Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Because the BBB and BCSFB are responsible for ligand transfer among the circulating blood, brain parenchymal tissue and CSF, it has been assumed that the BBB and BCSFB represent a predominant pathway for ligand distribution in the brain and CSF, respectively. However, if we consider the fact that 1) there is free ligand exchange between brain ISF and CSF across the ependymal surface and 2) several specific transport systems have been reported to play important roles at the BBB (Pardridge and Oldendorf, 1977; Pardridge, 1983; Smith et al., 1987; Terasaki and Tsuji, 1994) and BCSFB (Spector, 1982, 1986) in terms of ligand transport, it is necessary to examine the contribution of these two pathways to the distribution of drugs in brain tissue and CSF by using the pharmacokinetic model. Considering drug diffusion in brain parenchymal tissue, Collins and Dedrick (1983) and Collins (1983) have demonstrated that a distributed model is useful to analyze drug distribution in brain tissue and CSF. Using this model, we have recently shown that the organic anion efflux transport systems across the BBB and BCSFB play a dominant role in the elimination of -lactam antibiotics from the CSF (Ogawa et al., 1994; Suzuki et al., 1997).

For the effective treatment of AIDS-ADC or AIDS encephalopathy, anti-HIV drugs such as AZT and DDI need to reach the CNS in significant amounts. However, several investigators have demonstrated that the CNS distribution of AZT (Terasaki and Pardridge, 1988; Galinsky et al., 1990; Masereeuw et al., 1994; Wang and Sawchuk, 1995) and DDI (Anderson et al., 1990; Hoesterey et al., 1991) is restricted. Although the probenecid-sensitive efflux transport system across the BCSFB has been reported to be responsible for the restricted distribution of DDI both in CSF and brain parenchymal tissue (Galinsky et al., 1991), there have also been reported that BBB efflux transport is responsible for the restricted cerebral distribution of AZT (Wang and Sawchuk, 1995; Dykstra et al., 1993). Moreover, we have shown that AZT and DDI are transported from brain tissue into the circulating blood across the BBB via a probenecid-sensitive carrier-mediated system (Takasawa et al., 1997).

Accordingly, the purpose of the present study is to provide kinetic evidence of the contribution of these two barriers, by means of a distributed model, on the restricted distribution of AZT and DDI in brain tissue and CSF.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. [Methyl]-³H-3-azido-3-deoxythymidine ([³H]AZT; 20 Ci/mmol) and [2,3-³H(N)]-2,3-dideoxyinosine ([³H]DDI; 42 Ci/mmol) were purchased from Moravek Biochemicals Inc. (Brea, CA). Carboxyl-¹⁴C-inulin (2.6 mCi/g) and D-[1-¹⁴C]-mannitol (56.7 mCi/mmol) were purchased from New England Nuclear (Boston, MA). 4-(2-Hydroxyethyl)-1-piperazineethanesulfonic acid from Dojin Chemicals (Kumamoto, Japan) was of analytical grade. All other reagents were commercial products of reagent grade and were used without further purification. Male Wistar rats weighing 230 to 270 g were used throughout the experiments which were carried out according to the guidelines provided by The University of Tokyo Animal Care Committee.

Influx of [³H]AZT and [³H]DDI into brain tissue and CSF. Rats were anesthetized with urethane (ethylcarbamate) (1.5 g/kg) and their femoral artery and vein were cannulated with polyethylene tubing, PE50 (Becton Dickinson, Parsippany, NJ). [³H]AZT (44 µCi/kg, 2.2 nmol/kg) or [³H]DDI (88 µCi/kg, 2.1 nmol/kg) dissolved in 250 µl of saline was administered i.v. via the femoral vein through the cannula. An aliquot of blood (0.5 ml) was withdrawn from the femoral artery through the cannula 1, 3, 5, 8 and 10 min after administration. Then 2, 5 or 10 min after i.v. administration an aliquot of CSF specimen (50-150 µl) was taken by cisternal puncture by using a syringe (27.5G; Terumo, Tokyo, Japan) (Suzuki et al., 1988, 1989). Immediately after the collection of CSF, rats were decapitated and the cerebrums were removed. The ligand concentrations in the brain, CSF and plasma specimens were determined according to the method described in the following section.

(1)

(2)

(0t)

(3)

(4)

(0t)

View larger version (13K): [in this window] [in a new window]   Fig. 2.   Integration plots of AZT and DDI representing apparent uptake at the BBB (A) and BCSFB (B) after i.v. bolus administration. The solid lines were obtained by the nonlinear least squares program. The symbols represent the observed values of the mean. All values of S.E. are within the symbols. A, Brain (:AZT, :DDI), B, CSF (:AZT, :DDI).

Efflux of [³H]AZT or [³H]DDI from CSF. The efflux of [³H]AZT or [³H]DDI from CSF was examined after i.c.v. bolus administration as described in a previous report (Suzuki et al., 1985, 1989). The rats were fixed in a stereotaxic apparatus and a needle (350 µm o.d.; Seiseido Medical Industry, Tokyo, Japan) connected to silastic tubing was inserted into the left lateral ventricle through a hole drilled in the skull. [³H]AZT (0.8 µCi) or [³H]DDI (0.8 µCi) and [¹⁴C]mannitol (0.018 µCi) dissolved in 10 µl physiological buffer were administered through the cannula. The physiological buffer contained 122 mM NaCl, 25 mM NaHCO₃, 10 mM glucose, 3 mM KCl, 1.4 mM CaCl₂, 1.2 mM MgSO₄, 0.4 mM K₂HPO₄ and 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, pH 7.3 and was equilibrated with 95%O₂-5%CO₂ gas at 37°C. After i.c.v. administration, aliquots of CSF (50-150 µl) were withdrawn as described above at appropriate times.

Uptake of [³H]AZT or [³H]DDI by brain slice. The distribution volume in the brain of [³H]AZT or [³H]DDI was determined in an in vitro brain slice uptake study, as reported previously (Kakee et al., 1996) with minor modification (Newman et al., 1991). After decapitating rats, the cerebrums were immediately removed and dissected in ice-cold oxygenated physiological buffer. Using a microslicer, model DTK-2000, (Dosaka, EM, Kyoto, Japan), five pieces of 300 µm in thickness were obtained at 2.0 mm anterior to the bregma from one cerebrum, kept in ice-cold oxygenated physiological buffer equilibrated with 95%O₂-5%CO₂. After preincubation for 5 min at 37°C, the uptake study was initiated by transferring the slice to 50 ml of the oxygenated physiological buffer containing [³H]AZT, [³H]DDI or [¹⁴C]inulin, at 4 × 10³ µCi/ml, 4 × 10³ µCi/ml or 1 × 10² µCi/ml, respectively. At appropriate times, brain slices and an aliquot of 500 µl of the uptake medium were collected. In the [¹⁴C]inulin uptake study, brain slices were dissolved in 2.5 ml 2 N NaOH at 50°C for 3 hr. The radioactivity of dissolved tissue and uptake medium were determined in liquid scintillation cocktail by a liquid scintillation counter (LC6000, Beckmann, Instruments, Inc., Fullerton, CA). In the [³H]AZT and [³H]DDI uptake study brain slices and uptake medium were stored at 20°C and the determination of unmetabolized drug was carried out by HPLC.

Plasma protein binding of [³H]AZT and [³H]DDI. The unbound fraction of [³H]AZT and [³H]DDI in plasma was determined by ultrafiltration method as described previously (Shimamura et al., 1994). Plasma samples containing 0.1 µM of AZT and DDI were incubated at 37°C for 30 min and then centrifuged (2,000 × g) for 10 min through a suitable membrane (MPS-3; Amicon, Division, W. R. Grace & Co., Danvers, MA). The filtrate was collected and an aliquot of 10 µl analyzed by HPLC and counted in a liquid scintillation counter.

Analysis of [³H]AZT and [³H]DDI concentrations in the specimen. Unmetabolized [³H]AZT and [³H]DDI in brain tissue, brain slice, plasma, CSF and uptake medium were determined by HPLC. For the analysis, 0.6 ml physiological buffer was added to 0.6 g brain and then homogenized in an ULTRA-TURRAX T25 (Janke & Kunkel, IKA-Labortechnik, Germany). Then 2.1 ml acetonitlile were added to the homogenate, mixed well and kept at 4°C for 1 hr. After centrifugation for 15 min at 3000 rpm, the supernatant was evaporated to dryness under a stream of N₂ and then reconstituted with 180 µl mobile phase. After filtration through a membrane filter (0.45 µm, pore size; Millipore, Bedford, MA), 100 µl of aliquot were loaded onto a reversed-phase HPLC column, TSK gel ODS-80 (15.0 cm × 4.6 mm i.d., Toso, Tokyo, Japan). A constant flow solvent delivery system, model 655A-11 (Hitachi, Tokyo, Japan) was used. It was equipped with an injection system, Rheodyne model 7125, with a 200-µl injection loop and a UV detector (model 638-41, Hitachi). A guard column (1.5 cm × 3.2 mm i.d., Toso, Tokyo, Japan) was placed between the injector and the analytical column. The mobile phase was methanol-0.01 M ammonium acetate [23:77 (v/v), AZT; 11:89 (v/v), DDI], pH 6.5 at a flow rate of 0.8 ml/min. The column and solvent were kept at ambient temperature. The absorbance of the eluent was monitored continuously at 254 nm. Eluent was collected using a fraction collector, Model 2110 (Bio-Rad, Richmond, CA) and 10 ml liquid scintillation cocktail were added to each fraction. The radioactivity of the resulting mixture was measured in a liquid scintillation counter.

Theoreticals. Figure 1 illustrates a distributed model that has been proposed previously (Collins and Dedrick, 1983; Collins, 1983). Analysis was performed according to the method described previously (Suzuki et al., 1997) with some modifications. To derive the mass balance equation, we have assumed that 1) brain tissue and CSF are described by a one-dimensional slab of tissue; 2) a drug molecule diffuses through brain tissue in accordance with Fick's law of diffusion; 3) a drug is transported by an influx and efflux system across the BBB and BCSFB; 4) there is a bulk flow of CSF at a constant rate; 5) a drug distributes into the intracellular fluid space in the brain.

View larger version (28K): [in this window] [in a new window]   Fig. 1.   Distributed model for pharmacokinetic analysis of AZT and DDI distribution in brain and CSF.

Model analysis after i.v. administration. Equation 4 represents a mass balance equation describing drug concentration in brain tissue and CSF after i.v. administration.

(4)

where C_br(x, t) is the drug concentration in brain tissue at distance x from the surface of the ependymal cell layer at time t, D_t is the apparent diffusion coefficient in brain tissue, PS_BBB and PS_BBB,eff are the symmetrical permeability clearance across the BBB and the efflux transport clearance from the brain into the blood across the BBB, respectively, Cp,u(t) is the unbound drug concentration in plasma, and V_br is the distribution volume defined as the ratio of the drug concentrations in brain tissue and ISF space. As an initial condition of equation 4, C_br(x, 0) is zero at time t = 0 was assumed.

(5)

(6)

(7)

(8)

(9)

(10)

(11)

Model analysis after i.c.v. administration. Equation 12 represents a mass balance equation describing drug concentration in brain tissue and CSF after i.c.v. administration.

(12)

Taking x and t of equation 12 as zero, as an initial condition, the following relation is obtained for the CSF concentration. Assuming equation 5, the following relation is obtained.

(13)

As a boundary condition where the distance x is zero, the followingrelation is derived.

(14)

As a boundary condition where the distance x is significantly greater than x*, the following relation is obtained.

(15)

Taking the Laplace transform of equations 12 and 14, the following equation can be derived.

(16)

AUC_CSF after i.c.v. administration is obtained as,

(17)

The efflux clearance from the CSF after i.c.v. administration can be described by the following equation.

(18)

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Disposition of [³H]AZT and [³H]DDI in the CNS. The apparent influx clearance across the BBB was determined from the slope of the brain-to-plasma concentration ratio (K_p,app.) vs. the ratio of the area under the plasma concentration-time curve from time 0 to t (AUC_(0t)) to the plasma concentration after i.v. administration. As shown in figure 2A, CL_uptake values for [³H]AZT and [³H]DDI were found to be 2.88 ± 0.43 and 0.716 ± 0.172 µl/min/g brain (mean ± S.D.), respectively. The apparent intercept at zero time was found to be 25.3 ± 3.1 µl/g brain for [³H]AZT and 16.0 ± 2.2 µl/g brain (mean ± S.D.) for [³H]DDI.

(0t)

Uptake of [³H]AZT and [³H]DDI by brain slice. Figure 3 illustrates the time course of the apparent slice-to-medium concentration ratio (S/M ratio) for [³H]AZT, [³H]DDI and [¹⁴C]inulin. The adherent water volume was determined as the zero time intercept of [¹⁴C]inulin, i.e., 0.104 ml/g brain. No significant difference was observed in the S/M ratio of [³H]DDI uptake at 30 min and 60 min. Subtracting the adherent water volume from the apparent S/M ratio 30 min after incubation, the in vitro distribution volume of [³H]AZT and [³H]DDI were found to be 1.07 ± 0.09 ml/g brain and 0.727 ± 0.030 ml/g brain (mean ± S.E., n = 4, 5), respectively.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Time courses of apparent uptake of [14C]inulin, [3H]AZT and [3H]DDI by brain slices. Each point represents the mean ± S.E. of four to eight experiments.

Pharmacokinetic analysis of [³H]AZT and [³H]DDI distribution in brain tissue and CSF based on the distributed model. The concentration-time profiles of [³H]AZT and [³H]DDI in brain tissue and CSF after i.v. bolus administration and in CSF after i.c.v. bolus administration are shown in figures 4 and 5, respectively. Pharmacokinetic parameters describing plasma unbound concentration-time profiles were found to be A = 1.59% of dose/ml, B = 0.768% of dose/ml,  = 4.14 min¹ and  = 0.0629 min¹ for [³H]AZT and A = 1.29% of dose/ml, B = 0.512% of dose/ml,  = 0.964 min¹ and  = 0.0636 min¹ for [³H]DDI, respectively. The influx and efflux clearances of [³H]AZT and [³H]DDI across the BBB and BCSFB were obtained by fitting the data to equations 10, 11 and 16 using the nonlinear least squares method combined with a fast inverse Laplace transform (MULTI (FILT)) (Yano et al., 1989). The initial and model-fitted parameters are listed in table 1. Figures 4 and 5 show the good agreement between the observed and model fitted concentration-time profiles of [³H]AZT and [³H]DDI, respectively.

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Time courses of brain and CSF concentrations of AZT after i.v. or i.c.v. administration to rats. A, Brain concentration after i.v. administration. B, CSF concentration after i.v. administration. C, CSF concentration after i.c.v. administration. The symbols and solid line represent the observed values and the generated values using the best fitted parameters with a distributed model as listed in table 1.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Time courses of brain and CSF concentrations of DDI after i.v. or i.c.v. administration to rats. A, Brain concentration after i.v. administration. B, CSF concentration after i.v. administration. C, CSF concentration after i.c.v. administration. The symbols and solid line represent the observed values and the generated values using the best fitted parameters with a distributed model as listed in table 1.

Simulation study. The effects of efflux transport across the BBB and BCSFB on the distribution of AZT and DDI in brain tissue and CSF were examined under several conditions. As shown in figure 6A, a significant increase in brain concentration was observed when no efflux transport across the BBB was assumed (PS_BBB,eff = 0). However, no significant effect of efflux transport across the BBB was observed on the CSF concentration-time profiles shown in figures 6B and C. Although a significant decrease in the brain tissue concentration-time profile was observed when no permeation across the BBB was assumed (PS_BBB = PS_BBB,eff = 0; fig. 6A), no significant alteration were observed in the CSF concentration-time profiles (figs. 6B and C). As shown in figures 6E and F, significant increase in the CSF concentration-time profile was observed when no efflux transport across the BCSFB was assumed (PS_CSF,eff = 0). However, no significant effect of efflux transport across the BCSFB was observed on the brain tissue concentration-time profile (fig. 6D). When no permeation across the BCSFB was assumed (PS_CSF = PS_CSF,eff = 0), significant decrease (fig. 6E) and increase (fig. 6F) in the CSF concentration-time profile after i.v. and i.c.v. administration, respectively, was observed. No significant effect was observed for the brain tissue concentration-time profile when no permeation across the BCSFB was assumed, (PS_CSF = PS_CSF,eff = 0; fig. 6D).

View larger version (30K): [in this window] [in a new window]   Fig. 6.   Effect of the BBB and BCSFB permeability on the time courses of brain and CSF concentrations of AZT after i.v. or i.c.v. administration to rats. A and D, Brain concentration after i.v. administration. B and E, CSF concentration after i.v. administration. C and F, CSF concentration after i.c.v. administration. The symbols and solid line represent the observed values and the generated values using the best fitted parameters with a distributed model as listed in table 1. The dotted lines represent the simulated values. The parameters used for the simulation are listed in table 1, unless indicated as follows. A-C, PSBBB,eff = 0 (-·-·-), PSBBB = PSBBB,eff = 0 (- - -); D-F, PSCSF,eff = 0 (-·-·-), PSCSF = PSCSF,eff = 0 (- - -).

View larger version (31K): [in this window] [in a new window]   Fig. 7.   Effect of the BBB and BCSFB permeability on the time courses of brain and CSF concentrations of DDI after i.v. or i.c.v. administration to rats. A and D, Brain concentration after i.v. administration. B and E, CSF concentration after i.v. administration. C and F, CSF concentration after i.c.v. administration. The symbols and solid line represent the observed values and the generated values using the best fitted parameters with a distributed model as listed in table 1. The dotted lines represent the simulated values. The parameters used for the simulation are listed in table 1, unless indicated as follows. A-C, PSBBB,eff = 0 (-·-·-), PSBBB = PSBBB,eff = 0 (- - -); D-F, PSCSF,eff = 0 (-·-·-), PSCSF = PSCSF, eff = 0 (- - -).

View larger version (29K): [in this window] [in a new window]   Fig. 8.   Effect of diffusion coefficient on the time courses of brain and CSF concentrations of AZT after i.v. or i.c.v. administration to rats. A, Brain concentration after i.v. administration. B, CSF concentration after i.v. administration. C, CSF concentration after i.c.v. administration. The symbols and solid line represent the observed values and the generated values using the best fitted parameters with a distributed model as listed in table 1. The lines represent the simulated values. The parameters used for the simulation are listed in table 1, unless indicated as follows. Dt = Dw (-·-·-), Dt = Dt (), Dt = Dt/100 (- - -).

View larger version (28K): [in this window] [in a new window]   Fig. 9.   Effect of diffusion coefficient on the time courses of brain and CSF concentrations of DDI after i.v. or i.c.v. administration to rats. A, Brain concentration after i.v. administration. B, CSF concentration after i.v. administration. C, CSF concentration after i.c.v. administration. The symbols and solid line represent the observed values and the generated values using the best fitted parameters with a distributed model as listed in table 1. The lines represent the simulated values. The parameters used for the simulation are listed in table 1, unless indicated as follows. Dt = Dw (-·-·-), Dt = Dt (), Dt = Dt/100 (- - -).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The endothelial cells of brain capillaries and epithelial cells of choroid plexus are known to regulate the transfer of many compounds among brain ISF, CSF and circulating blood (Pardridge and Oldendorf, 1977; Pardridge, 1983; Smith et al., 1987; Terasaki and Tsuji, 1994; Spector, 1982, 1986). Although it is assumed that the transport across the BBB rather than BCSFB plays a predominant role in determining the brain-to-plasma concentration ratio of ligands after i.v. administration since the surface area of the BBB is 5000-fold greater than that of the BCSFB (Pardridge, 1983), an analysis with the distributed model in which the anatomical features of the CNS is considered (Collins and Dedrick, 1983; Collins, 1983; Suzuki et al., 1997) will enable us to gain a deeper insight into drug distribution mechanism in CNS. Several investigators have examined the distribution of AZT and DDI in CNS by means of a compartmental model (Tuntland et al., 1991), brain microdialysis (Hoesterey et al., 1991; Kakee et al., 1996; Wong et al., 1993) and an in vitro transport study (Masereeuw et al., 1994; Zimmerman et al., 1987). The present investigation, employing a distributed model, has clarified the kinetic features of the restricted distribution of AZT and DDI in brain tissue and CSF as follows; 1) efflux transport across the BBB plays a dominant role in the apparently restricted distribution of AZT and DDI in brain parenchymal tissue, but not in CSF; 2) efflux transport across the BCSFB plays a dominant role in the restricted distribution of AZT and DDI in CSF, but not in brain parenchymal tissue.

One of the advantages of employing a distributed model is that the effect of drug diffusion through brain tissue can be investigated quantitatively in terms of distribution in CNS (Collins and Dedrick, 1983; Collins, 1983; Suzuki et al., 1997; Dykstra et al., 1993). According to a previous report (Fenstermacher and Kaye, 1988), the apparent diffusion constants in brain tissue were calculated as 5.38 × 10⁵ (cm²/min) for AZT and 6.63 × 10⁵ (cm²/min) for DDI, by considering the reduced diffusion in the brain parenchymal tissue compared with that in water (table 1).

Only a slight effect was shown on the brain tissue and CSF concentration-time profiles by using a 100-fold smaller value of D_t and the diffusion coefficient in water, D_w (figs. 8 and 9). These results suggest the minimal significance of diffusion of drug molecules between brain ISF and CSF on the brain and CSF disposition. As HIV, in most cases, has been found in brain capillary endothelial cells and parenchymal cells surrounding the capillary endothelial cells and macrophages (Wiley et al., 1986; Koenig et al., 1986), drug transfer across the BBB is essential to increase AZT and DDI concentrations in brain tissue. Moreover, these results suggest that the CSF concentration cannot be used as an index of the brain tissue concentration of AZT and DDI, if HIV infection does not cause any significant increase in drug permeability via the BBB and/or the BCSFB.

The distributed model gave equation 21, which can be used to describe the brain-to-plasma unbound concentration ratio at steady state, K_p,u,br,ss (see Appendix). Using the pharmacokinetic and physiological parameters listed in tables 1 and 2, the values of K_p,u,br,ss were calculated to be 0.063 for AZT and 0.018 for DDI, comparable with those in previous reports (Galinsky et al., 1990; Anderson et al., 1990). In the same manner, the values of K_p,u,CSF,ss, calculated with equation 22 in Appendix, were 0.102 for AZT and 0.027 for DDI, which were very similar to those in previous reports (Galinsky et al., 1990; Anderson et al., 1990).

By analyzing the ligand amount remaining in the ipsilateral cerebrum after microinjection into the cerebral cortex (BEI method; Kakee et al., 1996), we have previously determined the apparent efflux transport rate of AZT and DDI from brain to circulating blood across the BBB (Takasawa et al., 1997). As shown in table 1, the BBB efflux clearance obtained by distributed model analysis was approximately 10-fold greater than that obtained by the BEI study (Takasawa et al., 1997). One possible explanation to explain this contradiction is that the rate-limiting process for the efflux of AZT and DDI from brain tissue to circulating blood is ascribed to the abluminal membrane transport.

Regarding the characteristics of the efflux transport system across the BBB, we have reported that AZT remaining in the ipsilateral cerebrum was significantly increased from 53.8 ± 1.8% to 69.3 ± 6.0%, 67.6 ± 3.3%, and 64.2 ± 3.5% of the administered dose in the presence of probenecid (50 nmol/0.5 µl injectate), PAH (500 nmol/0.5 µl injectate) and DIDS (5 nmol/0.5 µl injectate), respectively, 20 min after intracerebral microinjection of [³H]AZT in rats (Takasawa et al., 1997). The inhibitory effect of probenecid on the efflux of AZT is consistent with a previous observation obtained by using brain microdialysis (Dykstra et al., 1993). Moreover, DDI remaining in the ipsilateral cerebrum was significantly increased from 72.4 ± 3.9% to 92.6 ± 0.4% in the presence of probenecid (50 nmol/0.5 µl injectate) 20 min after intracerebral microinjection of [³H]DDI in rats (Takasawa et al., 1997). These results suggest that a carrier-mediated system shared by probenecid is responsible for the efflux transport of both AZT and DDI across the BBB (Takasawa et al., 1997). Regarding BCSFB transport, we have previously reported that the intact DDI remaining in the CSF was significantly increased from 1.66 ± 0.64 to 28.3 ± 3.3% of the administered dose when probenecid (0.35 µmol) was coadministered with [³H]DDI into the lateral ventricle (Takasawa et al., 1997). These results suggest that an organic anion transport system at the choroid plexus is responsible for elimination of DDI from CSF (Takasawa et al., in press). Such an organic anion transport system located at the basolateral membrane of the proximal tubule has been demonstrated as being responsible for the elimination of AZT from the kidney (Chatton et al., 1990; Griffiths et al., 1991).

Together with the previous findings, the results of our study suggest that an efflux transport process across the BBB plays an important role in determining the brain concentration of AZT and DDI. This suggestion should be taken into consideration in designing chemotherapeutic strategies and in developing new derivatives of AZT and DDI for the treatment of ADC.