TOXICOLOGY

AND

APPLIED

PHARMACOLOGY

A Physiologically

116, 177-188

( 1992)

Based Pharmacokinetic Model for Nicotine Disposition in the Sprague-Dawley Rat

DAVID R. PLOWCHALK* *t MELVIN E. ANDERSEN,#I AND J. DONALD DEBETHIZY T,’ + Duke

University Institute

Medical Center, Integrated Toxicology Program, Box 300.5, Davison Building, of To.xicology, Research Triangle Park, North Carolina; and 7 R. J. Rqvnolds Building 630-2, Winston-Salem, North Carolina

Durham, North Carolina 27710; Tobacco Company, Pharmacology 27102

jzChemica1 Division,

Industry

Received September 3, 199 1; accepted June 2, 1992

A Physiologically Based Pharmacokinetic Model for Nicotine Disposition in the Sprague-Dawley Rat. PLOWCHALK, D. R., ANDERSEN, M. E., AND DEBETHIZY, J. D. (1992). Toxicof. Appl. Pharmacol. 116, 177-188. A physiologically based pharmacokinetic (PBPK) model was developed to describe the disposition of nicotine in the Sprague-Dawley (SD) rat. Parameters for the model were either obtained from the literature (blood flows, organ volumes) or determined experimentally (partition coefficients). Nicotine metabolism was defined in the liver compartment by the first-order rate constants KNC and KNp which control the rate of nicotine metabolism to cotinine and “polar metabolites” (PM), respectively. These rate constants were estimated by optimizing the model fit to pharmacokinetic data obtained by administering an intraarterial (S)-[ 5-3H]nicotine bolus of 0.1 mg/ kg to 6 rats. Model simulations that optimized for the appearance of cotinine in plasma estimated KNC and KNp to be 75.8 and 24.3 hr-’ , respectively. Use of these constants in the model allowed us to accurately predict nicotine plasma kinetics and the fraction of the dose eliminated by renal (8.5%) and metabolic (91.5%) clearance. To validate the model’s ability to predict tissue kinetics of nicotine, 21 male SD rats were administered 0.1 mg/kg (S)-[ 53H] nicotine intraarterially. At seven time points following treatment, 3 rats were euthanized and tissues were removed and analyzed for nicotine. Model-predicted nicotine tissue kinetics were in agreement with those determined experimentally in muscle, liver, skin, fat, and kidney. The brain, heart, and lung exhibited nonlinear nicotine elimination, suggesting that saturable nicotinic binding sites may be important in nicotine disposition in these organs. Inclusion of saturable receptor binding expressions in the mathematical description of these compartments resulted in better agreement with the experimental data. The B,,, and KD estimated by model simulations for these tissues were brain, 0.009 and 0.12; lung, 0.039 and 2.0; and heart, 0.039 nmol/tissue and 0.12 nrvr, respectively. This PBPK model can successfully describe the tissue and plasma kinetics of nicotine in the SD rat and ’ To whom correspondence should be addressed.

will be a useful tool for pharmacologic studies in humans and experimental animals that require insight into the plasma or tissue concentration-effect relationship. 0 1992 Academic

Press, Inc.

Human exposure to nicotine can occur through the consumption of tobacco products, exposure to environmental tobacco smoke (ET’S), pharmaceuticals such as Nicorette gum, use of nicotine-based insecticides, and dietary sources (Sheen, 1988 ) . Tobacco use represents the most substantial exposure to nicotine. Surveys by the Centers for Disease Control indicate that 26.5% of the adult US population are smokers (CDC, 1987). This extensive use of tobacco in our society has generated concern over the potential health risks associated with nicotine (DHHS, 1989). Accurately predicting nicotine uptake and disposition in humans is a necessary step in assessing the potential adverse health effects of nicotine. Although plasma pharrnacokinetics have been extensively documented in animals( Kyerematen et al., 1982, 1987, 1988; Adir et al., 1976; Rotenberg et al., 1980) and humans (Rosenberg et al., 1980; Scherer et al., 1988; Kyerematen et al., 1990), the relationship between plasma concentrations, tissue concentrations, and pharmacologic or toxicologic effects are not well understood. The principal pharmacologic actions of nicotine are mediated through central and peripheral nicotinic receptors (Su, 1982; Lippiello and Fernandes, 1986). These effects have been reported to be dose dependent and include modulation of cardiac functions (Rosenberg et al., 1980), brain electrophysiology (Clarke, 1990)) and systemic vascular blood flow (Benowitz et al.. 1990; Henrich et al., 1984). In addition, recent evidence that chronic nicotine administration results in an increase in the density of nicotinic cholinergic receptors in the mammalian brain has resulted in the experimental administration of nicotine to Alzheimer’s patients (Newhouse et al., 1988; Sahakian et al., 1989). To understand how these pharmacologic effects are related to nicotine ex-

177

004 1-008X/92 $5.00 Copyright 0 1992 by Academic Press. Inc. All rights of reproduction in any form reserved.

178

PLOWCHALK.

ANDERSEN,

posure, the resulting nicotine concentration must be known in the target organ. Early autoradiographic studies indicated that nicotine accumulates in a variety of tissues (Hansson and Schmiterlow, 1962; Andersson et al., 1965; Waddell and Marlowe, 1976 ), which may account for its large volume of distribution ( l4 liters/ kg) (Svensson, 1987; Benowitz et al., 1982). Affinity for these tissues may be a function of the chemical properties of nicotine in viva At physiologic pH, approximately 7088% of nicotine is ionized at the pyrrolidine nitrogen (Benowitz, 1986) allowing nicotine to be sequestered into “acidic tissue” compartments (e.g., kidney and stomach ), similarly to other weak bases. Although this nonspecific accumulation has no apparent pharmacologic importance, it may be significant in the overall disposition of nicotine. In addition to nonspecific binding, nicotine tissue kinetics may also be affected by binding to specific nicotinic receptors in organs such as the brain ( neuronal), heart (peripheral), and muscle (peripheral) ( Wonnacott, 1990; Sleight and Widdicombe, 1965). The objective of this study was to develop a physiologically based pharmacokinetic model (PBPK) that describes the tissue and plasma pharmacokinetics of nicotine in the SpragueDawley rat. Although physiologic flow models for nicotine have been published (Schwartz et al., 1987; Benowitz et al., 1990)) to our knowledge these models were not tested against experimental data and only the results of simulations were reported. However, a three-compartment mammillary model that was based on physiologic parameters has been developed for nicotine and was used to assessplasma nicotine kinetics in rabbits (Porchet et al., 1987 ) . When considering the number of humans that are exposed to nicotine from a variety of sources, it is apparent that a model which describes nicotine tissue dosimetry would be a useful tool for relating nicotine exposure with specific pharmacologic effects. A validated model for nicotine disposition in the rat will be an important first step in the process of developing models that can be used to assessnicotine tissue kinetics in humans. A number of investigators (Ramsey and Andersen, 1984; Andersen et al., 1987) have extrapolated PBPK models developed in rodents to humans using allometric scaling, illustrating the value of such models. Moreover, a variety of current issues, such as the selection of appropriate dosing regimens for pharmacology studies conducted in animals, nicotine pharmacodynamics in humans, and nicotine exposure from ETS and other environmental sources can be examined in greater detail with this type of model. MATERIALS

AND METHODS

Chemicals. ( - )- [ Pyrrolidine-2- 14C] cotinine ( 56 mCi / mmol ) . pyrrolidine-2’-“‘C)nicotine (56 mCi/mmol) and (S)-[ 5-3H]nicotine hydrochloride (25.8 mCi/mmol) were purchased from Amersham (ArlingDL(

AND DEBETHIZY

ton Heights, IL). Radiochemical purities determined by high-pressure liquid chromatography were 96.6. 95.8, and 98.6%, respectively. (S)-[ 5‘H] Nicotine hydrochloride was synthesized by Amersham using the procedure described by Shigenaga et al. ( 1987). The starting material for this synthesis, (S)-5-bromonicotine, was obtained from Dr. Peter Crooks (University of Kentucky). Nicotine was purchased from Eastman Kodak (Rochester. NY) and distilled from NaOH in vacua (bp = 89.0-9O.O”C at l-2 Torr) to obtain a purity of 99+%. Unlabeled nicotine was used to dilute labeled nicotine to the desired specific activity (S.A.). Cotinine (99% purity) was purchased from Sigma Chemical (St. Louis, MO). HPLC grade methanol and acetonitrile were obtained from Burdick & Jackson (Muskegon, Ml) and triethylamine (99+% purity) was purchased from Aldrich Chemical Co. (Milwaukee, WI). Animals. Male Sprague-Dawley rats [ Crl, CD / BR] weighing between 175 and 200 g were purchased from Charles Rivers Laboratories (Raleigh. NC) and were allowed feed (Purina 5002) and water ad libitum. The rats were kept on a 12-hr light/dark cycle and acclimated to their environment for 2 weeks prior to use. All animals were housed and cared for in accordance with the Animal Welfare Act of 1970 and amendments (Public Law 91579) as specified in CFR Title 9, Part 3 Sub-part E (ILAR, 1985). Reference was also made to the DHHS document Guide for the Cure and Use o/ Luborafory Animals (NIH publication 86-23). Surgery. Rats were anesthetized with a 0. I mg/kg ip injection of a 1:1 solution of 100 mg/ml ketamine HCI (Aveco Co., Fort Dodge, IA) and 20 mg/ml xylazine (Mobay Corp., Shawnee, KS). When the animals reached stage III (surgical stage) general anesthesia (5-10 min), preparation for surgery was initiated. The incision sites (neck and back) were first prepared by shaving and use of a depilatory, followed by disinfection of the exposed skin with Betadine. The rats were restrained on their backs on a heated surgical platform. An incision was made in the skin 1 cm on the right and left side of the trachea, and the right jugular vein and left common carotid artery were blunt dissected from surrounding tissues. The right jugular vein was cannulated with 0.04-mm OD Micro-renathane (Braintree Scientific, Braintree, MA) surgical tubing. Similarly, the left carotid artery was cannulated with 0.033-mm OD Micro-renathane tubing. The free ends of the cannulas were run subcutaneously around the neck and dorsally to the back where they were exteriorized through a small incision between the scapula. All incisions were closed with either wound clips or surgical silk sutures (3-O). Following surgery. the rats were allowed to recover for 48-72 hr before use in an experiment. The cannulae were checked daily and were kept patent with heparinized saline solution (25 Units/ml). Determination of partition coefficients. Nicotine tissue:plasma partition coefficients were determined by infusing rats with nicotine and measuring nicotine tissue and plasma concentrations at steady-state (C,,). Six male rats weighing between 230 and 280 g were cannulated in the carotid artery and jugular vein as described above. The arterial cannula was connected to a Harvard infusion pump (Harvard Apparatus. Inc., South Natwick, MA) and the rats were infused with lactated Ringer’s solution containing 30 Units/ ml sodium Heparin at a rate of 0.2 ml/hr during the 48- to 72-hr recovery period. Each rat then was infused intraarterially for 6 hr (steady-state) with 0.25 mg/kg/hr (S)-[5-3H]nicotine hydrochloride (S.A. = 1.82 pCi/pg) and venous blood samples were collected at I. 2,3.4,5, and 6 hr. Following the last blood sample, the rats were anesthetized with 70% COz in air and a midline incision was made into the chest cavity to expose the heart. The rats were simultaneously exsanguinated and perfused by severing the right atrium and infusing ice-cold 0.9% saline ( 15 ml) through the arterial cannula. Following this procedure, tissue samples were removed, rinsed of blood. and frozen to -70°C. The tissues collected included muscle (gluteus). fat (epididymal). brain, skin (back). heart, liver, kidney, and lung. Plasma and tissue samples were prepared and analyzed for nicotine and cotinine concentrations as described below. Partition coefficients were calculated for non-eliminating and eliminating organs as specified by Chen and Gross ( 1979). For non-eliminating organs

PBPK MODEL (skin. fat, muscle, lung, heart, and brain), the concentration of nicotine in the tissue, determined after a steady state infusion of nicotine, was divided by the plasma concentration of nicotine at that time. For eliminating organs, (liver and kidney), the partition coefficients were adjusted to account for intrinsic clearance (Chen and Gross, 1979) using the following equation:

p=(I+% .g Q1 p where P is the tissue:plasma partition coefficient, CL, is the intrinsic clearance of the organ, Q is the plasma flow rate through the organ, and Cp and C; are the steady-state concentrations of the nicotine in tissue and plasma, respectively. Nicotine/cotinine pharmacokinetic studies. Six male rats weighing between 250 and 300 g were cannulated as described above. Each rat was administered 0.1 mg/ kg of(S) - [ 5- ‘H ] nicotine hydrochloride ( S.A. = 9.2 1 fiCi/pg) intraarterially in a volume of 0.9% sterile saline (pyrogen-free) equivalent to 1.O ml/kg. After dosing, each rat was housed in a metabolism cage and was allowed free accessto feed and water. The mine and feceswere collected on dry ice for the duration of the experiment. Venous blood samples (0.25 ml) were drawn from the jugular vein at 0,0.083,0.167.0.25,0.333, 0.5, 1, 2, 3, 4, 5, 6, 8, 10, 12, and 24 hr and placed in Microtainer blood collection vials containing sodium EDTA ( Becton-Dickinson, Rutherford, NJ). Sterile, pyrogen-free 0.9% saline (0.25 ml) was administered following each blood sample to replace the blood volume removed during sampling. Samples were centrifuged at 10,500g for 5 min and the plasma was removed and stored at -70°C. Nicotine tissue disposition experiments. Twenty-one male rats weighing between 230 and 300 g were anesthetized and a cannula was surgically implanted in the carotid artery as described above. Three rats were randomly assigned to each of seven groups representing 0.25-, 0.5-, l-, 2-, 3-, 5-, and 8-hr time points. Each rat was administered 0.1 mg/ kg of(S)- [ 5- ‘H ] nicotine hydrochloride (S.A. = 2.19 &i/pg) intraarterially in 0.9% saline ( 1.0 ml/ kg). At the selected time points, the rats were anesthetized with 70% CO2 in air, exsanguinated, and perfused as described above. Following this, tissue samples were removed, rinsed of blood, and frozen to -70°C. Determination of nicotine and cotinine in plasma and tissues. Plasma samples were thawed at room temperature and absolute ethanol was added to an equal volume of plasma to precipitate proteins. The mixture was vigorously shaken and allowed to stand on ice for several hours followed by centrifugation at 10,500g for 20 min. The supematant fraction then was pipetted into an Ultrafree-MC 0.45-pm filter unit (Millipore) and centrifuged at 2000g for 5 min. The filtrate (200 ~1) was placed in 250-J auto-sampler vials (Waters, Millford, MA) and stored at -20°C until HPLC analysis. For determination of nicotine and cotinine in tissues, approximately 0.5 g of frozen tissue was minced and placed into 3 ml of ice-cold methanol containing 0.127 &i [ pyrrolidine-2’-?Z]cotinine and 0.126 &i racemic [ pyrrohdine-2’-‘4C]nicotine as internal standards. The tissue was homogenized with a Tekmar Tissumizer (Tekmar, Cincinnati, OH) while the sample vial was cooled in a methanol/dry-ice bath. The resulting homogenate was centrifuged at 5000g for 30 min at 2°C. The supernatant fraction was concentrated to approximately 400 ~1 using a Speed Vat concentrator (Savant Instrument Co., Farmingdale, NY ), and then filtered through an UltrafreeMC 0.45~pm filter unit (Millipore, Bedford, MA). All tissue samples except for skin were prepared by this method. Because of the fibrous nature of skin, this tissue was digested for 24 hr in Unisol tissue solubilizer (Isolab, Inc., Akron, OH). The analytes and internal standards were extracted directly from the resulting basic solution into methylene chloride followed by a solvent exchange with methanol. All plasma and tissue samples were analyzed for nicotine and cotinine concentrations using reverse-phase HPLC with a Supelcosil LC-8DB (25 cm X 4.6-mm) column (Supelco. Inc., Bellefonte, PA) and mobile phase

OF NICOTINE

179

consisting of methanol:water:acetonitrile:O. 1 M ammonium acetate buffer (pH = 4.5) (23:5 1:6:20). The water:ammonium acetate buffer mixture was adjusted to pH 7.4 with triethylamine. The flow rate was 1 ml/min and the retention times for nicotine and cotinine were approximately 6 and 10 min, respectively. The retention times of nicotine and cotinine were determined using authentic nicotine and cotinine standards prepared in rat plasma as described above and detected by uv absorbance at 260 nm. Radiochromatograms reconstructed from 0.2-min fractions collected with a Foxy fraction collector (Isco, Inc., Lincoln, NE) were then compared to the uv chromatogram of the authentic nicotine and cotinine standards. Based on these retention times, two fraction collector windows were set to collect nicotine and cotinine peaks. Standards containing nicotine and cotinine were analyzed approximately every 10 samples to ensure consistent retention times. A third window also was set to collect a group of unidentified “polar metabohtes” (PM) which eluted shortly after the solvent front. An aliquot from each window fraction was counted for either ‘H (plasma samples) or ‘H / 14C(tissue samples) using a Packard Tri Carb CA2000 liquid scintillation counter (Packard Instrument Co., Sterling, VA). General model stracfure. This PBPK model contains 11 anatomical compartments (Fig. I ), each of which was considered to be important in either the absorption, distribution, metabolism, or excretion of nicotine, or in its pharmacologic action. The slowly perfused tissue in our model is a lumped compartment which represents the remainder of the body including both slowly perfused tissuesand various organs known to accumulate nicotine (e.g., stomach, salivary glands, adrenals, bone marrow, intestine, spleen. and melanin-containing tissues). Nicotine input to the model can be by intravenous, intraarterial, gastric, or inhalation route. For the experiments described here, however, only intraarterial administration will be discussed. For the purpose of this study, arterial administration was the most appropriate route since it closely approximates nicotine uptake by smokers, i.e., nicotine is rapidly absorbed in the lungs and appears essentially as an arterial bolus after passing through the pulmonary circulation. Following intraarterial input of nicotine, the

Hepatic

Metabolism

+

Urine

FIG. 1. Schematic of the physiologically based pharmacokinetic model for nicotine in the Sprague-Dawley rat (arterial and venous blood compartments not shown) The liver is the major site of nicotine metabolism where cotinine and “polar metabolites” are formed. See the Appendix for a list of the abbreviations and select mass-balance equations describing the model.

180

PLOWCHALK.

ANDERSEN.

model assumes instantaneous mixing of nicotine into the arterial compartment and each subsequent compartment it enters. Partition coefficients are assumed to be concentration and time independent and nicotine distribution is blood flow limited. Physiological constants (Table I ), such as organ volumes and blood flows. were obtained from the literature (Gerlowski and Jian. 1983; Gearhart et al., 1990). Organ and tissue volumes are expressed as a percentage of total body mass and blood flows are expressed as a percentage of cardiac output which was scaled to body weight using the allometric relationship described by Andersen et al. ( 1987 ). In general, the concentration of nicotine in an organ is a function of linear nonspecific binding of nicotine to tissue components, i.e.. the product of the partition coefficient and plasma concentration. However, in organs which have capacity-limited nicotinic binding sites (“nicotinic receptors”), total tissue partitioning was expressed as the sum of linear binding and saturable, high-affinity binding similar to the models developed for methotrexate (Bischaff et al., 197 1) and 2.3,7,8-TCDD (Leung et al.. 1990). To account for saturable binding in this model, a binding expression similar to that described by Bischoff ef ul., ( 197 1) was included for tissues known to possesslarge populations of nicotinic receptors (i.e., heart, lung, and brain). Once the overall model structure was formulated, mass-balance differential equations were written to describe the rate of change of nicotine mass in each anatomical compartment. These equations are similar to those described by Ramsey and Andersen ( 1984) except for brain, lung, and heart. which include saturable tissue binding. and kidney, which includes a term for renal clearance. A Listing of these equations can be found in the Appendix along with equations that describe the formation and elimination of cotinine and other more polar metabohtes. The model equations were “solved” by numerical integration using Gear’s method for stiff systems with the mathematical modeling software ACSL (Advanced Continuous Simulation Language, Mitchell and Gauthier Associates, Concord, MA) on a SUN SPARCstation 1 computer. Parameter estimations and final optimizations were performed with Simusolv (Dow Chemical Co., Midland, MI) which employs the maximum likelihood method. Model description ofnicotine melabolism. The metabolism of nicotine has been well established in animals (Kyerematen ef al., 1987; Gorrod and Jenner. 1975) and humans (Kyerematen et al., 1990; Benowitz. 1986). Nicotine is metabolized to cotinine (Fig. 2) and a variety of other polar metabolites and is also excreted unchanged in the urine. Cotinine is a major metabolite of nicotine and is formed by a two-step reaction catalyzed by cytochrome P450 (rate-limiting step) and aldehyde oxidase (McCoy PI al., 1986: Kyerematen er al. 1988; Booth and Boyland, 1971). Although the liver is the primary site for this biotransformation, other organs such as the lung and kidney also can metabolize nicotine, but to a lesser extent (Hansson et al.. 1964). N-oxidation of nicotine to form nicotine-w-oxide has been reported as an important metabolic pathway in many species (Booth and Boyland, 1970). Other minor metabolites reported to arise from nicotine include nornicotine, nicotine glucuronide (Byrd et al.. 1992), demethylcotinine, and y-( -3-pyridyl)-oxybutyric acid. Clearance of cotinine is accomplished by either renal excretion or metabolism to more polar compounds such as truns-3’-hydroxycotinine, cotinine glucuronide (Byrd et al.. 1992). cotinine-N-oxide. cotinine methonium ion, y-(3-pyridyl)-r-methylaminabutyric acid, 3-pyridylacetic acid. and y-( 3-pyridyl)-y-oxo-N-methylbutyramide (Kyerematen et ul , 1988). The combination of nicotine and cotinine metabolites (excluding cotinine) is defined as PM in this model and represents the unidentified PM peak measured by HPLC. Based on this well-established metabolic scheme, a mathematical description of nicotine metabolism was formulated for the model. The model assumes that nicotine clearance occurs solely by metabolism in the liver and renal excretion from the kidney compartment. Nicotine metabolism is described as a function of two biotransformation pathways: ( I ) the conversion of nicotine to cotinine and (2) the conversion of nicotine to PM (Fig. 2 ). The former reaction was addressed in the model by including a saturable

AND DEBETHIZY

TABLE 1 Physiologic Constants” Used for PBPK Model for Nicotine in the Sprague-Dawley Rat Organ/tissue

Muscle Skin Fat Liver Kidney Brain Heart Lung Slowly perfusedd Arterial blood Venous blood

Organ/tissue volume

Blood flow

(Percentage of BW)b

(Percentage of CO)’

55 10 1

4 0.75 0.55 0.55 I 15 3 3

20 5 9 25 20 5 2 co 14 co co

’ Gerlowski and Jian, 1983; Gearheart et al., 1990; Andersen et a/ , 1987. b Percentage of body weight (0.269 kg). ‘Percentage of cardiac output (CO = 5.3 hter/hr). d Slowly perfused tissue compartment represents the remaining body mass not accounted for by the other compartments.

Michaelis-Menten expression in the liver compartment similar to that described by other investigators (Ramsey and Andersen, 1984; Bischoff et al., 197 I ). The rate of polar metabolite formation from nicotine was defined by the first-order rate constant KNp Cotinine disposition in the model is controlled by: ( 1) the rate of nicotine metabolism to cotinine (Km and V,,,,,. or KNc; see below for definition of KNc): (2) cotinine volume of distribution ( VdcoT ) and; ( 3) total clearance rate of cotinine (C&or ). The metabolism of cotinine to PM was initially separated from the total clearance term. This metabolic pathway was defined by the first-order rate constant Kcp. We found through model simulations that when Kc, was set to a value that had an effect on PM kinetics it resulted in both PM and cotinine kinetic profiles that were radically different from the experimental data. Upon this observation, we set Kcp to 0.001 so that it made cotinine clearance via this pathway insignificant and allowed CLor to account for all clearance mechanisms of cotinine. AUC. tllz and 6 were obtained by first stripping the plasma cotinine concentration vs time data to estimate initial parameters which described a biexponential function and then optimizing the parameters to allow for the best fit to the experimental data. This analysis was performed with RSTRIP software (MicroMath, Salt Lake City, UT). C&o, and b’dcor were obtained from the literature (Gabrielsson and Bondesson. 1987; Miller et nl.. 1977). It should be noted that the portion ofthe model which describes cotinine distribution and elimination is based on classical pharmacokinetic techniques. The overall rate of PM formation is described by the first-order rate constants J(Nr and Kcp which specify the rates of conversion from nicotine and cotinine. respectively. The rate of PM appearance is of interest because it directly affects the amount of nicotine remaining for elimination by other pathways (e.g., metabolism to cotinine or renal clearance). However. the exact composition and disposition of these metabolites are not included since knowledge of the Vd and CL for each component would be required. There are no in vivo data available that allow for calculation of Km and Vl%Uor KNp directly; therefore these kinetic constants were estimated by parameter optimization. Furthermore, we found the ratio of K,,, and v,,, to be more important than their absolute values, an observation similar to one made by Reitz et al. ( 1989) for methylene chloride metabolism This

PBPK MODEL

_

“Polar

OF NICOTINE

Aldehyde

Metabolites”

181

Oxidase

(PM) in Plasma

FIG. 2. Schematic of nicotine clearance on which this PBPK model is based. As illustrated here, nicotine is either metabolized to cotinine and a variety of other “polar metabolites” (shaded rectangle) or is excreted unchanged in the urine. Based on this scheme, the model includes a definition of: ( 1) hepatic metabolism to cotinine, (2) hepatic metabolism to “polar metabolites”, and (3) renal excretion from the kidney compartment. (Abbreviations used: POXB, y-( 3-pyridyl)-y-oxobutyric acid; PMOB, y-( 3-pyridyl)-y-oxo-N-methylbutyramide; PMAB, y-( 3-pyridyl)-y-methyaminobutyric acid; and PAA, 3-pyridylacetic acid). observation also implied that at the nicotine dose examined (0.1 mg/kg) K,,, % Cv,. Therefore, the Michaelis-Menten expression was reduced to a first-order rate constant, KNc, proportional to V,,IK,. The following procedures were used to estimate KNc and KNp. Renal clearance ( CLRENAL)was determined experimentally to be 0.215 liters/hr/ kg by measuring the amount of unchanged nicotine excreted into the urine. The volume of distribution of cotinine Vhor was set to 1.2 liters/kg and total cotinine clearance (CL,,, = 0.2 11 liters/ hr/ kg) was calculated as the product of VdcoT and &or. Assuming these constraints, only the rates at which nicotine is metabolized to cotinine ( KNc) and PM (KM) are unknown. These kinetic constants were subsequently estimated by varying these parameters to optimize the model fit to plasma cotinine concentrations. This procedure allowed us to estimate the rate of nicotine metabolism to both cotinine and PM based on the appearance of one metabolite, cotinine.

RESULTS Partition Coeficients At the end of a 6-hr infusion of nicotine (0.25 mg/kg/ hr ), plasma nicotine reached a steady-state concentration of 85.5 t 34.9 rig/ml, with a range of 55 to 149 rig/ml. The tissue and plasma concentrations at steady state were used to generate the tissue:plasma partition coefficients listed in Table 2. Nicotine partitioned most notably into kidney and liver, achieving concentrations greater than 24 and 7 times that of plasma, respectively. Although these values are not identical to those reported by Benowitz et al. ( 1990), also listed in Table 2, they do agree with the relative affinity of nicotine they report for each tissue. Plasma Pharmacokinetics Classical pharmacokinetic parameters were determined in order to assessthis study’s overall agreement with those pre-

viously reported in the literature and to generate kinetic parameters for nicotine and cotinine which were incorporated into the model. Nicotine plasma kinetics followed a simple biexponential decay composed of a rapid distribution phase followed by a single elimination phase (Fig. 3). The appearance of cotinine in the plasma was rapid and reached a maximum of 47.4 * 5.8 rig/ml at 1.4 k 0.2 hr, followed by an elimination phase with a tl,2P of 3.9 + 0.5 hr. Similar to cotinine, PM quickly rose to a maximum of 2 1.4 + 3.1 ng/ ml at 0.7 + 0.2 hr. Elimination of PM followed a biexponential decay with t, ,2‘s of 2.4 and 42.9 hr for LYand p elimination phases, respectively. Pharmacokinetic parameters generated from analyses of these nicotine concentration vs time data are listed in Table 3. Surprisingly, there was little variability within nicotine, cotinine, or PM plasma concentrations among the six rats as shown in Fig. 3. A comparison of actual nicotine and cotinine plasma concentrations with model predicted values is illustrated in Fig. 4. The model successfully predicted the plasma kinetics of intraarterially administered nicotine and the appearance, distribution, and clearance of cotinine. The model was not designed to describe PM disposition, which was included solely to help estimate the metabolic clearance of nicotine. Tissue Kinetics Figure 4 shows the concentration time-course of nicotine for eight tissue compartments following an intraarterial bolus of 0.1 mg/kg nicotine. The nicotine tissue concentrations predicted by the model were in accordance with experimentally determined values for all organs.

182

PLOWCHALK.

ANDERSEN.

AND DEBETHIZY

Kidney tissue had the highest measured nicotine concentration, with nicotine concentrations of 969 ngjg tissue followed by liver (171), brain (49), lung (33), muscle (29). heart (24)) skin (2 1)) and fat ( 8). It should be recognized that these are tissue concentrations measured at 15 min and do not necessarily correspond to the organs which achieved the highest concentrations immediately after dosing. However, the model predicts that the peak tissue concentrations occur between 1- 15 min and are in the same order. Nicotine was rapidly distributed to most organs within 15-30 min and then was followed by a single elimination phase. 0: Model-Estimated

Physiologic Constants

Heart, lung, and brain exhibited a nonlinear nicotine elimination phase. Figure 5 illustrates the elimination of nicotine from brain, lung, and heart with and without the inclusion of saturable, high-affinity binding sites. The values of &ax (nmol/ tissue) and KD (nM) used to obtain these fits were 0.009 and 0.12 for brain, 0.039 and 2.0 for lung, and

TABLE 2 Tissue:Plasma Partition Coefficients for (S)-[5-3H]Nicotine Determined in Viva for the Sprague-Dawley (SD) Rat Partition coefficients Benowitz et al. (1990) Organ

Our lab” SD ratb

Rat’

Rabbitd

Muscle Skin Fat Liver Kidney Brain Heart Lung Slowly perfused/ Arterial bloodg Venous bloodg

1.1 + 0.3 1.1 * 0.3 0.2 k 0.1 7.0 2 0.5 24.8 t 9.4 I .4 2 0.1 0.6 k 0.1 0.9 f 0.4 6.4 1 1

1.6 ND’ 0.5 3.5 13.3 2.8 1.9 2.1 ND I I

2 ND 0.5 3.7 21.6 3 3.1 2 ND I 1

’ Determined after a 6-hr intraarterial nicotine infusion (4.17 rg/min/ kg). b Mean and standard deviation for six animals. ’ Determined after a 72-hr subcutaneous nicotine infusion with osmotic minipumps (2.8 pg/min/kg) (rat strain not stated). d Determined after a 24-hr intravenous nicotine infusion (2.5 wg/min/ kg). e ND, Not determined. ‘The partition coefficient for the slowly perfused tissue compartment was estimated by holding all parameters constant and adjusting the partition coefficient to acheive the best model fit with the initial plasma distribution phase of nicotine. g Not experimentally determined, but equal distribution of nicotine occurs throughout the vascular compartment and plasma protein binding is less than 10% (Rotenberg, 1978).

*

15 20 25 30 Hours FIG. 3. Plasma concentrations of nicotine, cotinine and “polar metabohte” (PM) following a 0.1 mg/kg intraarterial bolus of (S)-[ 5-3H] nicotine hydrochloride. All data points represent the mean analyte concentration & standard deviation of six rats. Model simulations are represented with solid lines. A simulation was not attempted for PM because the model description only accounted for PM appearance and not distribution or elimination. 0

5

10

0.039 and 0.12 for heart, respectively. These values were estimated by holding all model parameters constant and varying KD and B,,, to optimize (visually) for either the concentration of nicotine in the brain, lung, or heart. Clearly, the addition of saturable binding enhanced the overall agreement between the model prediction and experimental data. The optimized rate constants KNC and KNp were estimated to be 75.8 and 24.3 hr-‘, respectively, and are expressed with respect to the free concentration of nicotine in the liver (Table 4). Incorporation of these values into the model predicts the proportion of the nicotine dose eliminated by CLRENAL, metabolism to cotinine and metabolism to PM, to be 8.4,69.3, and 22.2%, respectively. The metabolic clearance of nicotine in the rat has been shown to be linear over the dose range of 0.01 to 1.0 mg/kg (Kyerematen et al., 1988; Rotenberg, 1978 ) . These data suggest that the hepatic metabolic oxidation of nicotine by flavin monooxygenase and cytochrome P450 is not saturable up to near lethal doses. The inability to saturate the metabolism of nicotine in vivo makes it impractical to determine accurately the in vivo K, and V,,, for nicotine metabolism. To circumvent this problem and obtain an estimate of the lower limits of K, and V,,,,, we have used the PBPK model to simulate the nicotine plasma concentrations from the highest nonlethal dose of nicotine which was used by Kyerematen et al. ( 1988 ) ( 1.O mg/kg). In these simulations, K, and I’,,, were lowered while maintaining a constant ratio of V,,,/K, (0.84) until the plasma elimination of nicotine exhibited nonlinear ki-

PBPK MODEL

TABLE 3 Pharmacokinetic Parameters for Nicotine and Cotinine Determined from Six Male Sprague-Dawley Rats Administered 0.1 mg/kg (S)-[S3H]nicotine Intraarterially Pharmacokinetic parameter t1/20 (W’

AUC (ng . hr/ml)” (3 (hr-‘) CLr,,, (liters/hr/kg) (liters/hr/kg) CLMetabolic Chenal (liters/hr/kg) Vd, (liters/kg)

Nicotine 0.9 34.9 0.8 2.9 2.7 0.2 3.1

i + + f f f t

0.1 5.5 0. I 0.46 0.5’ O.ld 0.6’

Cotinine 3.9 t 0.5 341.1 Ik 53.3 0.18 r 0.03

o.21fKLoT) 1.29

a Determined from a biexponential curve fit of plasma nicotine concentration vs time data using RSTRIP, a polyexponential curve stripping and parameter estimation program. b CLota, = dose/AUC$ c CLenal = XuF/AUC$, Xu$ = amount of nicotine excreted in the urine fromf=O+w. d CLMerablic = cbolal - cLRmal. e Vd, = CLotsdP~,c. /CL,,, = C’dcOT*BCOT g Gabreilsson and Bondesson, 1987; Miller ef al., 1977.

183

OF NICOTINE

where the model predicted experimental data sets from the literature (Miller et al., 1977). The nicotine partition coefficients agreed fairly closely with the relative tissue affinities within the species previously examined (Benowitz et al., 1990). A notable and potentially important difference was the partition coefficient determined for the brain. Benowitz reports partition coefficients of 2.8 and 3 for the rat and rabbit, respectively, whereas we estimated a partition coefficient of 1.4 for the Sprague-Dawley rat. Using this twofold greater value in the model would predict brain concentrations twice as high as we determined experimentally. It is also worthy to note that Benowitz found different partition coefficients between the rat and rabbit for heart and kidney tissues. Interspecies differences in nicotine tissue affinity have been reported by other investigators. Leed and Turner ( 1977) noted a greater than twofold difference

I

Kidney

netics. This occurred at 9.0 PM and 7.6 pmol/ hr for K, and Vmax> respectively. Based on these findings, we know that the apparent in viva K,,, and V,,, must be greater than these values. Additionally, we can conclude that nicotine metabolism is linear even at a near lethal iv dose ( 1.O mg/kg), and it is not possible to saturate the metabolic clearance of nicotine in vivu to obtain a more precise determination of K,,, and V,,,,,. I

DISCUSSION Several investigators have previously proposed PBPK models for nicotine. Benowitz et al. ( 1990) described a perfusion model constructed of compartments similar to those described in our model. However, no validation of either tissue nicotine concentrations or plasma kinetics of nicotine and cotinine have been published. Similarly, a model by Schwartz d al. ( 1987) was theoretical in nature and described only simulation results. Although it was used to simulate several experimental data sets (personal communication), to our knowledge no validation of this model has been published. Unlike these previous models, we have described a model and calibrated it by determination of nicotine concentrations in venous plasma and in eight tissues. Overall, this PBPK model successfully describes nicotine concentrations in both plasma and tissue compartments, providing good agreement with the experimentally determined data. The model also accounts for the plasma kinetics of cotinine. The general utility of the model is best illustrated in Fig. 5

I

Skin

FIG. 4. Time-course concentrations of nicotine in eight organs following a 0.1 mg/kg intraarterial bolus of (S)-[ 5-‘HInicotine hydrochloride. Each data point represents the mean nicotine concentration * standard deviation of three individual rats. Solid lines are simulations without saturable tissue binding. Dashed lines in the brain, heart, and lung panels illustrate the simulated nicotine kinetics with the saturable nicotine binding in these tissues.

184

PLOWCHALK,

0 W 0 0 -Model

0

2

4

6

Nicotine Nicotine Cotinine Cotinine

ANDERSEN,

(Dose = 0.06 mg/kg) (Dose = 0.4 mg/kg) (Dose = 0.06 ma/kg) (Dose = 0.4 mglkg) Prediction

6

10

12

Hours FIG. 5. Data sets from the literature were also simulated in order to test the model’s predictability with different doses and rat strain. In a study by Miller et al., 1977, Fischer-344 rats were administered (iv) either 0.08 (circles) or 0.4 (squares) mg nicotine/ kg. The model accurately predicted (solid line) both plasma nicotine (filled symbols) and cotinine (open symbols) concentrations in these experiments.

in kidney and brain affinities for nicotine between Wistar rats, White Carneau pigeons, and cats. Tsujimoto et al. ( 1975) reported that the skeletal muscle of the dog has a greater affinity for nicotine than that in the rhesus monkey. Such differences may become especially critical when trying to establish tissue concentrations of nicotine associated with a pharmacologic effect. Furthermore, this implies that model scale-up to other species, including humans, may require direct determination of partition coefficients or binding in tissues of the species of interest. The greatest tissue:plasma partition coefficient, 24.7, was observed in the kidney. Accumulation and excretion of nicotine by the kidney appear to be a function of urinary pH. Several investigators (Rosenberg et al., 1980; Benowitz and Jacob, 1985; and Feyerabend and Russell, 1978 ) have demonstrated that by reducing urinary pH, renal clearance of nicotine is increased. In this study, high variability in both kidney partition coefficients and renal clearance rates (coefficients of variation, 40.1 and 47.9%, respectively) may be explained by differences in urinary pH among the experimental animals (not determined). Nicotine, which is a weak base, is ionized at physiologic pH and has been reported to be sequestered by acidic sites within the body (Waddell and Marlowe, 1976; Andersson et al., 1965). In whole body autoradiography studies, substantial amounts of nicotine are found in the stomach lumen (Andersson et al., 1965) and melanin-containing tissues (Waddell and Marlowe, 1976). Other organs also were found

AND DEBETHIZY

to accumulate nicotine, e.g., salivary glands, adrenals, bone marrow, bronchial epithelium, intestine, spleen, and bladder. In part, these organs comprise the slowly perfused tissue compartment which accounts for the remaining body mass not accounted for by individual organ compartments. It is not unreasonable to speculate that the large partition coefficient for this compartment (6.4) estimated using the model may be a reflection of the affinity of nicotine for these sites. When evaluating the in vivo experimental data and model predictions it became obvious that saturable tissue binding was required to describe tissue nicotine kinetics adequately in the heart, lung, and brain of the rat. These organs exhibited nonlinear elimination of nicotine, and the addition of high affinity nicotine binding sites permitted a good fit to the experimental data at time points greater than 3 hr. These tissues appear to possess nicotine-binding sites which have a significant impact on tissue kinetics, although the B,,, is low enough that nicotine binding in these tissues does not affect the overall plasma pharmacokinetics at time points most often examined in pharmacokinetic studies. The authors recognize that only three or four data points are present in the region of the tissue elimination curves from which the binding constants are derived and, therefore, it should be stressed that the values reported for B,,, and Kn are provisional. However, because high- and low-affinity nicotinic receptors have been reported in mammalian brain and heart ( Wonnacott, 1990; Lippiello and Fernandes, 1986; Sleight and Widdicombe, 1965 ), the use of a saturable binding term in the model is biologically plausible. Recent reports of a second elimination phase for nicotine in human plasma (t, ,2 = 5-6 hr), which is longer lived than generally reported, may be explained by these tissue kinetics (R. J. Reynolds, 1988; Benowitz et al., 199 1). The KD and B,,, determined

TABLE 4 Metabolic Rate Constants and Tissue-Binding Constants Estimated for the Nicotine PBPK Model Metabolic constants Km (PM) V,, (rmollhr) KNc (hr-‘)” KNp (hr-I)” KCP

W’)

Nicotine binding constants B,, (nmol/heart) B,, (nmol/brain) B,, (nmol/lung) KD (heart, nM) KD (brain, nM) KD (lung. nM)

a-9.0 27.6 75.8 f 1.5 24.3 k 0.5