Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

There is growing interest in the use of the dual perfused liver preparation to explore the relative contributions of the hepatic artery (HA) and portal vein (PV) to events occurring within the liver, particularly given that intestinally absorbed compound enters the liver only via the PV, whereas systemically administered compound perfuses the liver via both vascular routes. One aspect of this dual input is the anatomical and functional relationships between these two vessels. Observations obtained from such efforts favor the idea that there are both common and separate channels within the liver. Although the majority of the sinusoids are perfused by the mixed blood, a small fraction of the vascular bed remains separate (about 10%; Field and Andrews, 1968; Ahmad et al., 1984; Sahin and Rowland, 1998a) and receives 17% of the arterial input (Sahin and Rowland, 1998a). Another aspect is the influence of route of input on the hepatic elimination of drugs.

Although various compounds, including hormones (e.g., insulin), dyes (e.g., indocyanine green), vitamins (e.g., vitamin D₃), and drugs (e.g., lidocaine, meperidine) have been used to investigate the influence of route of input on hepatic elimination, reports from these studies are currently controversial. Some authors (Ahmad et al., 1984; Gardemann et al., 1991; Meyerholz et al., 1991; Zimmermann et al., 1992; Pang et al., 1994; Shiotani et al., 1995; Iwasaki et al., 1998) report a difference between arterial and venous administrations, whereas others (Brauer et al., 1959; Hollenberg and Dougherty, 1966; Lautt and Daniels, 1983; Lautt et al., 1984; Gascon-Barre et al., 1988; Ikeda et al., 1993; Kuan et al., 1996) have observed no such difference. Nevertheless, choice of compound and method of assessment may be important issues when attempting to demonstrate whether any difference in recovery occurs with respect to site of vascular input.

The current study was designed to critically evaluate the hepatic disposition and elimination of diazepam in the in situ dual perfused rat liver. This compound was chosen for two reasons. First, being a drug of high intrinsic clearance, with an extraction ratio after PV input of almost 1 in the absence of albumin, its plasma binding protein (Rowland et al., 1984), diazepam is a sensitive marker to explore the influence of route of vascular input on effluent fractional recovery. Second, the greater sensitivity of cellular uptake to protein binding than cellular permeability (Diaz-Garcia et al., 1992) also allows an assessment on the influence of altered binding on the uptake kinetics for the two routes of input.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals

Iodine-125-labeled albumin [¹²⁵I-human serum albumin (HSA), 1.02 mCi/mg], was purchased from ICN Biomedicals (Costa Mesa, CA) and [2-¹⁴C]diazepam (57 mCi/mmol) from Amersham International (Horsham, UK). Diazepam was obtained from Sigma Chemical (St. Louis, MO). HSA was a gift from Pharmacia Upjohn AB (Uppsala, Sweden). All other chemicals were of analytical grade and were obtained from commercial sources.

Perfusion Procedure

All animals were handled in compliance with the UK Home Office guidelines. They were fed on a normal laboratory diet with free access to drinking water and kept under a 12-h light/dark cycle in a temperature-controlled environment.

The single-pass dual perfused in situ rat liver system, with male Sprague-Dawley rats (330-380 g; wet liver weight, 11-15 g) was essentially the same as that described previously (Sahin and Rowland, 1998b). Krebs-bicarbonate buffer containing HSA and/or drug (concentrations specified below) was used as the perfusion medium. Under anesthesia, the bile duct was cannulated, and loose ligatures were placed around the PV ensuring exclusion of the HA. The abdominal contents were displaced to the animal's right, and branches of the celiac artery (i.e., left gastric and lineal arteries) were tied very close to their junctions to the celiac artery. The gastroduodenal artery (branch of the common hepatic artery) was also ligated. After all these steps only the HA was left patent. The PV was cannulated with a 16GA catheter (Argyle Medicut, o.d. 1.7 mm × 45 mm), and the perfusion was started immediately at a flow rate of 12 ml/min. Exsanguination of the liver was facilitated by inserting a tubing into the vena cava via the right atrium. The HA was cannulated with an 18GA (Argyle Medicut, o.d. 1.3 mm × 45 mm) or 20GA (Argyle Medicut, o.d. 1.1 mm × 45 mm) catheter, and the second perfusion started immediately at a flow rate of 3 ml/min. Liver viability was routinely assessed by monitoring bile flow, perfusate recovery, and arterial perfusion pressure, and by macroscopic appearance.

Injection Preparation

One drug ([¹⁴C]diazepam) and one extracellular marker (¹²⁵I-HSA) were injected together as a bolus [50 µl in saline containing HSA (at a concentration equal to that in the perfusate) and unlabeled drug (1 mg/l diazepam)]. Doses used were as follows: 0.18 to 0.26 µCi of [¹⁴C]diazepam in the presence of 1% HSA and 0.90 to 0.95 µCi in the presence of 0.25% HSA; and 0.06 to 0.27 µCi of ¹²⁵I-HSA.

Experimental Procedure

The disposition kinetics of diazepam was investigated under both unit impulse-response and steady-state conditions. Preliminary experiments indicated that up to 10 min was required to achieve steady state with regard to diazepam extraction. Therefore, the fraction escaping unchanged was determined from the total outflow samples collected between 10 and 20 min for PV and 15 and 25 min for HA infusion. After stabilization of the exposed liver with drug- and protein-free perfusate, the animals were allocated to one of two groups, A or B.

In group A (n = 5), two different perfusates each containing 1 mg/l diazepam were used; one was protein-free and the other contained 0.25% HSA. In the absence of protein, diazepam infusion was alternated between the PV and HA inputs; diazepam was delivered into the liver initially via the PV for 20 min and then via the HA for 25 min. During diazepam infusion via one of the inputs (e.g., PV) the alternate vessel (e.g., HA) was perfused with drug-free perfusate. Subsequently, the perfusion medium was changed, and the liver was perfused with diazepam in the presence of 0.25% HSA. The infusion period and order of diazepam delivery to the liver were the same as in the absence of protein; first PV and then HA. During constant perfusion with unlabeled diazepam in 0.25% HSA, a bolus (50 µl) containing [¹⁴C]diazepam and ¹²⁵I-albumin was injected into the liver via the vessel in which unlabeled diazepam was infused. Immediately after an injection, the total effluent from the liver was automatically collected every 2 s for 2 min, and thereafter (into silanized test tubes) at increasing time intervals up to 6 min. Perfusate samples were also collected before and after the bolus injection to determine the extraction of unlabeled diazepam at steady state. In group B (n = 6), the design of the experiment was the same as with group A except a different protein concentration (i.e., 1% HSA) was used.

Radiochemical Analysis

The activities of [¹⁴C]diazepam and ¹²⁵I-albumin in effluent samples were determined by radiochemical analysis after the addition of 4 ml of liquid scintillation fluid, with the results expressed as dpm.

HPLC Assay of Diazepam

Diazepam was analyzed by HPLC using a modification of the method of Raisys et al. (1980). The HPLC system consisted of a Kontron model 420 HPLC pump (Zurich, Switzerland), which delivered mobile phase (acetonitrile:water with 1% triethylamine adjusted to pH 3.0 with 85% orthophosphoric acid; 50:50, v/v) at a flow rate of 1 ml/min; an autosampler (Kontron model 360); and a Jasco model 100-IV variable wavelength UV detector (Tokyo, Japan). A Newguard column (Brownlee Labs, UK) with a disposable cartridge (Newguard RP-18, 15 × 3.2 mm, 7 µm) was also used to protect the analytical column. Separation of diazepam was performed at ambient temperature on a Spheri-5, RP18 cartridge column (25 cm × 4.6 mm; 5 µm), with detection at 254 nm. Quantitation of diazepam was determined by peak area ratio of drug to internal standard (prazepam) with reference to an appropriate calibration curve. The retention times for diazepam and prazepam were 9 and 17 min, respectively.

When the concentration of diazepam in the outflow samples was high (i.e., in the presence of HSA in the perfusate), samples were measured directly after precipitation of HSA with acetonitrile. Briefly, 75 µl of internal standard solution (prazepam; dilution of 2 mM stock solution) and 500 µl of acetonitrile were added to 500 µl of outflow sample. After vortex mixing (1 min) and centrifugation at 3500 rpm for 20 min, the supernatant was injected into the HPLC system. The calibration curves were constructed over the range of 100 to 1500 ng/ml. The determination coefficient (r²) for the calibration curve was always greater than 0.95.

Hepatic outflow samples obtained using protein-free perfusate contained concentrations of diazepam too low to be measured accurately using the method described above. Therefore, it was necessary to perform a simple extraction procedure before HPLC analysis. An aliquot (6 ml for PV and 8 ml for HA samples) of the outflow sample was transferred into a silanized test tube, and 50 µl of internal standard solution (prazepam; dilution of 2 mM stock solution) and 6 ml of hexane:ethyl acetate (8:2, v/v) were then added. The samples were mixed using a horizontal mixer for 45 to 60 min and then centrifuged (3500 rpm; 20 min). The lower aqueous layer was frozen by immersion in liquid nitrogen, and the upper organic layer was then transferred to another silanized tube and evaporated to dryness under a smooth nitrogen stream (35°C). The residue was reconstituted in mobile phase and vortexed immediately before injection into the HPLC system. Calibration curves were constructed over the range of 5 to 150 ng/ml. The determination coefficient (r²) for the calibration curve was always greater than 0.95.

To avoid nonspecific adsorption (binding) of diazepam, all glassware was silanized by immersion in 4% dichlorodimethylsilane in chloroform for 24 h followed by rinsing with methanol. Furthermore, protein (HSA) used throughout the study was defatted and freeze-dried.

Purification of HSA. The method for removal of fatty acids from serum albumin was a modification of that of Chen (1967). Two liters of HSA solution (20% w/v) was diluted with the twice the volume of distilled water and then placed in an ice bath. After mixing in activated charcoal powder (200 g; British Drug House, Poole, UK) and lowering the pH to 3.0 by addition of 1 N HCl, the solution was magnetically stirred for 1 h, at a constant temperature of 4°C. The charcoal suspension was then further diluted with distilled water to obtain a concentration of 4% HSA (w/v) and centrifuged (3000 rpm) at 4°C for 20 min to remove the majority of the charcoal. The remaining finely suspended charcoal was removed by filtration through a bed of Kieselguhr (British Drug House), followed by filtration through 0.8-, 0.45-, and 0.22-µm filters (Millipore, Watford, UK), respectively. The solution was then brought up to pH 7.4 by the addition of 1 M NaOH and stored at 4°C. Because the albumin solution was prepared once as a large batch, to ensure stability, its water content was removed by freeze-drying technique. The defatted freeze-dried albumin powder was stored at 4°C in tightly closed containers until used.

Equilibrium Dialysis. Equilibrium dialysis was performed using two-chambered Teflon dialysis cells (Dianorm, Switzerland), each of 1-ml capacity, separated by a membrane (Spectropor-2; Spectrum Medical Industries Inc., CA). To assess the unbound fraction of [¹⁴C]diazepam (fu), 1 ml of drug solution [unlabeled, 1 mg/l, and [¹⁴C]diazepam, 0.045 µCi/ml, in HSA-buffer solution; HSA concentrations (g/dl): 0, 0.1, 0.25, 0.5, and 1] was dialysed against 1 ml of protein-free perfusate at 37°C in a thermostatically controlled water bath, with the cell system continuously rotated at a fixed rate of 20 rpm. At the end of the equilibration time (4 h), the solution from each chamber was expelled by pushing air through the cells, and collected into test tubes. After the addition of 4 ml of liquid scintillation fluid (OptiPhase "Hisafe" II; Wallac, Gaithersburg, MD), ¹⁴C activity was determined in duplicate aliquots (200 µl) from each compartment by radiochemical analysis. The unbound fraction at equilibrium was calculated as follows:

(1)

where dpm_buffer and dpm_protein are the radioactivity (dpm) in the buffer and protein compartments.

Data Analysis

Bolus Experiments. After bolus administration the outflow data were transformed to frequency data [f(t), l/s] using the following equation:

(2)

where C(t) is the concentration of radioactivity associated with the unchanged compound, and Q is the total perfusate flow (ml/s). Time was taken as the difference between the midtime of the sampling interval and the lag time, corresponding to the delay in the nonhepatic region of the experimental system (catheter and tubing: PV, 2.5 s and HA, 2.7 s). The maximum frequency output (f_max) and the time reached (t_max) were determined experimentally.

(3)

(4)

(5)

(6)

(7)

(8)

(9)

Steady-State Experiments. The fractional hepatic recovery at steady state, F, is calculated from the following relationship (Ahmad et al., 1984):

(10)

where C_in and C_out are the perfusate drug concentrations entering and leaving the liver, and Q_in and Q_out are the influent (3 ml/min for HA and 12 ml/min for PV) and effluent (15 ml/min) flow rates, respectively. With diazepam infused into either the HA or PV, the rate of presentation is Q_HAC_in or Q_PVC_in as appropriate.

(11)

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Protein Binding. The fractions of diazepam unbound in fresh perfusate at different defatted HSA concentrations are summarized in Table 1. These results are in a close agreement with those of Diaz-Garcia et al. (1992). It has been shown previously that equilibrium is reached within 4 h, and that diazepam does not bind to the membrane and dialysis apparatus. Furthermore, the albumin-to-buffer volume ratio for diazepam 4 h after dialysis with different albumin concentrations was about 1, indicating that the aqueous volume shift was negligible.

Outflow Profiles. Figures 1 and 2 show representative frequency outflow versus time profiles for [¹⁴C]diazepam obtained in the presence of various HSA concentrations (0.25 and 1%), after bolus administration into the PV and HA, respectively. In all cases the shape of the outflow profile was dependent upon the albumin concentration used. In the presence of 1% HSA, the hepatic output profiles of [¹⁴C]diazepam appeared as a sharp peak (f_max = 0.0147 ± 0.0035 l/s at 12.2 ± 4.7 s for PV administration, and 0.0043 ± 0.0003 l/s at 19.7 ± 3.2 s for HA administration), which eluted over the first 25-s interval, followed by a slowly eluting tail. With a decrease in the perfusate protein concentration from 1 to 0.25%, the earlier first peak disappeared and the profiles displayed a unimodal characteristic with a later and much flatter peak (f_max = 0.0029 ± 0.0006 l/s at 60.1 ± 9.4 s for PV administration; 0.0012 ± 0.0001 l/s at 76.8 ± 15.5 s for HA administration). In contrast, ¹²⁵I-albumin displayed a unimodal outflow profile irrespective of both HSA concentration and route of input (Fig. 3). Regardless of both the HSA concentration (0.25 and 1%) and compound (¹²⁵I-albumin and [¹⁴C]diazepam), the outflow profiles after arterial input were flatter than those after venous input (e.g., for diazepam Fig. 1 versus Fig. 2, and for ¹²⁵I-albumin f_max, l/s: 0.045 ± 0.001 versus 0.098 ± 0.011 for 1% HSA; Fig. 3).

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Fractional rate of efflux of 14C radioactivity obtained in the presence of 0.25% () and 1% () HSA, after bolus injection of [14C]diazepam into the portal vein. Profiles are from an individual rat liver preparation, operating in the dual perfusion mode.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Fractional rate of efflux of 14C radioactivity obtained in the presence of 0.25% () and 1% () HSA, after bolus injection of [14C]diazepam into the hepatic artery. Profiles are from an individual rat liver preparation, operating in the dual perfusion mode.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Fractional rate of efflux of albumin obtained after bolus injection into the portal vein () and hepatic artery (). Profiles are obtained in the presence of 1% perfusate albumin concentration.

Albumin. Moment analysis results for ¹²⁵I-albumin are summarized in Table 2. Regardless of both route of input (HA and PV) and perfusate albumin concentration (0.25 and 1%), MTT after arterial administration was longer than that after venous administration (P < .05 and P < .001 for the corresponding albumin concentrations): the transit times also increased with an increase in the albumin concentration from 0.25 to 1% (P < .05 for PV and P < .01 for HA; Table 2). For a given protein concentration, the distribution volume (V_H) of albumin was slightly but significantly larger after arterial administration (P < .05 for 0.25% HSA and P < .001 for 1% HSA; Table 2). For a given route, the volume terms were increased with an increase in the perfusate albumin concentration (P < .05 for PV and P < .01 for HA). When administered via the PV, spreading (CV²) of labeled material (i.e., 0.55 ± 0.11 versus 0.50 ± 0.05) within the liver was not influenced by the changes in the perfusate protein concentration. In contrast, when injected via the HA, there was a small but not significant increase in the CV² values with an increase in the HSA concentration (i.e., from 0.60 ± 0.03 to 0.73 ± 0.08 for 0.25 to 1% HSA, respectively). Furthermore, irrespective of perfusate albumin concentration used, no significant difference was observed between HA and PV inputs with respect to CV² estimates.

Diazepam. Results of the moment analysis for labeled diazepam are shown in Table 2. Regardless of the perfusate protein concentration, MTT after arterial administration was longer than that after venous administration (P < .05 for 0.25% HSA and P < .01 for 1% HSA). On the other hand, the transit times decreased with an increase in the albumin concentration from 0.25 to 1% (P < .001; Table 2). When administered via the HA, spreading of labeled diazepam within the liver was not influenced by the changes in the perfusate albumin concentration. In contrast when administered via the PV, there was a small but significant increase in the CV² values with an increase in the HSA concentrations in the perfusate (P < .05). However, for a given perfusate protein concentration, no significant difference was obtained between HA and PV inputs with regard to CV² estimates.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Choice of Compound The liver is a heterogeneous organ in terms of its flow distribution: 83% of the arterial flow mixes with the venous flow in a common space, with the remainder perfusing a specific space (Sahin and Rowland, 1998a). Despite such heterogeneity, currently it is not known how these two spaces differ with respect to enzyme content. Unlike PV administration, hepatic extraction after HA input will be governed by events in both the common and specific spaces. Any difference in fractional hepatic recovery as a function of route of input could therefore be attributed to the specific space and its enzyme content because the extraction of compound from the common space will be the same whether administered via the HA or PV. Nevertheless, demonstration of any differences in fractional hepatic recovery requires the use of highly cleared compounds (e.g., ones with an extraction ratio in excess 0.90-0.95; Ahmad et al., 1984) because the specific space receives only 17% of the arterial flow and hence the arterial dose. At the extreme, if no enzyme is present in this space 17% of arterial dose will escape elimination. However, in the case of slowly extracted compounds, problems may arise in demonstrating a difference in hepatic extraction.

Albumin. Comparison of the V_H values indicates that labeled albumin has access to larger space after arterial than venous administration, confirming our previous observation made with labeled albumin and other reference markers, including erythrocytes, sucrose, urea, and water (Sahin and Rowland, 2000). This excess space was about 15 to 20% of the total extracellular space. Furthermore, longer MTT values and also slightly higher CV² values obtained after arterial administration could be attributed to the presence of more tortuous pathways (i.e., peribiliary capillary plexus) taken by the arterial blood.

Outflow Profiles Regardless of route of input, changes in albumin concentration had a clear effect on the hepatic outflow profiles of labeled diazepam. When 1% HSA was used, [¹⁴C]diazepam outflow profiles were characterized by a sharp early peak (throughput component) followed by a slowly eluting tail, produced by the material returning from the cellular space having escaped metabolism and biliary excretion (returning component). This is in accordance with previously observations with diazepam in the single PV perfused rat liver and is indicative of a compound whose radial distribution from the vascular space within this organ is not instantaneous (Diaz-Garcia et al., 1992). When the perfusate HSA concentration was decreased to 0.25%, the outflow profiles displayed monophasic characteristics. Such discrepancies in outflow profiles as a function of protein concentration can be attributed to the hepatic uptake processes involved.

Effect of Route of Administration on the Hepatic Recovery. A survey of available data on the hepatic extraction or clearance of various compounds in relation to the route of presentation to the liver is currently ambiguous. Hepatic clearances of both bromosulphthalein and CrPO₄ colloid (Brauer et al., 1959) and of ⁸⁵Kr (Hollenberg and Dougherty, 1966) in dog liver were not found to be influenced by route of hepatic input. When equal doses of bile salts were infused into the hepatic artery or portal vein, taurocholic acid produced equal effects on bile flow regardless of the infusion site (Lautt and Daniels, 1983). Hepatic extraction of indocyanine green (Lautt et al., 1984), and extraction ratio of vitamin D₃ (Gascon-Barre et al., 1988) were unchanged whether administered into the HA or PV. Ikeda et al. (1993) also observed no difference in the hepatic extraction of arterially or portally delivered insulin, glucagon, and epinephrine. Furthermore, no significant difference was reported between regional kinetics (i.e., hepatic extraction and hepatic clearance) and systemic exposure of 5-fluorouracil and 5-bromo-2'-deoxyuridine after hepatic arterial versus portal venous infusions of drugs (Kuan et al., 1996). In contrast, more epinephrine (Meyerholz et al., 1991) and norepinephrine (Gardemann et al., 1991; Zimmermann et al., 1992) were extracted when administered via the HA. Furthermore, hepatic recovery was 18 and 3 times greater for lidocaine and meperidine, respectively, when infused via the HA compared with the PV (Ahmad et al., 1984). Recently, small but significant decreases were also observed in the extraction ratios of [¹⁴C]phenacetin and tritiated acetaminophen delivered simultaneously into the HA and PV, with increment of HA flow, within the same liver preparation (Pang et al., 1994). Additionally, two chemotherapeutic agents, Adriamycin (Shiotani et al., 1995) and doxorubicin (Iwasaki et al., 1998), were more effectively extracted by the liver when administered via the HA than PV, leading to the conclusion that intra-arterial administration of these drugs are superior to intraportal administration in terms of reduction of systemic drug exposure and systemic toxicity.