0.31 0.87 0.88 0.90 0.10 0.11 0.92 0.50 0.56 0.92 0.52 0.24 0.98

a References used for each API are indicated in the table. For further details the reader is referred to the Investigated Compounds section. bPK parameters not estimated from human intravenous data. P-SI = proximal small intestine, D-SI = distal small intestine, LI = large intestine.

Peff =

CLabs surface area

analyses, the Peff estimates from the D-SI and the LI were normalized to the respective P-SI value. When no P-SI value was available (only theophylline), the LI value was normalized to the D-SI value. Alternatively, when adopting the same assumptions, eq 8 can be used to calculate the site-specific Peff instead of the stepwise approach described above (eqs 5−7).

(7)

The surface area was estimated as the lateral surface area of a geometric tube with the same volume as Vdose, assuming that the shape of the Vdose in the intestine could continuously be described by this geometry during the time of analysis (1 h). The radii of the duodenal, jejunal, ileal, and colon intestinal segments used in the calculations were 1.75, 1.75, 1.5, and 2 cm, respectively.28 Thus, a site-specific Peff−time profile up to 1 h was acquired for each data set analyzed. This time period was chosen based on the objective to obtain a site-specific Peff. Time-point estimates of Peff equal to zero were excluded, and to minimize the influence of outliers, the median value from the acquired Peff−time profile was considered to be a representative value of the overall permeability. To enable general trend

Peff =

absorption rate × radius API in lumen × 2

(8)

This equation is somewhat more straightforward and will generate the same result, but it is more difficult to interpret. However, it clearly illustrates that the volume of the lumen is not relevant when adopting these calculations for the estimation of in vivo relevant Peff. This can be explained by 2028

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics that CLabs (unit: volume time−1) is estimated by dividing the absorption rate (unit: amount time−1) by the luminal concentration (unit: amount volume−1) (eq 6). Peff is then obtained by correcting for the surface area. As the area can be described by its relationship to volume (i.e., tube area = tube volume × 2/radius), volume can be canceled out in the calculations. Investigated Compounds. Selection Details. The relevant physicochemical and biopharmaceutical properties as well as classification according to BCS and the Biopharmaceutics Drug Disposition Classification System (BDDCS) of the investigated APIs are summarized in Table 1. PK parameters used for calculations are displayed in Table 2. Only clinical studies reporting systemic concentration−time profiles after site-specific single-dose administration of the API as a solution were selected in this study to estimate the regional intestinal permeability. This definitive criterion limited the number of available clinical studies. Deconvolution analysis should be performed using absorption and IV data from the same individual to minimize bias, but such studies were not found in the literature. One optimal set of data for the adopted approach would be individual absorption data plus impulse response data obtained concomitantly via an IV microtracer. In this study, however, we used average or median absorption and IV data from different studies under the assumption that all data were representative of the population. Only PK-data from singledose studies were applied. Compartmental PK analysis, with up to three compartments, was adopted, and the best model was selected based on the Aikaike information criterion and precision (CV%) in parameter estimates. Some extra APIs were included despite there being no human IV data (bevirimat, fenofibric acid, and ipsapirone) in order to increase the number of investigated compounds. The absolute permeability values for these three compounds are consequently less reliable. However, these estimates are still valid for the purpose of comparing Peff between the intestinal regions. Bevirimat (Protease Inhibitor, Anti HIV-Drug). Absorption data after administration of bevirimat in the jejunum, ileum, and colon were available from the literature.29 The absorption study used a remote-controlled capsule technique, which facilitated site specific administration of 25 mg of bevirimat in solution (10% hydroxyl propyl-β-cyclodextrin). The volume administered was not specified in the report and was therefore assumed to be 1 mL, as this is the volume capacity of the capsule’s drug reservoir. Because no IV data for bevirimat were found in the literature, PK parameters were estimated from the concentration−time profiles after ileal administration (highest relative F) adopting first-order absorption and one-compartment systemic kinetics. The absolute values of Peff should be considered less reliable, as it was not possible to compensate for first-pass elimination in the estimations, especially since bevirimat is mainly eliminated through hepatobiliary routes.30 Budesonide (Glucocorticoid Steroid, Crohn’s Disease Treatment). Plasma concentration−time profiles after regional intestinal administration of budesonide to the jejunum, ileum, and colon were retrieved from the literature.31 Single doses of 3 mg of budesonide aqueous solutions (3 mg/mL) were administered by GI intubation. The data for this investigation were acquired after pretreatment with a single low dose of ketoconazole (8 mg) to only inhibit the intestinal first pass extraction. The systemic PK of budesonide were unaffected by the single dose (8 mg) of ketoconazole. Deconvolution was

performed using systemic PK parameters estimated from the concentration−time plasma profiles after a 5 min IV infusion of 0.1 mg of budesonide.32 Information on colonic stability was acquired from published in vitro faecal degradation data.33 Cyclosporin (Cyclic Peptide, Immunosuppressor). Whole blood concentrations after local instillation of 150 mg of cyclosporin in emulsion to the duodenum, jejunum, ileum, and colon were obtained from the literature.34 The emulsion consisted of polyethylene glycol hydrogenated castor oil, middle chain triglycerides, and low molecular weight glycols and was administered by intubation using a modified onelumen GI tube. The volume administered was unspecified in the report. However, as the same dose was also administered in three capsules, the volume of the emulsion was assumed to be 1.5 mL (adopting a common capsule size of 0.5 mL). In addition, saline solution was used to rinse the drug ampules and the tube (1.5 mL + 15 mL and 1.5 mL + 10 mL for the SI and LI, respectively). The PK parameters were estimated from data from an interaction study with grapefruit juice.35 SI EG, which is mediated by cytochrome P450 (CYP)3A4, was reported to be approximately 0.68, assuming an fabs of 0.9. However, a maximum value for EG of 0.54 was used in this study, to ensure that fabs did not exceed 1. Final SI and LI values for F of 0.40 and 0.84 were adopted. A direct reference value for the jejunal Peff for cyclosporin (1.65 ± 0.53 × 10−4 cm s−1), measured using the double balloon technique, was obtained from the literature.36 Fenofibric Acid (Peroxisome Proliferator-Activated Receptor Activator, Cholesterol and Lipid Modulation). GI absorption of fenofibric acid administered as a nanocrystal dispersion suspension was studied using remote-controlled capsule system.37 It was assumed that the suspension was immediately dissolved in the dose volume (1 mL) as the total concentration of the suspension (130 mg/mL) was less than the reported solubility of fenofibric acid (300 mg/mL).37 Plasma concentration−time data were available for administration of fenofibric acid to the P-SI, D-SI, and LI. However, no data from IV administration were reported. The PK parameters were therefore estimated from the P-SI plasma profile, compensating for the reported value for absolute F of 0.879, assuming complete absorption. Absolute Peff values should be viewed with some caution because of the lack of IV impulse response function data. Fexofenadine (Antihistamine, Allergy Symptoms Treatment). Plasma concentration−time profiles after a single dose of 60 mg fexofenadine to the jejunum, ileum, and colon were collected from the literature.38 Although individual absorption profiles were accessible, median plasma data were used for the estimations because of the absence of individual IV PK information. Fexofenadine was delivered in solution (4 mL dose + 3.5 mL flush) to the intestine using aGI intubation technique. Deconvolution was carried out using the impulse response data acquired from an IV microdose study with simultaneous oral administration.39 The absorption calculations included compensation for an EH of 0.10, estimated from reported clearances (CLH = CL − CLR). A reference value (0.07 ± 0.07 × 10−4 cm s−1) for the jejunal Peff determined using the double balloon technique was obtained from the literature.11,40 Ipsapirone (Seratonin Receptor Agonist, Antidepressant). Absorption data after a single dose of 4.62 mg of ipsapirone in aqueous solution (0.7 mL) to an unspecified region of the SI as well as to the proximal (ascendens) and distal (transversum or 2029

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics

concentrations of polyethylene glycol and polysorbate. Deconvolution was carried out using individual data; however, the final calculations of Peff used average data because of the low number of subjects (n = 4), adopting a Ffirstpass of 0.46. Some caution with regard to the absolute values is advised since individual day-to-day variability has been reported for nifedipine as well as a pronounced risk of considerable firstpass metabolism in the gut, mediated by CYP3A4.52,53 Ranitidine (H2-Receptor Antagonist, Treatment of Peptic Ulcer). Absorption data after a single dose of ranitidine to different regions of the intestine were obtained from two studies. In the first study, 150 mg doses of ranitidine in solution (6 mL dose + 8 mL flush) were administered via a nasoenteric tube to the jejunum and the cecum.54 A remote-controlled capsule technique was used in the second study to deliver 121 mg doses of ranitidine in solution (0.6 mL) to the jejunum, ileum, and colon.55 Parameters for the impulse response function were acquired from a single-dose human IV PK study.56 A reference measurement (0.27 ± 0.25 × 10−4 cm s−1) for the jejunal Peff using the double balloon technique was obtained from the literature.11,57 Rivastigmine (Acetylcholinesterase Inhibitor, Treatment of Dementia). Plasma concentration−time profiles after instillation of 3 mg doses of rivastigmine in solution (10 mL + 10 mL flush) were obtained from the literature.58 The drug was administered via a naso-intestinal tube to the jejunum, ileum, and ascending colon. PK parameters were estimated from clinical plasma concentrations after administration of 2 mg of rivastigmine as an IV infusion over 1 h.59 Rivastigmine is completely absorbed along the GI tract. It is also subject to regional nonhepatic first-pass extraction.58 The absorption parameters were compensated for this to acquire complete absorption in all segments over 6 h. The Ffirstpass values adopted in the calculations of Peff were 0.435, 0.49, and 0.52 for jejunum, ileum, and ascending colon, respectively. Sumatriptan (Serotonin Agonist, Migraine Treatment). Estimation of the absorption rates of sumatriptan in the jejunum and cecum was made using the available plasma concentration−time profiles.60 A sodium chloride solution containing 50 mg of sumatriptan (4.2 mL dose + 10 mL flush) was instilled into the intestine via an oroenteric tube. Deconvolution used the impulse response data calculated after administration of 3 mg of sumatriptan as a 15 min IV infusion. 61 Calculation of the P eff was done with a compensation for a Ffirstpass of 0.24. Potential gut metabolism mediated by monoamine oxidase A was not accounted for since no direct measurements of first-pass gut extraction were available.60,62 Theophylline (Xanthine Derivate, Respiratory Disease Therapy). Concentration−time profiles in serum after administration of theophylline to different locations in the terminal ileum and colon were obtained from the literature.63 Sitespecific delivery of a 104 mg/mL theophylline solution containing salicylamide sodium salt as a solubility mediator was achieved using the high-frequency capsule technique. Mean data were used as the number of individuals in the available study was low (n = 3). The colonic data were analyzed together for the same reason, although different locations were specified. PK parameters were estimated from the serum theophylline concentrations after a 20 min IV infusion of 351 mg of theophylline.64 The calculations were compensated for a Ffirstpass of 0.98.

descendens) parts of the colon were collected from the literature.41 The drug was administered using the high frequency capsule technique. PK data from rhesus monkeys were used in combination with terminal half-life data from the absorption study for the PK calculations since there were no human IV PK data in the literature.42 As substantial first-pass metabolism has been reported in dogs, all calculations were also compensated for a Ffirstpass of 0.1.41 The lack of human IV PK input means that the ipsapirone Peff values should be considered less reliable. Lisdexamfetamine Dimesylate (Psychostimulant Prodrug, Attention Deficit Hyperactivity Disorder Treatment). Plasma concentration−time profiles after single 50 mg doses of lisdexamfetamine dimesylate in aqueous solution to the P-SI, D-SI, and LI (ascending colon) were obtained from the literature.43 Regional administration of the drug (dissolved in an assumed volume of 0.75 mL) was accomplished using a remote-controlled capsule technique. Human PK parameters were estimated following IV administration of 50 mg of lisdexamfetamine dimesylate.44 Lisdexamfetamine dimesylate is rapidly hydrolyzed in multiple organs, e.g., blood and kidney, to the active metabolite d-amphetamine.45 The reported exposure to d-amphetamine was higher after regional intestinal administration to the P-SI (AUC0−24h = 755.1 ng h mL−1) and D-SI (AUC0−24h = 731.8 ng h mL−1) than after IV administration (AUC0−24h = 562.7 ng h mL−1).43,44 Based on these data, the fabs in the SI segments was assumed to be 1; the fabs in the LI was assumed to be 0.66 since this was the approximate relative d-amphetamine exposure (AUC0−24h = 489 ng h mL−1) compared to the SI administrations. Lumiracoxib (Cyclo-oxygenase-2 Inhibitor, Anti-inflammatory). Plasma concentration−time profiles after single 100 mg doses of lumiracoxib to the P-SI, D-SI, and LI (ascending colon) were obtained from the literature.46 Lumiracoxib was delivered with a remote-controlled capsule technique as a PEG 4000 solution (0.75 mL). IV PK parameters adopted for the calculations were recalculated from previous reports.47,48 Under the assumption that complete absorption occurred in the SI, the Ffirstpass values adopted for P-SI, D-SI, and LI were 0.90, 0.99, and 0.92, respectively. Metoprolol (β1-Receptor Antagonist, Treatment of Cardiovascular Diseases). Plasma concentration−time profiles after single 25 mg doses of metoprolol tartrate (corresponding to 19.5 mg of metoprolol) in solution to the jejunum, ileum, and colon were obtained from the literature.49 A GI intubation system was used to instil the metoprolol solution into the specific intestinal regions. The administered volume was not specified but was assumed to be a 2.5 mL dose + 1 mL flush. Plasma PK parameters following IV administration in humans were attained from the literature. 50 The permeability calculations were compensated for a Ffirstpass of 0.50 considering that renal excretion accounts for 10% of the total elimination of metoprolol.7 A reference measurement (1.34 ± 1.0 × 10−4 cm s−1) of the jejunal Peff made using the double balloon technique was obtained from the literature.11,51 Nifedipine (Calcium Channel Blocker, Hypertension Treatment). The local absorption of nifedipine from the jejunum and two sites in the colon (ascendens and descendens/transversum) was investigated using the high-frequency capsule system.52 Plasma concentration−time profiles following IV infusion (60 min) of nifedipine were available from the same study. The formulation used for intestinal administration (0.7 mL) was an aqueous solution containing unspecified 2030

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics

Table 3. Estimates of Regional Intestinal Effective Permeability (Peff) and the Corresponding Fraction of the Active Pharmaceutical Ingredient (API) Absorbed at 1 h ( fabs.1h) from the Proximal Small Intestine (P-SI), Distal Small Intestine (DSI), and Large Intestine (LI)a P-SI −4

−1

D-SI −4

−1

LI −4

cm s−1)

API

class

Peff (10

cm s )

fabs.1h

bevirimat budesonide cyclosporin

VH H H

0.79 0.75 0.27

0.21 (0.16−0.23)c 0.59 (0.34−0.84)c 0.0001 (--0.03)c

0.060c 0.17c 0.004c

VH L L

0.89 0.49 0.50 0.56b 0.95 0.12 0.13

3.7 (2.2−4.6) 3.4 (2−4.2) 0.84 (0.59−0.96)

fenofibric acid fexofenadine ipsapirone

5.9 (5.5−7.3) 1.9 (1.2−2.2) 2.2 (1.3−2.8) 2.6 (1.7−2.8)b 8.6 (1.7−16) 0.27 (0.15−0.92) 0.31 (0.22−0.66)

2.3 (0.77−4.6) 0.062 (0.035−0.62)

0.7 0.044

lisdexamfetamine lumiracoxib metoprolol nifedipine

VH VH H VH

3.1 3.7 1.5 4.4

(1.0−3.7) (2.7−6.4) (1.2−2.9) (1.9−11)

0.71 0.81 0.45 0.89

3.3 (1.1−4.1) 7.2 (0.87−19) 2.0 (0.85−2.4)

0.68 1 0.54

ranitidine (tube) ranitidine (capsule) rivastigmine sumatriptan theophylline

L L VH H H

0.39 (0.22−0.59) 0.21 (0.16−0.25) 16 (5.4−26) 1.7 (0.17−1.9)

0.15 0.078 1 0.39

0.11 (0.080−0.30) 12 (1.1−14)

0.054 1

0.92 (0.56−1.8)

0.39

0.48 (0.41−0.72)c 0.025 (0.018−0.59)c 0.33 (0.31−0.51)e 0.42 (0.2−0.74)f 0.11 (0.010−0.15)e 2.5 (1.2−3.3)e 1.6 (0.74−2.4)c 1.2 (0.050−1.6)e 2.0 (0.41−2.1)f 0.071 (0.062−0.074)d 0.060 (0.036−0.23)c 10 (3.4−12)e 0.27 (0.056−0.84)d 0.74 (0.56−2.0)d

0.17c 0.033c 0.12e 0.15f 0.054e 0.58e 0.47c 0.38e 0.51f 0.024d 0.029c 0.95e 0.070d 0.22d

Peff (10

cm s )

fabs.1h

Peff (10

fabs.1h

a

The P-SI refers to the jejunum if not indicated otherwise. The location of administration in the LI is indicated. Peff estimates are reported as medians with the range. bDuodenum. cUnspecified location in colon. dCecum. eColon ascendens. fColon descendens/transversum.

Figure 1. Deconvolution profiles for fexofenadine and metoprolol (upper plots), with the corresponding intestinal effective permeability (Peff)−time profiles (lower plots). Green, red, and blue colors represent the proximal small intestine, distal small intestine, and large intestine, respectively. The filled circles and the dotted lines in the deconvolution profiles represent observations and the response function, respectively. In the lower plots, the solid and dotted lines represent the Peff and the fraction absorbed ( fabs) over time, respectively.

■

LI, respectively. The adopted approach generates a Peff−time profile for each analyzed API and intestinal location. No general trends were found in the shapes of the acquired Peff−time profiles for the 14 investigated APIs (see Supporting Information). Graphs showing the deconvoluted concentration−time profiles and the corresponding Peff−time profiles for two APIs (fexofenadine and metoprolol) are displayed in

RESULTS

General Results. Peff estimates for the P-SI, D-SI, and LI are summarized in Table 3. Forty-three regional Peff values were estimated, comprising 15, 11, and 17 estimates from the P-SI, D-SI, and LI, respectively. The Peff estimates ranged from 0.21 × 10−4 to 16 × 10−4 cm s−1, 0.062 × 10−4 to 12 × 10−4 cm s−1, and 0.0001 × 10−4 to 10 × 10−4 cm s−1 in the P-SI, D-SI, and 2031

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics Figure 1. The dynamic ranges in the Peff estimates for each investigated API are summarized in Table 3. The jejunal Peff values for four of the investigated compounds (fexofenadine, metoprolol, cyclosporin, and ranitidine) were also directly measured using the double balloon technique. There was a good correlation between these measurements (r2 = 0.96, slope = 1.24, intercept = 0.030; Figure 2). All estimates of Peff fell within a range considered reasonably similar to previous direct measurements acquired using the single-pass perfusion double balloon technique.11

appeared to be independent of the permeability class of the drug. For some APIs that were in the very high permeability class (bevirimat and fenofibric acid) the Peff values in the LI were less than 5% of those in the P-SI. The most significant reduction in Peff in a distal direction along the intestine was observed for cyclosporin, with estimated jejunal and colonic Peff values of 2.2 × 10−4 cm s−1 and practically zero, respectively. There were no differences between the intestinal segments for two APIs (ipsapirone and metoprolol). Nevertheless, the average relative difference between the LI Peff and the P-SI Peff was 0.30, 0.47, and 0.44 for the APIs in the very high, high, and low permeability classes, respectively. The relative difference in Peff between the P-SI and the D-SI could depend on the permeability class, as the observed differences were 0.93, 1.15, and 0.38 for the APIs in the very high, high, and low permeability classes, respectively. For all APIs combined, the relative Peff estimates in the D-SI and the LI were generally lower 0.90 and 0.40, respectively (Figure 4d).

■

DISCUSSION This study has reanalyzed historical plasma concentration−time data following intestinal and IV administration by using a deconvolution process to obtain 43 new in vivo Peff estimates for 14 APIs in the human P-SI, D-SI, and LI. This collection of APIs represents a wide range of biopharmaceutical and physiochemical properties as summarized in this article. The presented data enable a direct comparison of regional permeability along the human GI tract for the first time. These novel Peff estimates also provide new opportunities to further develop predictive GI absorption tools, such as in silico GI absorption models, in respect of regional differences in absorption along the GI tract.65 Current approaches for the estimation of regional permeability include extrapolations and assumptions related to differences in surface area, physicochemical characteristics of the API and empirical correlations to preclinical species.66−68 A generic module based on human regional Peff estimates is still missing. Such a model would increase the reliability and accuracy of predictions for drug products with potential for drug absorption from the distal parts of the intestinal tract. This includes in particular MR formulations and APIs that are incompletely absorbed in the proximal intestine due to poor solubility and permeability. However, experimental and theoretical investigations are

Figure 2. Correlation plot (r2 = 0.96, slope = 1.24, intercept = 0.030) for the intestinal effective permeability measured using the double balloon technique (DB-Peff) and that estimated using the evaluated deconvolution method (Peff). The DB-Peff data are displayed as means ± SD, while the Peff data are shown as medians and the range. The correlation is shown as a solid line, and the line of unity is shown as a dotted line. The colors indicate the respective active pharmaceutical ingredient: cyclosporin (green), metoprolol (orange), ranitidine (blue), and fexofenadine (red).

Comparison of Regional Intestinal Effective Permeability (Peff). On the basis of these results, the APIs were classified into three classes representing the relevant intestinal permeability to the API: very high (Peff > 3 × 10−4 cm s−1), high (Peff between 1 × 10−4 and 3 × 10−4 cm s−1), and low (Peff < 1 × 10−4 cm s−1). This classification was done to evaluate whether the differences in regional Peff were dependent on the molecules’ potential for membrane permeation. The permeability classes for the respective APIs are indicated in Table 3. The estimated Peff values for each class and intestinal segment are displayed in Figure 3. The corresponding plots for the relative Peff (normalized to the respective P-SI Peff value) are shown in Figure 4. The reduction in the Peff of the LI compared to the SI

Figure 3. Regional Peff estimates for the investigated active pharmaceutical ingredients (APIs), displayed according to the extent of intestinal permeability: very high (a), high (b), and low (c). The intestinal regions are represented as follows: green bars = proximal small intestine; red bars = distal small intestine; blue bars = large intestine. The average value was used for APIs with several estimates from the same region. The data are displayed as medians. 2032

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics

Figure 4. Peff estimates in the proximal small intestine (P-SI), distal small intestine (D-SI), and large intestine (LI), normalized to the respective Peff estimate in the P-SI (Peff/Peff,P‑SI) for active pharmaceutical ingredients (APIs) in the (a) very high, (b) high, and (c) low permeability classes; (d) the overall result for all investigated APIs. The data are displayed as means ± SD.

that study corresponded well with our results (1.46 × 10−4 cm s−1 vs 1.5 × 10−4 cm s−1), although there was a wide discrepancy between the studies for the sumatriptan P-SI Peff (0.28 × 10−4 cm s−1 vs 1.7 × 10−4 cm s−1) and the theophylline D-SI Peff (5.0 × 10−4 cm s−1 vs 0.92 × 10−4 cm s−1).70 The discrepancies may be the result of factors such as differences in deconvolution technique and unit impulse response, but the main reason is probably the source of the data and differences in the route of administration (regional instillation vs oral administration of single doses). The data used in our study are likely to be more specific to the region, not only because of the site-specific administration but also because of the lower risk of confounding factors (e.g., stomach emptying, intestinal transit, and luminal stability) that might affect the analyses after oral administration. There was a general trend regarding the site-specific permeability of the intestine in our study indicating that the Peff and fabs.1h in the LI were lower than those in the SI (Table 3). This was independent of the permeability class associated with the APIs in the SI (very high, high, or low), as demonstrated in Figure 4. The observed variability in the extent of the differences between the Peff values in the various regions was also independent of the permeability class. Both modest (−80%) decreases in the LI Peff compared to the SI Peff were observed for individual APIs in all three permeability classes. Numerous factors, such as molecular properties, luminal instability, or CM mass membrane transport, could potentially have contributed to this variability, as discussed later in this report. However, it does strongly indicate that specific evaluations, in which factors additional to permeability were taken into account, would be necessary to predict colonic absorption when using this kind of approach. Because of the gaps in the relevant information, it is difficult to establish a universal LI Peff−SI Peff relationship based on the data presented in this study. Four APIs with very high SI Peff values [rivastigmine (16 × 10−4 cm s−1), lumiracoxib (3.7 ×

needed to further explore how to make use of this novel data in an optimal way This study was completely dependent on the availability of appropriate clinical studies in the literature; studies were included if they used site-specific single-dose (i.e., not intestinal perfusions) administration of drug solutions and reported the plasma concentration−time profiles. Only a few (n = 15) such studies were accessible, which limited the possibility of excluding suboptimal cases. For instance, no study included information on the disposition PK following IV administration to individual subjects via a crossover study design. Further, no human IV PK were available for some APIs (bevirimat, ipsapirone, and fenofibric acid), and consequently, the absolute values for Peff should be considered as approximations. Nevertheless, all the Peff data acquired in this study are valid for the regional comparison. The good correlation between the four Peff estimations in the P-SI that were compared with direct determinations of jejunal Peff obtained using the single-pass perfusion double balloon technique (for fexofenadine, metoprolol, cyclosporin, and ranitidine; Figure 2) is strong support for the validity of the novel approach applied in this study.11 Furthermore, the P-SI Peff estimates reported in this study were also within the range of previous direct measurements of jejunal Peff using the single-pass perfusion double balloon technique.11 This suggests that the methodology used in our study is not only robust but can also generate accurate values of Peff from different intestinal sites. It is recommended that additional APIs with corresponding direct Peff determinations using the double balloon technique should be investigated, as only four APIs could be included in our direct correlation.69 Data for three of the APIs investigated in this study (metoprolol, sumatriptan, and theophylline) have also been reported in a study in which the Peff for the SI was calculated using deconvolution from the absorption data after oral singledose administration.70 The P-SI Peff for metoprolol estimated in 2033

DOI: 10.1021/mp500834v Mol. Pharmaceutics 2015, 12, 2026−2039

Article

Molecular Pharmaceutics 10−4 cm s−1), fenofibric acid (8.6 × 10−4 cm s−1), and bevirimat (5.9 × 10−4 cm s−1)] demonstrate this. The estimated LI Peff was high for rivastigmine (10 × 10−4 cm s−1) and lumiracoxib (2.5 × 10−4 cm s−1) but low for fenofibric acid (0.48 × 10−4 cm s−1) and bevirimat (0.21 × 10−4 cm s−1). A straightforward general explanation for these observations cannot be made with the available information. Nevertheless, these LI Peff estimates suggest adequate absorption of rivastigmine and lumiracoxib in the LI, but a potential risk of insufficient colonic absorption of fenofibric acid and bevirimat. These predictions are correct for rivastigmine and bevirimat, as the plasma exposure after administration in the LI relative to that in the P-SI were >1 and 0.32, respectively.29,58 However, the reported relative (LI/ SI) plasma exposure for lumiracoxib and fenofibric acid were 0.75 and 0.88, respectively, which appears contradictory.37,46 It is difficult to assess why the LI Peff for bevirimat is so much lower than the SI Peff. Cyclodextrin was used to ensure solubilization, bevirimat is not a P-gp substrate, and there are there no indications of luminal instability in the colon. Potentially, an as yet unknown CM mass membrane transport process, either influx in the SI or efflux in the colon, might be the explanation. Also, discounted colonic gut wall metabolism could have led to an overprediction of colonic Ffirstpass, which would result in a lower estimation of LI Peff (eq 3). In the case of lumiracoxib, Peff decreased over time in the estimated Peff −time profile for the LI. This decline in the Peff−time profile could be the result of colonic luminal instability, which has been reported for other drugs in human colon.33 However, no substance specific evidence has been reported that supports this theory. Like bevirimat, no information on fenofibric acid (e.g., CM transport or luminal instability) supported any theories explaining the reduction in LI Peff. However, the long residence time in the colon could explain the adequate colonic absorption of fenofibric acid over time, as this could have been sufficient to compensate for the low LI permeability.15,37 One could also speculate that the extended residence time in the LI was not enough to compensate for the even lower LI Peff of some APIs, such as bevirimat or the poorly permeable compounds (BCS III and IV). It is generally understood that the permeability of the LI to BCS Class I drugs will be sufficiently high to justify the development of extended-release formulations.5,9 The overall conclusions of this study are in agreement with this understanding. This study included APIs with different protolytic properties. However, molecular charge appears not to explain the observed differences in Peff between the SI and the LI. Nonetheless, the increased risk of precipitation of the acids in the occasionally slightly more acidic proximal LI environment compared to the SI cannot be completely ruled out.15,71,72 However, it has been shown that, overall, the solubilization capacity of the colonic fluid, like that of the SI fluid, is higher than that of buffer.73 The risk of precipitation is also decreased by the solubilizing excipients included in some of the formulations. In summary, it is unlikely that precipitation is the reason for the lower Peff values in the LI than in the SI. Degradation of the API in the lumen would result in smaller amounts remaining unabsorbed than indicated by the deconvolution process. Consequently, the Peff estimates would be underestimated, unless this is corrected for in the calculations (eq 8). Unaccounted-for microbiotic degradation in the LI could therefore have been a reason for the reduction in the estimated LI Peff.33,74,75 Theoretically, all the investigated APIs had molecular structures that were considered susceptible

to bacterial-enzyme catabolism.74,75 Ranitidine and budesonide have been reported to be unstable in faecal material in in vitro assays, and the peptide bond in lisdexamfetamine is particularly vulnerable to proteolysis.33,76,77 However, microbiotic degradation is probably not a common problem, as 77% of the 133 studied drugs in a recent investigation remained stable (