Estimation of the Unbound Brain Concentration of P‑Glycoprotein Substrates or Nonsubstrates by a Serial Cerebrospinal Fluid Sampling Technique in Rats T. Thanga Mariappan,*,† Vishwanath Kurawattimath,† Shashyendra Singh Gautam,† Chetan P. Kulkarni,† Rajareddy Kallem,† Kunal S. Taskar,† Punit H. Marathe,§ and Sandhya Mandlekar‡ †
Pharmaceutical Candidate Optimization, Biocon Bristol-Myers Squibb R&D Centre (BBRC), Syngene International Ltd., Biocon Park, Plot 2 & 3, Bommasandra IV Phase, Bangalore 560 099, India ‡ Pharmaceutical Candidate Optimization, Biocon Bristol-Myers Squibb R&D Centre (BBRC), Bristol-Myers Squibb India Ltd., Bangalore 560099, India § Pharmaceutical Candidate Optimization, Metabolism and Pharmacokinetics, Bristol-Myers Squibb, Pennington, New Jersey 08534, United States ABSTRACT: The unbound concentration in plasma drives the transport of the drug into the brain, and the unbound drug concentration in the central nervous system (CNS) drives the interaction with the target eliciting the pharmacological eﬀect. Delivery of the drug to the CNS is a challenge because of the unique neurovascular unit, which restricts the passage of drugs into the brain. The eﬄux transporters [especially P-glycoprotein (P-gp)] present at the blood−brain barrier (BBB) act as one of the major detractors for keeping drugs outside the CNS. The cerebrospinal ﬂuid (CSF) drug concentration has been used as a surrogate for unbound brain concentrations and has proven to be a good indicator to relate to CNS activity. Herein, we have established a serial CSF sampling technique in rats, which allowed CSF sampling from a single animal and reduced the number of animals required, as well as the interanimal variance associated with a composite/terminal study design. Concentrations in the CSF sampled from the cisterna magna serially from the same rat were compared with the concentrations obtained from discrete CSF sampling and with brain concentrations. The serial CSF sampling technique was also authenticated by ensuring no change in the barrier without any indication of damage caused by the repeated puncture of cisterna magna. This technique was corroborated using three passively permeable compounds (carbamazepine, theophylline, and propranolol), three P-gp substrates (quinidine, verapamil, and digoxin), and one L-amino acid uptake transporter substrate (gabapentin). The P-gp substrates were also used in separate studies with the P-gp inhibitor elacridar to assess the eﬀect on CSF concentration versus brain concentration on P-gp inhibition. The CSF concentration and unbound brain concentration were comparable (within 3-fold) for all compounds, including P-gp substrates even in the presence of elacridar. Therefore, this technique can prove to be beneﬁcial for predicting the unbound drug concentrations in the brain from the CSF concentrations and reduce the cost incurred in preclinical animal models. Chemical inhibition by elacridar and prediction of the brain unbound concentrations from the serial CSF sampling of P-gp substrates in the rat may be an attractive alternative to the use of genetically knocked out rodents. KEYWORDS: serial CSF sampling, P-glycoprotein, blood−brain barrier, blood−cerebrospinal ﬂuid barrier, unbound brain concentration
projected human dose is adequate for testing the proof of concept. The estimation of the unbound drug concentration in the CNS is complicated. On one hand, it is diﬃcult to replicate the complicated disease pathology in brain, and on the other hand, there is diﬀerential expression of transporters and various physiological perpetrators at the blood−brain barrier (BBB) or even in the brain parenchyma. Hence, the detection of active concentrations in the brain is very critical when the target organ Received: Revised: Accepted: Published: 477
July 26, 2013 November 5, 2013 December 31, 2013 December 31, 2013 dx.doi.org/10.1021/mp400436d | Mol. Pharmaceutics 2014, 11, 477−485
surrogate for weakly permeable compounds, such as ceftazidime.15 In addition, the location of the transporters present at the BBB and BCSFB also varies. P-gp is reported to be present at the basolateral side of the BBB and pumps the substrates from the brain ISF to the blood. However, the P-gp present at the BCSFB pumps the substrates from the blood to the CSF.16,17 Because of the diﬀerential orientation, P-gp substrates are expected to show diﬀerential distribution in the brain ISF and CSF. In this work, we describe a simple technique for sampling CSF at multiple time points from the same rat without employing either stereotaxic apparatus or any specialized surgery for cannulation. This experimental protocol was established in rats with the advantage of adequate CSF volume and serial CSF sampling over shorter time spans.18−20 Although CSF sampling reports exist in the literature, in this work we have tried to establish a serial CSF sampling technique that has the advantage of using fewer animals and reducing the interanimal variance. Such a serial sampling technique can be further applied in various CNS animal models. This technique was validated using three passively permeable compounds, three P-gp substrates, and one L-amino acid uptake transporter substrate. Additionally, the utility of this serial CSF sampling technique was explored to evaluate whether the CSF concentration can be used as a surrogate for Cu,brain for both P-gp and P-gp nonsubstrates. The study also evaluated the eﬀect of a P-gp inhibitor (elacridar) on the plasma, brain, and CSF distribution of P-gp substrates. The use of chemical inhibition has an economic advantage over the use of the P-gp knockout mice.21 Compounds were selected on the basis of their behavior to be P-gp substrates (quinidine, verapamil, and digoxin) or P-gp nonsubstrates (carbamazepine, propranolol, and theophylline). Additionally, one compound (gabapentin) was selected as it is a substrate for the L-amino acid uptake transporter at the cell membrane. All the seven distinct compounds were selected to compare the CSF drug concentration with the brain unbound drug concentration, irrespective of the chemical class or therapeutic area or whether it is a substrate for transporters at the BBB.
is the CNS or even more so when minimal CNS concentrations are favorable for avoiding neurotoxicity. The brain is separated from systemic circulation by the blood−brain barrier (BBB) and the blood−cerebrospinal ﬂuid barrier (BCSFB). The BBB restricts the passage of a number of drugs, and frequently, their brain disposition is diﬀerent from what is observed in the peripheral organs. Molecular characteristics, protein binding, and involvement with eﬄux/inﬂux transporters determine the distribution in the brain. Multiple eﬄux and uptake transporters are present at the BBB, of which P-glycoprotein (P-gp) is a major eﬄux transporter that belongs to the ATP binding cassette (ABC) family. It has a wide substrate range and is responsible for the low level of brain penetration of various drugs. Drug disposition in the brain is determined by the exchange of drug among blood, brain interstitial ﬂuid (ISF), brain cells, and cerebrospinal ﬂuid (CSF). Several direct and indirect methods for the estimation of Cu,brain are reported in the literature.2−4 Microdialysis is considered to be the gold standard method for measuring the unbound concentration in the ISF,5 but this method is highly resource-intensive and not suitable for compounds with high lipophilicity because of the nonspeciﬁc binding to the device. Because of the technical challenges, direct measurement of the brain ISF concentration in a clinical setting is highly limited. The CSF concentration is generally used as a surrogate for the ISF concentration to assess unbound drug exposure at the central target sites both in experimental animals and in humans.6,7 Sampling of CSF can be done from various sites, such as ventricles or the cisterna magna, or by puncturing the lumbar membrane (a common procedure in humans). ISF and CSF can be sampled at multiple time points from the same animal to build a concentration−time proﬁle, to reduce interanimal variability, and to reduce the number of animals required for the study. In a preclinical setting, CSF can be sampled by various methods, such as cisterna magna cannulation, intraventricular cannulation, or cisterna magna puncture.6,8 All these techniques involve a specialized surgical technique for implanting the cannula and use of a stereotaxic apparatus to locate the site for cisternal puncturing. Cisternal puncturing is usually terminal and requires separate sets of animals for each time point. There exists a controversy in the literature regarding the usage of the CSF concentration as a surrogate for the unbound concentration in the brain.6,9−11 Although the CSF concentration is reported to be a reliable surrogate for the ISF concentration for passively permeable compounds, conﬂicting reports are available in the literature for P-gp and breast cancer resistance protein (BCRP) substrates. Kodaira et al. showed that predictability of Cu,brain by the CSF unbound concentration (Cu,CSF) decreases with the net transport activities by P-gp and BCRP at the BBB.12 They showed that Cu,CSF of only P-gp and BCRP nonsubstrates can be a reliable surrogate of Cu,brain for lipophilic compounds. However, a recent report by Xiao et al. showed that the level of CSF exposure of P-gp and BCRP substrates (digoxin and dantrolene) increased to an extent similar to that in the brain13 and found that the level of CSF exposure was primarily determined by the rapid transport of compounds from the brain to CSF. This report suggests that the CSF concentration can be used as a surrogate to assess unbound exposure in the brain. Along with these conﬂicting reports, limitations for the use of CSF include regional variation in CSF concentration,9,14 and it is not reported to be a reliable
EXPERIMENTAL SECTION Materials. Carbamazepine (>95%), theophylline (>98%), propranolol (>98%), quinidine (>97%), verapamil (>99%), digoxin (>98%), gabapentin (>95%), and alprazolam (>95%) were purchased from Sigma-Aldrich Chemie, GmbH. Elacridar (>98%) was synthesized by the Discovery Chemical Synthesis department at Biocon-Bristol Myers Squibb Research and Development Centre (BBRC), Bangalore, India. MultiScreen Solvinert ﬁlter plates (0.45 lm, low binding hydrophilic PTFE) were purchased from Millipore (Carrigtwohill, Ireland). Highperformance liquid chromatography-grade methanol was purchased from Merck (Mumbai, India). Formic acid was purchased from Fluka Chemie, GmbH. Milli-Q water from a Milli-Q system (Millipore SAS, Molsheim, France) was used. PE-50 tubing was purchased from Smiths Medical (ASD Inc.), and 22 gauge needles were purchased from Becton Dickinson India Pvt. Ltd. (Bangalore, India). Instrumentation. The Acquity Ultra Performance LC system (Waters, Milford, MA) consisting of a degasser, a binary gradient pump, an autosampler (10 °C), and a column oven (40 °C) was used. A 50 mm × 2.1 mm (inside diameter) Acquity UPLC BEH C18 column with 1.7 μm particles 478
microcentrifuge tubes and stored at −80 °C until the samples were analyzed. For serial sampling, a separate cannula was used for each time point. In the case of discrete sampling, a separate set of rats were used at each time point. The CSF samples were visually observed for any blood contamination, and the samples with blood contamination from the particular animal were not considered for further analysis. After the collection of the CSF from the rats used for discrete CSF sampling, the blood samples (0.3 mL) were collected by retro-orbital puncturing in tubes containing a 2% (w/v) potassium ethylenediaminetetraacetic acid solution. Immediately after blood collection, the rats were euthanized using carbon dioxide and the brain tissues were harvested and homogenized using 4 volumes of water. The brain homogenates and plasma samples were stored at −80 °C until they were analyzed. To understand the eﬀect of multiple puncturing for the collection of CSF samples, neurological scores, sensory motor deﬁcits, and CSF compositions were assessed. For this purpose, a separate study was conducted in which CSF samples were collected at 1, 3, 5, 7, and 24 h from each rat (n = 16). After the last sample collection and recovery from anesthesia, the rats were observed visually for Bederson’s neurological score.23 The rats were then equally divided into two groups. The ﬁrst group was assessed for motor coordination using the accelerating rotarod assay, and the other group was assessed with respect to locomotion and rearing using an actophotometer.24,25 The CSF samples obtained at each time point were used for glucose, protein, and chloride measurement using standard laboratory methods.26 In addition, the sampling site was visually observed for any inﬂammation. Estimation of the Free Fraction in Plasma, Brain, and CSF for Pgp Substrates. Free fractions of P-gp substrates (quinidine, verapamil, and digoxin) were determined using the conventional equilibrium dialysis method. The eﬀect of elacridar on the free fractions of P-gp substrates was also estimated. Undiluted plasma and brain homogenate (diluted 5fold with phosphate buﬀer) obtained from naı̈ve rats were used for the study. The compounds were externally spiked at a concentration of 5 μM in the respective matrix and dialyzed against blank phosphate buﬀer (pH 7.4) for 6 h at 37 °C in a CO2 incubator. The brain homogenate method for determining the free fraction was found to be suitable for the test compounds as they do not undergo cellular accumulation.3 Unbound concentrations in the brain and plasma were calculated by multiplying the total concentration in each matrix by the respective free fraction ( f u). The f u estimated in these experiments was used for P-gp substrates, while literature values were used for all the remaining compounds. The free fraction in CSF (f u,CSF) was calculated by the following equation,3 using the free fraction in plasma (f u,p) and assuming a rat protein CSF concentration of 0.3%.
(Waters) was used for separation. Analytes were detected by multiple-reaction monitoring (MRM) using electrospray ionization (ESI) mass spectrometry on an API-4000 triple quadrupole spectrometer (Applied Biosystems/MDS SCIEX) with Turbo Ion spray. Animals. PK studies were conducted in male SpragueDawley rats weighing 300−350 g (10−12 weeks of age), obtained from the Syngene in-house breeding facility (Bangalore, India). All animal experiments were conducted in an animal research facility of Syngene International Ltd. (Bangalore, India), which was registered by the Committee for the Purpose of Control and Supervision on Experiments on Animals (CPCSEA) and accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC), after gaining approval of the Institutional Animal Ethics Committee (IAEC). The animals were fed with a standard laboratory rodent diet (Tetragon Chemie Pvt. Ltd., Bangalore, India) and housed at room temperature (22 ± 3 °C) and a relative humidity at 50 ± 20% on a 12 h light−dark cycle. Water was provided ad libitum throughout the study. For discrete CSF sampling, 12 rats were used in each experiment. For serial CSF sampling, three rats were used for each experiment. Construction of the Cannula for CSF Collection. A 23 gauge hypodermic needle was cut to 1 cm and inserted into the PE-50 tube (20 cm), leaving ∼5 mm of the bevel end exposed. A silicone bead was ﬁxed (∼5 mm) behind the bevel end to restrict the depth of insertion of the needle at the puncture site. Procedure for CSF Collection. This CSF collection technique was established with the prepared cannula and did not involve any surgery or sophisticated instruments. Before each CSF collection, the rats were shaved and anesthetized using isoﬂurane. The head was ﬂexed downward at ∼45°, and CSF (∼30−40 μL) was collected by vertically puncturing cisterna magna using the prepared cannula. The silicon bead in the cannula helped to avoid deeper puncturing of the cisterna magna and damage of any barriers. The CSF ﬂowed through the cannula by capillary action, and the volume collected was controlled by marking the cannula and withdrawing the cannula once the CSF reached the desired mark. The volume of CSF collected was ∼30−40 μL at each time point, which was ∼15% of the physiological volume (CSF production rate of ∼3 mL/ day with a physiological volume of 250 μL).22 In Vivo Study. Formulations of carbamazepine (2.5 mg/ mL), theophylline (2.5 mg/mL), propranolol (2.5 mg/mL), elacridar (5 mg/mL), quinidine (12.5 mg/mL), verapamil (5 mg/mL), and digoxin (2.5 mg/mL) were prepared in 10% dimethyl acetamide, 40% polyethylene glycol-400, 30% hydroxypropyl-β-cyclodextrin, and 20% water. The formulation of gabapentin (25 mg/mL) was prepared using saline. Rats were kept in a restrainer, and the compounds were dosed via the tail vein as a bolus at a dose volume of 2 mL/kg. Thus, the doses used were 5 mg/kg for carbamazepine, theophylline, propranolol, and digoxin, 10 mg/kg for verapamil, 25 mg/kg for quinidine, and 50 mg/kg for gabapentin. For the combination study with P-gp substrates, elacridar (10 mg/kg) was co-administered in the same formulation.21 Two sets (serial and discrete sampling) of rats were used. Plasma, brain, and CSF samples were collected at each time point (1, 3, 5, and 7 h) from the rats used for discrete sampling. Only CSF samples were collected from the rats used for serial sampling at 1, 3, 5, 7, and 24 h, and plasma and brain samples were collected at 24 h. The CSF samples were transferred to
1 ⎛ 1 ⎞ 1 + 0.003⎜ f − 1 ⎟ ⎝ u,p ⎠
Using the equation given above, the values of the percentage free fraction in the CSF (% f u,CSF) of P-gp substrates quinidine, verapamil, and digoxin were estimated to be 99.39, 97.63 and 98.80%, respectively. As there was negligible protein binding in the CSF (≥98% free), the measured CSF total concentration was considered as the free CSF concentration. 479
Bioanalytical Method. A sensitive UPLC−MS/MS (ultra performance liquid chromatography−mass spectrometry) method for the quantitative determination in plasma, brain, and CSF was developed. The parameters used in the mass spectrometer are listed in Table 1. The analytes and the internal
(obtained via homogenization with 3 volumes of water) with blank plasma, resulting in a net dilution factor of 16.27 Pharmacokinetic Analysis. Unbound concentrations (Cu,brain or Cu,plasma) were calculated by multiplying the total brain or plasma concentration by the free fraction obtained from brain homogenate or plasma using the equilibrium dialysis method. The area under the concentration−time proﬁle (AUC) was calculated by noncompartmental approach and the mixed log linear method using Kinetica (version 4.4.1, Thermo Electron Corp.). AUCCSF was calculated using CSF concentrations, and AUCu,brain was calculated using unbound brain concentrations. The partition coeﬃcient (Kp,uu,CSF/brain) was calculated using the ratio of the AUCs obtained from the CSF concentration−time proﬁle and unbound brain concentration−time proﬁle (Kp,uu,CSF/brain = AUCCSF/AUCu,brain). Statistical Analysis. Statistical analysis was performed using GraphPad Prism version 5.02 (GraphPad Software, San Diego, CA). A Student’s t test was conducted (a) to compare the sensory motor deﬁcits (motor coordination, locomotion, and rearing) of control rats and rats from which ﬁve multiple samples were collected and (b) to compare the CSF concentrations of compounds obtained at each time point (1, 3, 5, and 7 h) by serial and discrete sampling. One-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test was conducted for comparisons among the glucose, chloride, and protein levels in the CSF obtained at 1, 3, 5, 7, and 24 h. p < 0.05 was regarded as signiﬁcant.
standard were eluted from a BEH-C18 column under a gradient condition using a mobile phase consisting of 5 mM ammonium formate with 0.1% formic acid in water and 0.1% formic acid in acetonitrile at a ﬂow rate of 0.6 mL/min. The column was maintained at 40 °C. The injection volume of the test sample was 3 μL. Each wash cycle consisted of 600 μL of strong wash solvent [acetonitrile and water, 9:1 (v/v)] and 800 μL of weak wash solvent [acetonitrile and water, 1:9 (v/v)]. Sample Preparation Method (plasma, brain homogenate, and CSF). Aliquots of standards and study samples (25 μL) were processed with 125 μL of acetonitrile (containing 150 nM alprazolam as an internal standard) in 96-well plates. The plates were vortexed on a plate shaker for 5 min at 300 rpm and then centrifuged at 2600g for 5 min. The concentrations of test compounds in the ﬁltrate were determined by UPLC−MS/MS. Calibration curves for each compound were made by diluting the samples serially with a dynamic range of 1.22−10000 nM. Calibration curves were prepared using blank rat plasma and artiﬁcial CSF. Brain samples were analyzed using the plasma calibration curve after a 4-fold dilution of the brain homogenate
RESULTS Establishment of the Serial CSF Sampling Technique in the Rat. Preliminary experiments were conducted to assess the success rate of CSF sampling using 30 rats. Initially, blood contamination was observed at later time points, but with practice, this was eliminated. After the serial CSF sampling technique had been established, various parameters (neurological score, locomotor activity, and composition of the CSF) were evaluated to assess the impact of multiple puncturing of cisterna magna. Thirteen of 16 rats did not show any observable deﬁcit. Only three rats showed a grade 1 (forelimb ﬂexion) score. No animal showed grade 2 (decreased resistance to lateral push and forelimb
Figure 1. Sensory motor deﬁcit assessment (motor coordination, locomotion, and rearing) of control rats (n = 6) and the rats after ﬁve multiple-CSF samples (n = 8) over a period of 24 h. 480
Table 2. Eﬀect of Multiple Puncturing of Cisterna Magna on the CSF Compositiona CSF composition after serial CSF sampling in SD rats (n = 4) parameter
glucose (mg %) protein (mg %) chloride (mequiv/L)
47.5 ± 2.4 7.7 ± 2.8 137.0 ± 13.7
49.1 ± 3.8 9.1 ± 0.9 150.0 ± 23.0
47.1 ± 1.7 13.3 ± 1.2b 141.3 ± 7.6
48.3 ± 1.8 12.3 ± 3.4b 146.0 ± 10.6
44.2 ± 2.0 12.7 ± 1.1b 141.3 ± 8.3
Data presented are mean values of four replicates with the standard deviation. Four samples were made by pooling the samples from CSF samples from 16 rats (one sample was made by pooling samples from four rats). bp < 0.05 by one-way ANOVA.
ﬂexion without circling) or grade 3 (same behavior as grade 2 with circling) level symptoms. In addition, there was no formation of visible edema at the site of puncturing in any of the rats. The sensory motor deﬁcit assessment showed no signiﬁcant change compared to the control rats for motor coordination, locomotion, and rearing (Figure 1). This assessment conﬁrmed that multiple puncturing of cisterna magna did not lead to any motor deﬁcits. Table 2 shows the composition of the CSF collected at each time point. As the volume of the CSF collected was small (∼30−40 μL), the samples collected at each time point were pooled from four rats. It was evident that none of the parameters deviated compared to that of the ﬁrst sample. Though there appears to be an increase in the protein levels at the later time points, the levels were similar to the literature values.26,28 Comparison of Serial CSF Samples with Discrete CSF Samples. As most of the literature reports terminal CSF sampling, an eﬀort was made to compare the CSF concentrations obtained from the serial sampling technique to the concentrations obtained from terminal sampling. Figure 2A shows the AUC0−7 h obtained from CSF concentrations after serial and discrete sampling (1, 3, 5, and 7 h) for all test compounds. The AUC0−7 h values for all compounds from serial and discrete sampling were within 25%. Figure 2B shows the CSF concentration−time proﬁle of carbamazepine. The CSF concentrations obtained from serial and discrete sampling were similar at each time point. A similar observation was also made for six other compounds. This clearly demonstrated that the serial sampling technique provided CSF exposure similar that of discrete sampling. Determination of Plasma and Brain Free Fractions for P-gp Substrates. Table 3 shows free fractions for three P-gp substrates in the plasma and brain. All three compounds showed brain free fractions lower than their plasma free fractions. Also, the free fractions were similar in the presence and absence of elacridar in both plasma and brain, which conﬁrmed that elacridar did not aﬀect the protein binding of these compounds. Eﬀect of Elacridar on the Plasma, Brain, and CSF Distribution of P-gp Substrates in Rats. Figure 3 shows the AUC of the plasma, brain, and CSF of P-gp substrates in the presence and absence of elacridar. The CSF levels in the ﬁgure are from serial sampling. In this study, AUC values after a single dose were used to reﬂect the brain and CSF partitioning at steady state, which are utilized to determine the active drug concentrations.12,13 Elacridar did not alter the plasma levels of quinidine and verapamil. However, there was a 2-fold increase in the plasma concentration of digoxin in the presence of elacridar. The brain and CSF levels of all the three P-gp substrates increased 21−26and 6−11-fold, respectively. The increase in the level of CSF exposure was ∼2−4-fold smaller than the increase in the level
Figure 2. CSF exposure obtained from serial (□) and discrete (■) sampling in rats. (A) Comparison of AUC0−7 h for CSF obtained from serial and discrete sampling. AUC0−24 h was not calculated for the discrete sample as the discrete study was not conducted for the 24 h time point. AUC0−7 h for serial sampling represents the mean ± the standard deviation. (B) Representative CSF concentration−time proﬁles for carbamazapine from serial and discrete sampling (n = 3). Concentrations represent means ± the standard deviation for each time point. Key: Q, quinidine; V, verapamil; D, digoxin; C, carbamazepine; P, propranolol; T, theophylline; G, gabapentin.
Table 3. Free Fractions of P-gp Substrates in Plasma and Brain Homogenate in the Presence and Absence of Elacridara free fraction in plasma
free fraction in brain
with 1 μM ELA
with 1 μM ELA
quinidine verapamil digoxin
0.198 0.111 0.337
0.263 0.116 0.470
0.038 0.025 0.065
0.027 0.026 0.057
ELA, elacridar. Data are means of two replicates.
of brain exposure in the presence of elacridar. The free concentrations of elacridar (Iu) in the plasma and brain were higher (14−37-fold) than its inhibitory concentration (Ki)21 for P-gp at its peak concentration (Cmax), and the levels were maintained above the Ki for up to 7 h. Thus, the dose of elacridar (10 mg/kg) used in this study showed suﬃcient brain and plasma exposure to inhibit P-gp at the BBB and BCSFB. 481
Figure 3. Plasma, brain, and CSF distribution of P-gp substrates in the absence (■) and presence (□) of elacridar. The bar diagram was made using the AUC0−24 h obtained from the plasma, brain, and CSF concentrations obtained from each time point. The method for calculation of AUC is given in the Experimental Section. Numbers atop the bars represent relative increases in the AUC in the presence of elacridar compared to the control (absence of elacridar).
in the neurological behavior of the animals caused by multiple puncturing at cisterna magna, which was further conﬁrmed from the sensory motor deﬁcit assessment. The absence of any change in the composition of the CSF, neurological behavior, and sensory motor deﬁcits supports the idea that there is no major eﬀect of multiple-CSF sampling. Further, this technique was validated by comparing the CSF concentrations obtained by serial sampling to that of discrete sampling using seven compounds. This is the ﬁrst report describing a simple technique for serial CSF sampling without the involvement of major surgery or the use of sophisticated instrumentation. This technique can be employed in a drug discovery setting to understand the CNS distribution of NCEs. It reduces the number of animals required for generating the CSF concentration−time proﬁle, as compared to the conventional method using terminal CSF samples. It further helps to reduce the interanimal variability observed when using separate animals. An additional focus of this work was to explore the eﬀect of elacridar on the plasma, brain, and CSF distribution of three Pgp substrates. There are very limited reports in the literature describing the eﬀect of P-gp inhibition on the CNS distribution of P-gp substrates in rats.14 However, there are several reports using P-gp gene knockout mice.13 A dose of 10 mg of elacridar/ kg was used in this study, which was found to be suﬃcient to inhibit P-gp at the BBB and BCSFB as the unbound plasma concentration of elacridar was more than its Ki for up to 7 h. The drug concentration in the CSF is considered to be in equilibrium with the interstitial ﬂuid concentration in the brain. A number of studies have reported the CSF concentration as a surrogate for the unbound brain concentrations. Thus, CSF sampling is commonly employed in humans for assessing the availability of drugs in the CNS.7,30,31 However, there are several reports suggesting that the CSF compartment is separate from the interstitial ﬂuid compartment of the brain
The results of this study by chemical inhibition of P-gp by elacridar in the rat were compared with the data obtained in the mouse from the literature reports.12,29 For quinidine and verapamil, data are available from wild-type and P-gp knockout mice.12,29 For digoxin, the reported data in the mouse are from chemical inhibition of P-gp by elacridar. P-gp inhibition, either by elacridar or by gene knockout, produced a similar eﬀect on the plasma, brain, and CSF levels in both rats and mice.13,21 Use of the CSF Concentration as a Surrogate for the Unbound Brain Concentration. An attempt was made to compare the CSF concentrations of the seven test compounds (three passively permeable, three P-gp substrates, and one uptake transporter substrate) to the unbound brain concentration. As shown above, the calculated free fractions in the CSF for digoxin, verapamil, and quinidine were >97%; hence, we assumed that total CSF concentrations were equivalent to free CSF concentrations. Figure 4A shows the comparison of AUCCSF and AUCu,brain. The AUCCSF was within 3-fold of the AUCu,brain. This suggested that the CSF concentration can be used as a surrogate for the unbound brain concentration. Figure 4B shows the comparison of AUCCSF and AUCu,brain for P-gp substrates administered with and without elacridar. The AUCCSF was within 3-fold of the AUCu,brain when the three P-gp substrates were administered in combination with elacridar.
DISCUSSION Literature reports describing the CSF drug concentration in rats employ terminal sampling that involves several animals sacriﬁced at multiple time points to generate a concentration− time proﬁle or infusion to reach steady state. This study established and validated for the ﬁrst time a simple technique for withdrawing multiple CSF samples from the same rat. This technique can be used to withdraw ﬁve consecutive samples within 24 h. This technique was also found to show no change 482
the CSF compartment is in equilibrium with the brain ISF compartment, the increase in the CSF concentrations is expected to be similar to the increase in brain ISF concentrations upon co-administration with elacridar. However, the increase in the CSF concentrations in combination with elacridar was found to be smaller than the increase in brain concentrations (Figure 3). Similar observations were also noted in the P-gp knockout mouse. This suggests that the distribution of P-gp substrates in the brain and CSF is similar in the mouse and rat and is aﬀected similarly, either by chemical inhibition or by gene knockout. The smaller increase in the CSF concentrations could be explained on the basis of the diﬀerential role of P-gp at the BBB versus the BCSFB. Quantitatively, two-thirds of the CSF volume is formed in the choroid plexus, and one-third enters from the ISF by bulk ﬂow.6 In the presence of elacridar, P-gp at both the BBB and BCSFB is expected to be inhibited to a similar extent as the blood and brain unbound concentrations of elacridar were greater than the in vitro Ki for P-gp. The inhibition of P-gp at the BBB leads to an increase in the brain concentrations, while inhibition of P-gp at the BCSFB leads to a decrease in the CSF concentration. The net result is an increase in the CSF concentrations but not to the same extent as the brain concentrations, because elacridar is not likely to aﬀect the bulk ﬂow of P-gp substrates from the brain ISF to the CSF. The CSF concentrations of all seven compounds tested in this study can predict the unbound brain concentrations within a 3-fold range. The 3-fold range was considered appropriate in this comparison because of a similar indication from results for the unbound concentration in the brain and the CSF concentrations from previous work.32 The results obtained from the serial CSF sampling were similar to the CSF concentrations reported in the literature using discrete terminal sampling. Friden et al. reported that the CSF concentrations can overpredict Kp,uu,brain for drugs with low Kp,uu,brain values and underpredict for drugs with higher Kp,uu,brain values.32 Overprediction of >3-fold was reported for some P-gp substrates (verapamil, loperamide, rifampicin, and nelﬁnavir). In this study, a slight overprediction of Kp,uu,brain (∼2-fold) by the CSF concentration was observed for the weakly brain permeable compounds when Kp,uu,brain < 1 (digoxin, verapamil, carbamazepine, and theophylline). Although there is some variation in the CSF concentrations and unbound brain concentrations, the variation was
Estimation of the unbound brain concentration of P-glycoprotein substrates or nonsubstrates by a serial cerebrospinal fluid sampling technique in rats.
The unbound concentration in plasma drives the transport of the drug into the brain, and the unbound drug concentration in the central nervous system ...