Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

It is well established that liver disease alters hepatic microcirculation (Varin and Huet, 1985) and impairs the disposition kinetics of many solutes, including drugs (Reichen et al., 1987; Gariepy et al., 1993). Callaghan et al. (1993) and Morgan and McLean (1995) have suggested that hypoxia, changes in enzyme activity, and alterations in plasma and intrahepatocellular proteins (albumin and ₁-acid glycoprotein) during liver disease may influence the hepatic disposition of drugs. Fenyves et al. (1993) have also postulated that the disposition of propranolol in cirrhosis is dependent, in part, on capillarization and intrahepatic shunts, oxygen delivery, and possibly acinar heterogeneity. Furthermore, induced pharmacokinetic changes due to liver disease also depend on drug properties (low or high hepatic extraction, extent of protein binding, etc.) as well as on the severity of liver disease (Rodighiero, 1999). At present, there appears to be no work quantitatively relating hepatic pharmacokinetics to both the drug properties and hepatic pathophysiology (Morgan and McLean, 1995).

In this work, we attempt to define the hepatic pharmacokinetics of a range of cationic drugs in perfused rat livers that had been treated with CCl₄, alcohol, or were untreated (controls) using the multiple indicator dilution technique (MID). The pharmacokinetics was then related to drug properties (log P_app and pK_a) and changes in liver pathophysiology. The latter changes were estimated in terms of: 1) physiological alterations in sinusoidal morphology as expressed by changes in the mean and normalized variance of sinusoidal transit times; 2) a degree of fibrosis index determined by computer-assisted image analysis; 3) the level of intrahepatocellular proteins [e.g., ₁-acid glycoprotein (AAG) and microsomal protein]; and 4) cytochrome P450 (P450) concentrations. We anticipated that the fibrosis index (FI) might be a surrogate measure for changes in plasma membrane permeability due to a decreased diffusivity or an increased diffusion path length. As variations in plasma AAG content in various diseases have been used to account for the fraction unbound in plasma of many drugs (Rowland and Tozer, 1995), we felt it was important to monitor and examine the role of AAG and other proteins in the intrahepatic disposition of drugs. The relationships between the pharmacokinetic parameters (e.g., permeability-surface area product) of a given solute and these histopathological data as well as the physicochemical properties of a solute (e.g., lipophilicity) were then defined by using stepwise regression analysis. Cationic drugs were studied in this work for four reasons: 1) they constitute 70 to 80% of all drugs; 2) they often have a limited therapeutic ratio (e.g., cardiovascular, analgesic, and psychotherapeutic drugs); 3) they constitute the majority of drugs showing a high first-pass effect; and 4) structure-hepatic disposition relationships of model cationic drugs with varying lipophilicity exist for normal rat livers.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals. Atenolol, antipyrine, prazosin, labetalol, propranolol, diltiazem, ethanol, and carbon tetrachloride were obtained from Sigma-Aldrich (St. Louis, MO) and used without any further purification. [U-¹⁴C]Sucrose and [³H]water were obtained from Amersham Biosciences UK, Ltd. (Little Chalfont, Buckinghamshire, UK).

CCl₄-Induced Liver Disease Rat Model. The CCl₄ rat model was established following a procedure outlined in detail previously (Proctor and Chatamra, 1982). Briefly, male Wistar rats (weighing approximately 150 g) were placed in all-wire mesh cages in groups of eight each (two groups) and given sodium phenobarbital (35 mg/dl) (Biotech International Ltd., Rocklea, Australia) in their drinking water ad libitum; the first dose of CCl₄ was given 10 days later. Group 1 animals were given CCl₄ (Sigma-Aldrich) once a week for 12 weeks. CCl₄ was dissolved in corn oil and given by intragastric gavage without anesthetic. The initial dose of CCl₄ was 0.04 ml. Body weight was monitored daily, and each subsequent dose was adjusted by the weight loss associated with the preceding dose. All doses were multiples of the original dose, and the total volume of dose was 1.0 ml. Group 2 animals were treated identical to group 1, except that they were given a 1.0-ml dose of corn oil weekly without CCl₄.

Alcohol-Induced Liver Disease Rat Model. The alcohol rat model was developed by using a procedure described previously (Takeyama et al., 1996). Briefly, male Wistar rats (weighing approximately 150 g) were placed in all-wire mesh cages in groups of eight each (two groups). Group 1 rats were single fed for 12 weeks on a nutritionally liquid high-fat alcohol diet (Dyets, Bethlehem, PA) containing ethanol (36% of the total caloric intake). The intake volume of those liquid diets single fed to the rats was 100 ml/day. Group 2 animals were treated identical to group 1, except that they were given a control diet (Dyets) that contained no ethanol.

Liver Biochemistry Determinations. Serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were measured on a Hitachi 747 analyzer (Hitachi Ltd., Tokyo, Japan) to assess the severity of disease by taking blood samples from tail vein every fortnight.

In Situ Rat Liver Perfusions. The in situ perfused rat liver preparation used in this study has been described previously (Cheung et al., 1996; Hung et al., 2001). Briefly, the portal vein was punctured with a 23-gauge needle and connected to a hydrostatic manometer to measure the in vivo portal venous pressure. The bile duct was cannulated with PE-10 (Clay Adams, Parsippany, NJ). The portal vein was then cannulated using a 16-gauge intravenous catheter, and the liver was perfused via this cannula with 2% bovine serum albumin MOPS buffer, which contains 15% (v/v) prewashed canine red blood cells (School of Veterinary Sciences, The University of Queensland, Brisbane, Australia) at pH 7.4, and oxygenated using a silastic tubing lung ventilated with 100% pure oxygen (BOC Gases, Brisbane, Australia). The perfusion system used was nonrecirculating and used a peristaltic pump (Cole-Parmer Instrument Co., Chicago, IL). After perfusion was effected, the animals were sacrificed by thoracotomy, and the thoracic inferior vena cava was cannulated with PE-240 (Clay Adams). The animal was placed in a temperature-controlled perfusion cabinet at 37°C. Liver viability was assessed by macroscopic appearance, bile production, oxygen consumption, and perfusion pressure as described by Cheung et al. (1996).

Bolus Studies. To reflect in vivo microcirculation, perfusion was initially conducted at a rate of 15 ml/min and was then adjusted to a pressure similar to the in vivo portal venous pressure (Varin and Huet, 1985). After a 10-min perfusion stabilization period, aliquots (50 µl) of perfusion medium containing a particular cationic drug [0.06-0.11 µmol of atenolol/propranolol, 0.09-0.15 µmol of antipyrine, 0.04-0.08 µmol of prazosin/labetalol, or 0.04-0.07 µmol of diltiazem as determined by high-performance liquid chromatography (HPLC) assay], [³H]water (3 × 10⁶ dpm), and [U-¹⁴C]sucrose (1.5 × 10⁶ dpm) was injected into the liver with outlet samples collected via a fraction collector over 4 min (1 s × 20, 4 s × 5, 10 s × 5, and 30 s × 5). In each liver, a maximum of six injections was made with the order of injection randomized and no repeat of the same injection in the same rat. A stabilization period of 10 min was afforded between two injections. The total perfusion time for each liver was less than 2 h. These samples were centrifuged at 1500g (25°C) for 3 min, and aliquots (100 µl) of supernatant (containing [³H]water and [U-¹⁴C]sucrose) were taken for scintillation counting using a MINAXI Beta TRI-CARB 4000 series liquid scintillation counter (Packard BioScience, Meriden, CT). The residue was vortexed and prepared for HPLC analysis to determine the outflow concentration of each cationic drug.

Analytical Procedure. The HPLC method used in this work has been described and validated previously (Hung et al., 2001). The within-day coefficients of variation for all the drugs studied were in the range of 0.6 to 4.4% (n = 3).

Histopathological Examination. For light microscopy, three to five slices of each liver were fixed in 10% neutral buffered formalin then routinely embedded in paraffin. Five-micrometer sections were prepared and stained with H&E for histopathological examination.

Quantitation of Fibrosis. Fibrous tissue was differentially stained pink in 5-µm paraffin sections by the hematoxylin van Gieson method (a solution of 1% aqueous acid fuchsin, saturated aqueous picric acid, and concentrated hydrochloric acid). The degree of fibrosis was quantified by computer-assisted image analysis (Image-Pro Plus version 3.0 for Windows; Media Cybernetics, Inc., Silver Spring, MD). For each rat, the area of stained fibrous tissue in five randomly selected fields was measured, and the average was expressed as fibrosis per unit area of liver tissue (termed the FI) as described by MacIntosh et al. (1992).

Determination of Hepatic Tissue AAG Level. Livers were harvested from the rats after MID studies and perfused with a mixed solution of calcium and magnesium-free Hanks' balanced salt solution, 5 mM EDTA, and 10 mM HEPES (all from Sigma-Aldrich) at 10 ml/min for 10 min to remove the protein and blood from sinusoidal beds. The liver (approximately 1 g) was then homogenized in 1 ml of MOPS buffer using tissue blender and centrifuged at 3000g for 10 min. A quinaldine red (Sigma-Aldrich) fluorometric titration method based on a method described previously was used for tissue AAG level analysis (Imamura et al., 1994).

Determination of Hepatic Cytoskeleton Residue (CR), Microsomal Protein (MP), and P450 Concentrations. The liver (approximately 1 g) was homogenized in 2.5 ml of ice-cold 0.25 M sucrose containing 50 mM Tris-HCl buffer (pH 7.4) for 3 to 5 min. The homogenates were centrifuged at 5,000 rpm for 20 min, and the pellets resuspended in 2.5 ml of Tris buffer as the cytoskeleton fraction. The supernatant (approximately 1 ml) from the 5,000 rpm centrifugation was centrifuged again at 50,000 rpm for 1 h. The resulting pellets were resuspended in 2.5 ml of Tris buffer and used as the microsomal fraction. The CR and MP concentrations in the respective fractions were determined by the method of Lowry et al. (1951). P450 content in MP was estimated from the dithionite-reduced difference spectrum of CO-bubbled samples using the molar extinction difference of 104 mM¹ cm¹ in absorption at peak position (about 450 nm) (Matsubara et al., 1976).

Investigation of Cationic Drug Binding to Hepatic Components. These experiments were carried out in 1) blank MOPS buffer (pH 7.4); 2) buffer containing 0.35 mg/ml rat AAG (Sigma-Aldrich); 3) buffer containing 0.35 mg/ml MP from normal livers; or 4) buffer containing 0.35 mg/ml CR from normal livers. The unbound fraction of cationic drug in each buffer solution (f_uT) was estimated using an ultrafiltration method. A known concentration of the cationic drug stock solution was added to 500 µl of each buffer solution to make a final concentration of 0.05 µM and placed in a centrifugal filter device (Microcon YM-30; Millipore Corp., Bedford, MA) and then centrifuged at 3000g for 10 min. The ultrafiltrate (in triplicate) was assayed by HPLC. The f_uT was determined as the ratio of the free concentration to total concentration of solute. Estimation of association constant K_a,T (K_a,AAG, K_a,MP, or K_a,CR) for binding to a given hepatic component (MP, AAG, or CR) to an individual cation was deduced using f_uT = 1/(1 + K_a,T · [C_T]) where [C_T] is the concentration of one of these hepatic components.

Hepatic Extraction Ratio and Mean Transit Time. Nonparametric estimates of hepatic extraction ratio (E), mean transit time (MT), and normalized variance (CV²) were determined from the outflow concentration versus time profiles for the model cationic drugs as described previously (Hung et al., 2001). The recovery for each drug is defined by its availability, which is defined by 1  E (Roberts and Rowland, 1986).

Estimation of Pharmacokinetic Parameters by Modeling and Data Fitting of the Outflow Concentration-Time Profiles of Extracellular and Cellular References after Impulse Dosing. A mixture of two inverse Gaussian density functions with correction for catheter effects was used to estimate the extracellular space (V_B, determined by [U-¹⁴C]sucrose) (Weiss et al., 1997). A barrier-limited plus space-distributed liver model with correction for catheter effects was used to estimate the total water volume (V_w, determined by [³H]water), V_w then being used to estimate the cellular water volume, V_C, defined as (V_w  V_B) (Weiss et al., 2000).

Modeling and Data Fitting of the Outflow Concentration-Time Profiles of Model Cationic Drugs. A heterogeneous (barrier-limited and space-distributed) transit time model (the "two-phase stochastic model") was used to estimate the pharmacokinetic parameters of hepatocellular influx, efflux, binding, and elimination for the permeating solutes (Hung et al., 2001). Briefly, the model (Fig. 1) assumes unbound drug transfer across the permeability barrier (plasma membrane) with influx and efflux rate constants k_in and k_out, respectively. An apparent distribution ratio (K_v) for an unbound solute between the cellular and extracellular space is defined by k_in/k_out. K_S is the equilibrium amount ratio for slowly equilibrating binding sites as defined by an on and off rate constant (Hung et al., 2001). K_S for a given drug, in theory, should be related to the product of the association constant K_a,p of the drug for protein p and the concentration of protein [P] i.e., K_S = K_a,p · [P] (Rowland and Tozer, 1995). The permeability-surface area product (PS) is given by k_in · V_B/f_uB where V_B is the extracellular space and f_uB is the fraction unbound in the perfusate, estimated previously by Hung et al. (2001). The intrinsic clearance, CL_int, was estimated as the product of the elimination rate constant k_e and the cellular distribution volume for water, V_C, following the approach of Pang et al. (1995). The stochastic approach represents the transit of a molecule through the organ as a series of sojourns in one of the two regions described by density functions. The sojourn time distribution f_y(t) of a molecule after a single excursion in the cellular space for the resulting two-compartment cell model can be obtained by standard methods in the Laplace domain, _y(s) = L¹[f_y(t)], as described earlier (Weiss et al., 2000; Hung et al., 2001).

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Typical outflow profiles for model cationic drugs in various animal models (data weighted 1/yobs2) in the regressions. A, atenolol (most hydrophilic) and B, diltiazem (most lipophilic). The open circles () represent the normal group. The filled triangles () represent the alcohol-treated group. The filled circles () represent the CCl4-treated group. The lines are the fits of the profiles to a heterogeneous (barrier-limited and space-distributed) transit time model. Inset, KS is the equilibrium amount ratio characterizing the slowly equilibrating binding sites; fuB is the fraction of drug unbound in perfusate; VB is the extracellular volume (vascular + Disse space); VC is the cellular water volume; and kin, kout, and ke represent the permeation, efflux, and elimination rate constant, respectively (modified from Hung et al., 2001).

Statistical Analysis. All data are presented as mean ± standard deviation unless otherwise stated. Stepwise regression analysis was performed using the program SPSS 10.1 for Windows (SPSS, Inc., Chicago, IL). Statistical analysis was performed with two-way analysis of variance, Student's t test, and regression analysis where appropriate. A p < 0.05 was taken as significant. Linear regression equations have been only considered when r² > 0.5.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Liver Biochemistry and Physiology. The CCl₄-treated group has a significantly lower perfusion rate (0.75 ± 0.02 ml · min¹ · g¹ of liver) than that of the alcohol-treated (0.89 ± 0.04 ml · min¹ · g¹ of liver, p < 0.05) and normal control (1.23 ± 0.09 ml · min¹ · g¹ of liver, p < 0.01) groups. Accordingly, the CCl₄-treated group had a significantly higher in vivo perfusion pressure (1.02 ± 0.08 cm of H₂O) than that of the alcohol-treated (0.92 ± 0.11 cm of H₂O, p < 0.05) and normal control (0.75 ± 0.09 cm of H₂O, p < 0.01) groups. The CCl₄-treated group also had a significantly lower bile flow (0.61 ± 0.05 µl · min¹ · g¹ of liver) than that of the alcohol-treated (0.83 ± 0.11 µl · min¹ · g¹ of liver, p < 0.05) and normal control (1.41 ± 0.15 µl · min¹ · g¹ of liver, p < 0.01) groups. The normal control group had significantly higher liver oxygen consumption (1.13 ± 0.19 µmol · min¹ · g¹ of liver) than that of CCl₄-treated (0.93 ± 0.11 µmol · min¹ · g¹ of liver, p < 0.05) and alcohol-treated (0.97 ± 0.06 µmol · min¹ · g¹ of liver, p < 0.05) groups. The liver biochemistry differed greatly between the three animal models. The CCl₄-treated group has a significantly higher mean ALT (1,027.40 ± 778.60 IU/l) and AST (740.11 ± 301.64 IU/l) level than those of the alcohol-treated (ALT, 109.60 ± 13.10 IU/l and AST, 90.94 ± 14.19 IU/l) and normal control (ALT, 54.00 ± 7.70 IU/l and AST, 46.39 ± 7.50 IU/l) groups (p < 0.001). Two other controls used in this study for completeness, CCl₄ control (phenobarbitone treatment only) and alcohol control (liquid diet only), yielded experimental parameters not significantly different than those for the normal control group (data not shown).

Histopathology. Sections of normal rat livers showed typical architecture under light microscopy, i.e., cords of hepatocytes one cell wide radiating out from hepatic venules toward portal tracts. Livers from rats fed an alcohol diet showed perivenular macrovesicular steatosis with minor fibrosis. The CCl₄-treated livers showed severe fibrosis, and fibrosis was perivenular with venous-venous fibrous linkage. FI, estimated by computer-assisted image analysis, for each of the animal models is shown in Table 1. The CCl₄-treated group had a significantly higher FI value than the alcohol-treated (p < 0.05) and control groups (p < 0.001).

Hepatic AAG, MP, and P450 Concentrations. Table 1 also shows that the alcohol-treated group has a significantly higher hepatic AAG, MP, and P450 content than those of the control (AAG, p < 0.05; MP, p < 0.05; and P450, p < 0.01) and CCl₄-treated groups (AAG, p < 0.01; MP, p < 0.001; and P450, p < 0.001). The AAG, MP, and P450 levels found in the CCl₄ control (phenobarbitone treatment only) and alcohol control (liquid diet only) were not significantly different than those for the normal controls.

Cationic Drug Binding to Hepatic Components. This study shows that the unbound fraction (f_uT) of cationic drug in the blank MOPS buffer containing MP (f_u,MP) or AAG (f_u,AAG) decreases directly with the lipophilicity of drug. Values of 0.98 and 0.96 for atenolol, 0.97 and 0.96 for antipyrine, 0.43 and 0.31 for prazosin, 0.25 and 0.17 for labetalol, 0.11 and 0.06 for propranolol, and 0.09 and 0.05 for diltiazem were found for MP (0.35 mg/ml) and AAG (0.35 mg/ml), whereas f_u,CR for various cationic drugs in the buffer of cytoskeleton residue remain relatively constant (0.97). The relationships between log f_uT and log P_app are 1) MP, log f_u,MP = 0.098  0.307 log P_app (p = 0.001, r² = 0.953, and n = 6) and 2) AAG, log f_u,AAG = 0.012 + 0.8 log P_app (p = 0.0001, r² = 0.999, and n = 6). Although both MP and AAG bind lipophilic cationic drugs (prazosin, labetalol, propranolol, and diltiazem) in hepatocytes, MP may be more dominant given its 10-fold greater concentration in the hepatocyte than AAG for all models (Table 1).

Hepatic Extraction Ratio and Mean Transit Time. Table 2 shows that hepatic extraction ratio (E) increases with the lipophilicity (log P_app) of drug in various animal models. The CCl₄-treated group has a significantly lower E than that of the normal control and alcohol-treated groups for the same lipophilic drug but not for the same polar drug. Interestingly, the alcohol-treated group has a significantly higher E than that of the normal control and CCl₄-treated groups for the same lipophilic drug but not for the same polar drug. Stepwise regression analysis examining relationships between hepatic extraction ratio (E) with lipophilicity of drug (log P_app) and P450 yielded a high correlation of predicted and observed E (Fig. 2A; log E = 0.35 + 0.07 P450 + 0.027 log P_app; p = 0.0001, r² = 0.932, and n = 30). FI was excluded from this regression as nonsignificant. It is also poorly correlated with P450 (FI = 4.082  3.186 P450, r² = 0.174, and n = 30).

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Comparison of predicted and observed pharmacokinetic parameters of the model cationic drugs in various animal models. A, E; B, PS; C, tissue-binding constant, Ks; D, CLint; and E, partition ratio of influx and efflux rate constant (Kv = kin/kout). The open circles () represent the normal group. The filled triangles () represent the alcohol-treated group. The filled circles () represent the CCl4-treated group.

Modeling and Data Fitting of the Outflow Concentration-Time Profiles. Figure 1, A and B, shows typical logarithms of normalized outflow concentration versus time data for atenolol (most hydrophilic) and diltiazem (most lipophilic) in various animal models. Also shown are the corresponding regression line fits using a heterogeneous (barrier-limited and space-distributed) transit time model and a data weighting of 1/y_obs². All MID data appeared to be adequately fitted by the models. The estimated model parameters for extracellular volume V_B and cellular water volume V_C were 1) V_B = 0.49 ± 0.15 ml g¹ of liver, V_C = 1.30 ± 0.44 ml g¹ of liver for the normal control group (n = 6); 2) V_B = 0.53 ± 0.19 ml g¹ of liver, V_C = 1.38 ± 0.39 ml g¹ of liver for the alcohol-treated group (n = 6); and 3) V_B = 0.57 ± 0.21 ml g¹ of liver, V_C = 1.34 ± 0.41 ml g¹ of liver for the CCl₄-treated group (n = 6). Although the V_B obtained for the CCl₄ control (phenobarbitone treatment only) and alcohol control (liquid diet only) were similar to the normal controls, the V_C for the CCl₄ control was larger (data not shown).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

This work has shown that the pharmacokinetics of cationic drugs in cirrhotic livers can be related to the liver histopathology and solute physicochemical properties. Care was taken to accurately define the vascular dispersion in the liver (Roberts and Rowland, 1986) using a heterogeneous (barrier-limited and space-distributed) transit time model as described in our previous work (Hung et al., 2001). In this work, robust and good fits were also found for cationic drugs in diseased rat livers (Fig. 1, A and B). An implicit assumption in this model is linearity of kinetics as defined by hepatocellular unbound concentrations being below the Michaelis constants for the enzymes metabolizing each drug. The modeling has also assumed that a rapid equilibrium occurs between erythrocytes and the plasma, and that the efflux of the cationic drugs from erythrocytes is sufficiently rapid so as to not lead to an erythrocyte carriage effect (Goresky et al., 2000).

The experimental parameters associated with the normal and CCl₄-treated perfusion studies, reflecting liver viability, are also comparable with previous reports (Varin and Huet, 1985; Hung et al., 2001). CCl₄ produces acute hepatocellular injury with centrilobular necrosis and stenosis (Hong et al., 2000). In contrast, alcohol produces chronic hepatocellular injury with inflammation and perivenular macrovesicular steatosis (Sherlock, 1993; Takeyama et al., 1996). These observations were confirmed in this work.

This work has shown that hepatic extraction ratio E and primary parameters PS, K_S, and CL_int depend on the lipophilicity of the cationic drug, irrespective of the disease state (Table 3). This work has also shown that a good correlation exists between the predicted and observed values for E and the parameter values (Fig. 2). E or availability F (= 1 E) can be related to f_uB, PS, perfusate flow rate, and CL_int in a number of models describing hepatic elimination (Pang and Rowland, 1977; Roberts and Rowland, 1986; Roberts and Anissimov, 1999; Roberts et al., 2000), including the dual inverse Gaussian model (Weiss et al., 1998). These models show E is related to f_uB, PS, CL_int, and perfusate flow rate, which are, in turn, related to log P_app (f_uB, PS, and CL_int), FI (PS), and P450 (CL_int). The good relationship between log E with log P_app and P450 (Fig. 2A) is therefore consistent with the E being defined by PS, CL_int, etc. A high correlation was found for E with log P_app and P450 (r² = 0.932) with the stepwise regression excluding FI as a determinant.

The underlying relationship between PS with FI and log P_app is consistent with the definition of PS, namely PS = K_mD_mS/h_m, where K_m is the plasma membrane-unbound perfusate concentration coefficient, D_m is the solute diffusivity in the plasma membrane, S is the surface area, and h_m is the membrane path length for diffusion. Hence, in the absence of transporters, K_m is defined by the solute lipophilicity and log P_app as a surrogate. An impaired uptake of solute across the capillarized endothelium in cirrhosis (Huet et al., 1985) is consistent with a slower solute diffusivity and longer diffusion path length as a consequence of collagenization of the Disse space. The FI is a surrogate measure for changes in D_m and h_m as a determinant of PS.

The extent of solute uptake into normal and cirrhotic livers depends on both tissue binding as well as ion trapping by organelles such as the mitochondria and lysosomes. The main measure of tissue uptake is slow binding as expressed by the equilibrium amount ratio for slowly equilibrating binding sites, K_S. In this work, we found that the alcohol-treated group has the highest intrahepatocellular AAG level and MP concentration, whereas the CCl₄-treated group has the lowest AAG level and MP concentration among various animal models (Table 1). One source for these binding sites is the soluble microsomal proteins, MP. The regression between log K_S with the determinants of the contribution of MP to K_S, its association constant K_a,MP (defined from in vitro binding, Table 2), and the concentration of MP in the liver, i.e., K_S,MP = K_a,MP · MP is log K_S = 0.027 + 0.618 log K_a,MP + 0.295 log MP (p = 0.001, r² = 0.947, and n = 30). Earlier, it was shown that the unbound fraction (and thus, K_a,p) of cationic drug in perfusate or buffers containing MP or AAG also correlated with the octanol-water partition coefficient log P_app. Hence, the derived expression log K_S = 0.017 + 0.242 MP + 0.022 log P_app (p = 0.0001, r² = 0.926, and n = 30; Fig. 2C) suggests that hepatic tissue binding can be predicted from a knowledge of the lipophilicity of the solute as defined by log P_app and binding protein content as defined by MP. As discussed earlier, AAG, one of the acute-phase proteins that increases in various conditions including inflammatory diseases (Brinkman-van der Linden et al., 1996), has concentrations in the liver too low to make a substantial contribution to K_S.

The K_v, or distribution ratio of a solute between the tissue and vascular space, is mainly dependent on ion trapping in organelles and asymmetric transport across the plasma membrane (Hung et al., 2001). We have previously shown that K_v increases with drug pK_a and an ion-trapping effect for cationic drugs related to the volume of and intraorganelle pH in the hepatic cytoplasm, mitochondria, and lysosomes (Hung et al., 2001). We have now shown that K_v is related to both pK_a and FI, enabling a lesser dependence of K_v on pK_a for the alcohol-treated and, even less so, for the CCl₄-treated group to be described. The decreased dependence is most likely due to a higher intracellular pH and a reduction in the ion trapping of the cations as a consequence of a reduction in ionization. Hence, the higher cytosolic pH in cirrhosis rats (7.45) than in normal rats (7.27), due to the activation of hepatocellular Na⁺/H⁺ exchange in cirrhosis (Elsing et al., 1994), would explain a lower K_v for the CCl₄-treated group relative to the normal control group as well as a loss of dependence of K_v on pK_a for the CCl₄-treated group. How fibrosis affected the organelle volumes (and potential ion-trapping volume) is less certain. For instance, CCl₄-induced cirrhosis reduces rat liver mitochondrial volume (Krähenbühl et al., 2000) (normally about 20% of hepatocyte volume; Hung et al., 2001) but increases the activity and numbers of lysosomes (Dufour et al., 1994) (normally about 1% of hepatocyte volume; Hung et al., 2001).

In this analysis, we estimated the intrinsic metabolic clearance CL_int as a product of k_e and V_C. This approach, used previously by Pang et al. (1995), ensures that solutes ion- trapped in organelles or bound to rapidly equilibrating binding sites are not determinants of CL_int. The good correlation of observed and predicted CL_int (Fig. 2D) derived from the expression log CL_int = 1.938 + 0.31 P450 + 0.112 log P_app (p = 0.011, r² = 0.799, and n = 30) suggests that intrinsic elimination clearance of a solute in the diseased liver can be predicted from lipophilicity of the solute and P450 activity of the liver tissue. It should be noted that P450 consists of a number of isozymes that, in response to a stimulant, may be differentially expressed. Better correlations may therefore be expected by relating CL_int to the individual isozymes responsible for the metabolism of a given drug. Relevant to this work, Kono et al. (1999) showed chronic exposure to alcohol caused induction of CYP3A, CYP2A12, CYP1A, CYP2B, and CYP2E1 expression in mouse livers. Bastien et al. (2000) showed CCl₄-induced cirrhosis was associated with reduction of CYP1A, CYP2C, CYP2E1, and CYP3A expression in rat livers. Given that P450 accounted for a low and variable fraction of MP (0.076 normal group; 0.093 alcohol-treated group; and 0.03 CCl₄-treated group), caution should be exercised in using MP as a surrogate for P450. It is therefore apparent, as noted by Obach (1997) and McLure et al. (2000), that much of the drug binding to MP is nonspecific and not necessarily a determinant of CL_int.

In summary, the disposition of cationic drugs may be directly related to the extent of fibrosis caused by the liver disease and the physicochemical properties of the drugs. The changes in the extraction ratio E are consistent with a reduction in PS, CL_int, and perfusate flow rate, each of which define E. These reductions arise from alterations in sinusoidal morphology and in the hepatic microcirculation and a decreased enzymatic activity due to loss of normal metabolic zonation and microsomal activity in the liver. In addition, hepatic distribution is reduced in cirrhosis due to reduced tissue binding (lower K_S) as a consequence of a lower synthesis of intrahepatocellular proteins as well as a loss of an ion-trapping effect (lower k_in/k_out) due to an elevation of intracellular pH and changes in organelle volumes. Drug lipophilicity is a key determinant of cation drug extraction. There were good correlations between the predicted and observed pharmacokinetic parameters (E, PS, K_S, CL_int, or K_v) of the model cationic drugs based on nature of drug (log P_app or pK_a) and the histopathological results (FI, MP, and P450). This prediction of E using various determining variables such as log P_app and P450 is likely to be improved by the use of larger data set (drugs and severity of disease state). These results should be relevant in better understanding how changes in fibrosis-induced hepatic diseases quantitatively affect hepatic drug pharmacokinetics and in assisting preclinical drug development through predictions of the first-pass effect.