Pharmacokinetics and Distribution over the Blood-Brain Barrier of Zalcitabine (2′,3′-Dideoxycytidine) and BEA005 (2′,3′-Dideoxy-3′-Hydroxymethylcytidine) in Rats, Studied by Microdialysis

Abstract

Microdialysis was applied to sample the unbound drug concentration in the extracellular fluid in brain and muscle of rats given zalcitabine (2′,3′-dideoxycytidine; n = 4) or BEA005 (2′,3′-dideoxy-3′-hydroxymethylcytidine; n = 4) (50 mg/kg of body weight given subcutaneously). Zalcitabine and BEA005 were analyzed by high-pressure liquid chromatography with UV detection. The maximum concentration of zalcitabine in the dialysate (C_max) was 31.4 ± 5.1 μM (mean ± standard error of the mean) for the brain and 238.3 ± 48.1 μM for muscle. The time to C_max was found to be from 30 to 45 min for the brain and from 15 to 30 min for muscle. Zalcitabine was eliminated from the brain and muscle with half-lives 1.28 ± 0.64 and 0.85 ± 0.13 h, respectively. The ratio of the area under the concentration-time curve (AUC) (from 0 to 180 min) for the brain and the AUC for muscle (AUC ratio) was 0.191 ± 0.037. The concentrations of BEA005 attained in the brain and muscle were lower than those of zalcitabine, with C_maxs of 5.7 ± 1.4 μM in the brain and 61.3 ± 12.0 μM in the muscle. The peak concentration in the brain was attained 50 to 70 min after injection, and that in muscle was achieved 30 to 50 min after injection. The half-lives of BEA005 in the brain and muscle were 5.51 ± 1.45 and 0.64 ± 0.06 h, respectively. The AUC ratio (from 0 to 180 min) between brain and muscle was 0.162 ± 0.026. The log octanol/water partition coefficients were found to be −1.19 ± 0.04 and −1.47 ± 0.01 for zalcitabine and BEA005, respectively. The degrees of plasma protein binding of zalcitabine (11% ± 4%) and BEA005 (18% ± 2%) were measured by microdialysis in vitro. The differences between zalcitabine and BEA005 with respect to the AUC ratio (P = 0.481), half-life in muscle (P = 0.279), and level of protein binding (P = 0.174) were not statistically significant. The differences were statistically significant in the case of the half-life in the brain (P = 0.032), clearance (P = 0.046), volume of distribution (P = 0.027) in muscle, and octanol/water partition coefficient (P = 0.019).

Various strategies are used in the development of antiviral drugs in the form of structural analogs of nucleosides. Most nucleoside analogs are first converted to their 5′-triphosphates, which then exert their biological effects as virus-specific transcriptase inhibitors or chain terminators. It has been shown that 2′,3′-dideoxynucleoside analogs in which the 3′-hydroxyl group in the ribose sugar moiety is replaced with a hydrogen (zalcitabine) or other chemical groups have a high level of antiviral activity, and a number of 2′,3′-dideoxy-3′-hydroxymethyl nucleosides have been synthesized as potential inhibitors of human immunodeficiency virus (HIV) (21, 22). Among these analogs, several were shown to have activity against HIV, but only 2′,3′-dideoxy-3′-hydroxymethylcytidine (BEA005) had a high level of inhibitory activity against HIV and a broad range of DNA viruses (21). In vivo, BEA005 has been found to be active against acute simian immunodeficiency virus and HIV type 2 infections in macaques in pre- and postexposure treatment experiments (2). Therefore, this compound may be a useful antiretroviral agent in humans. An important property of potential anti-HIV agents is their ability to penetrate into the brain since this is one of the potential sanctuaries of the virus. In order to investigate this we have used microdialysis (15, 16) which is a method for the continuous monitoring of unbound extracellular drug concentrations in soft tissues. Microdialysis has been used to study the distributions of a range of nucleoside analogs (1, 20).

Thus, the aim of the present study was to investigate the distribution to the brain and the pharmacokinetics of 2′,3′-dideoxy-3′-hydroxymethylcytidine (BEA005) in rats. Zalcitabine (2′,3′-dideoxycytidine) was included in the study as a reference compound. In addition, some physicochemical properties of the two compounds and degrees of binding in human plasma were investigated.

MATERIALS AND METHODS

Antiviral drugs.

Zalcitabine and BEA005 were generous gifts from Medivir AB, Huddinge, Sweden. Both compounds were dissolved in Ringer solution to a concentration of 50 mg/ml.

Animals.

Male Sprague-Dawley rats (weight, 180 to 260 g; B&K Universal AB, Sollentuna, Sweden) were used throughout the study. The rats had free access to tap water and standard lab chow and were housed at five to six rats per cage. Each rat was used only once.

Microdialysis.

Detailed accounts of the microdialysis method can be found elsewhere (12–15, 17–19, 25). In the present study the rats were anesthetized with halothane during the whole experiment and were placed in a David Kopf stereotaxic instrument with the bite bar 2.5 mm below the interaureal line. Dialysis probes (a membrane 0.50 mm in diameter and 4.0 mm long); CMA Microdialysis, Stockholm, Sweden) of the concentric type were implanted into the corpus striatum on one side of the brain (stereotaxic coordinates, 2.2 mm lateral and 1.3 mm anterior to the bregma and 7.5 mm ventral to the brain surface) and into the gastrocnemius muscle. Each dialysis probe was perfused with a Ringer solution at a rate of 0.50 μl/min, and samples of 15 μl were collected at 30-min intervals or at a rate of 1 μl/min and samples of 20 μl were collected at 20-min intervals. The variation in microdialysis parameters are motivated by the difficulty in obtaining a sufficiently sensitive analysis of the compounds.

The in vivo recovery of each dialysis probe was determined in the following way. After implantation and a 30-min washout period the perfusion medium was switched to a solution containing the compound at 10 μM, (zalcitabine or BEA005), which was subsequently given systemically. Two (at a rate of 0.5 μl/min) or three (at a rate of 1 μl/min) samples were collected, and the perfusion medium was switched back to the Ringer solution. After a second washout period of 40 to 60 min no drug was detectable in the dialysates. The proportion of drug lost over the dialysis membrane was taken as an estimate of the in vivo recovery over the dialysis membrane (17–19).

After a 60-min washout period, zalcitabine (50 mg/kg; n = 4) or BEA005 (50 mg/kg; n = 4) was injected subcutaneously, and samples were collected for 180 min postinjection. The concentration measured in the dialysate was converted to estimates of the unbound extracellular concentration as described above. The mean recoveries for zalcitabine were 0.170 (at a rate of 0.5 μl/min) and 0.045 (at a rate of 1 μl/min) in the brain and 0.275 (at a rate of 0.5 μl/min) and 0.155 (at a rate of 1 μl/min) in muscle. The mean recoveries for BEA005 (at a rate of 1 μl/min) were 0.129 in the brain and 0.362 in muscle.

Plasma protein binding.

The levels of plasma protein binding of zalcitabine and BEA005 were determined by microdialysis as described in detail elsewhere (9). Zalcitabine or BEA005 was added to human plasma to a total concentration of 10 μM. The microdialysis recoveries were determined by using a stirred 37°C Ringer solution containing a known concentration of the compound (C_m) into which the dialysis probe was inserted, and dialysis samples were collected and the concentration was measured (C_d). The recovery (R) is the ratio C_d/C_m. The recovery was about 0.950 in these experiments. Next, the dialysis probe was inserted into the plasma containing the compound at 10 μM and 37°C. The plasma was stirred in the same way as the Ringer solution, and the concentration in the dialysate was measured. The free concentration was then calculated as the concentration in the dialysate divided by R. The ratio of the free concentration to the total concentration is the unbound fraction (f_u), and 1 − f_u is the fraction bound to plasma protein.

Analysis of zalcitabine and BEA005.

Analysis of zalcitabine and BEA005 was performed by isocratic high-pressure liquid chromatographic separation and UV detection. Two different columns were used in the study: a 5-μm-particle-size C₁₈ column (Bioanalytical Systems) (100 by 2.1 mm) and a 3.5-μm-particle-size C₁₈ column (75 by 3.2 mm; ZORBAX). The mobile phase consisted of 0.05 M ammonium phosphate and 2 mM 1-octane-sulfonic acid at pH 3.0 containing 15% (vol/vol) methanol (16). Zalcitabine and BEA005 were easily separated from endogenous compounds both in the brain and in the muscle dialysates.

Octanol/water partition coefficient.

Equal volumes of octanol and a 100 μM aqueous solution of zalcitabine or BEA005 were mixed by shaking with a vortex mixer, and the phases were separated by centrifugation at 3,500 rpm for 2 min. The concentration of compounds in the aqueous phase was determined by high-pressure liquid chromatography before the mixing and after centrifugation. The concentration of each compound in the octanol phase was calculated as the difference between the concentrations in the aqueous phase before mixing the phases and after mixing and separation.

Experimental design and statistical analysis.

The experimental design was a simple time-concentration curve. Because the samples were collected for 20 or 30 min, the data are presented in the graphs as the midpoint of the collection period (i.e. 10, 30, 50,…170 min or 15, 45, 75,… 165 min), which is an acceptable approximation (12–15). Data for the individual rats are plotted in time-versus-log concentration diagrams. The area under the time-versus-concentration curve (AUC) from time zero to 180 min (AUC_0–180) was calculated by summing the products between the sampling interval (20 or 30 min) and the sample concentration (corrected for recovery) (14). The AUC ratio was calculated as the AUC_0–180 (brain)/AUC_0–180 (muscle). The AUC from time zero to infinity (AUC_0–∞) was calculated by extrapolation to infinity. The half-life was calculated as ln 2/slope, in which the slope of the elimination phase was determined by linear regression. The clearance (CL) was calculated as dose/AUC_0–∞ for muscle, and the volume of distribution (V) was calculated as CL/slope. Data are presented as means ± standard errors of means. Comparisons of the AUC ratio, half-life, CL, and V for zalcitabine and BEA005 were performed by Student’s t test. For hypothesis testing, a significance level of 5% was chosen.

RESULTS

The degrees of binding of zalcitabine and BEA005 to human plasma proteins were found to be 11% ± 4% and 18% ± 2%, respectively. The difference was not statistically significant (P = 0.174).

The octanol/water partition coefficients for zalcitabine and BEA005 were found to be −1.19 ± 0.04 and −1.47 ± 0.01, respectively. The difference was statistically significant (P = 0.019).

The time-concentration curves for zalcitabine for each rat are given in Fig. 1. The mean maximum concentration of zalcitabine in the dialysate (C_max) was 31.4 ± 5.1 μM for the brain and 238.3 ± 48.1 μM for muscle. The time to C_max in the dialysate was found to be from 30 to 45 min for the brain and from 15 to 30 min for muscle. The AUC_0–∞ was 295.2 ± 37.5 μmol · h/liter for muscle. Zalcitabine was eliminated from the brain and muscle with half-lives of 1.68 ± 0.56 and 0.76 ± 0.08 h, respectively. The AUC ratio between brain and muscle was 0.191 ± 0.032. CL calculated for the concentration in muscle was 0.86 ± 0.13 liters/h/kg. V was 0.95 ± 0.19 liter/kg for muscle.

FIG. 1.

Time-log (concentration) curves for muscle (——) and brain () from rats treated with zalcitabine (50 mg/kg subcutaneously; n = 4). The concentration is the free (unbound) extracellular concentration.

The time-concentration curves for BEA005 for each rat are given in Fig. 2. The maximum concentration of BEA005 in muscle (C_max = 61.3 ± 12.0 μM) was attained 30 to 50 min after subcutaneous injection, and BEA005 was then eliminated with a half-life of 0.64 ± 0.05 h. The maximum concentration of BEA005 in the brain (C_max = 5.74 ± 1.37 μM) was attained 30 to 70 min after injection. The AUC_0–∞ was 97.3 ± 14.4 μmol · h/liter for muscle. The half-life in the brain was 5.51 ± 1.26 h. The AUC ratio between brain and muscle was 0.162 ± 0.023. CL calculated for the concentration in muscle was 2.30 ± 0.56 liter/h/kg. V was 2.03 ± 0.32 liters/kg for muscle.

FIG. 2.

Time-log concentration curves for muscle (——) and brain (·····) from rats treated with BEA005 (50 mg/kg subcutaneously; n = 4). The concentration is the free (unbound) extracellular concentration.

The values of the pharmacokinetic parameters obtained for the two compounds are compared statistically in Table 1.

TABLE 1.

Values of pharmacokinetic parameters for zalcitabine and BEA005 in rats^a

Drug	t1/2 (CNS) (h)	t1/2 (muscle) (h)	CL (liter/h/kg)	V (liter/kg)	AUC ratio (CNS/muscle)
Zalcitabine	1.68 ± 0.56	0.76 ± 0.08	0.86 ± 0.13	0.95 ± 0.19	0.191 ± 0.032
BEA005	5.51 ± 1.26	0.64 ± 0.05	2.30 ± 0.56	2.03 ± 0.32	0.162 ± 0.023
P	0.032b	0.279	0.046b	0.027b	0.481

^aBoth drugs were given at 50 mg/kg to four rats each. Abbreviations: t_1/2, half-life; CNS, central nervous system. The other abbreviations were defined in the text. 

^bThe difference is statistically significant. 

DISCUSSION

Before discussing the specific findings of this study it is important that we clarify a technical issue. The concentrations in muscle referred to here should be understood to represent unbound concentrations in plasma, as indicated previously (1).

The pharmacokinetic properties of zalcitabine and BEA005 have not previously been investigated by microdialysis. This study shows that both 2′,3′-dideoxynucleosides penetrate the blood-brain barrier. The pharmacokinetic properties of zalcitabine obtained in this study are partly consistent with previously published results. It was previously reported that the CL of zalcitabine calculated from the concentrations in the plasma of dogs is about 5.0 to 5.6 ml/min/kg (6), which is lower than those that we obtained. However, the CL obtained after administration of a high dose (500 mg/kg) to rats was 0.75 liter/h/kg (10), which is very similar to the CL obtained in this study, suggesting a difference in CL between species. A second pharmacokinetic parameter whose values were not found to be consistent with those published previously was half-life. Previous studies showed that the rate of elimination of zalcitabine from plasma is similar among species; the half-lives were 1.29 h for rats, 1.5 h for mice, 1.7 h for monkeys, 1.8 h for microswine, and 1.2 h for humans (6, 10, 11, 24, 26). In the present study the half-life of zalcitabine was shorter, about 0.76 h in muscle.

It was reported previously that zalcitabine is not measurably transported into the brain in the rat (23). In that study the intracarotid injection technique was used to measure the transport of zalcitabine, azidothymidine, and dideoxyadenosine, but the method did not allow any of these nucleoside analogs to be found after decapitation and removal of the brain 15 s after carotid injection. However, it is well known that azidothymidine does penetrate the blood-brain barrier. Other methods of analysis of the distribution of zalcitabine in the brain have demonstrated its presence in the cerebrospinal fluid of monkeys, dogs, and humans (4, 5, 24). The level of penetration of zalcitabine into the central nervous system was reported to be low, and the drug levels in the cerebrospinal fluid were 3% (dog, monkey) and 20% (human) of the concentration in plasma after administration by intravenous injection. In this study the AUC ratio for zalcitabine between brain and muscle was found to be about 19%, which is somewhat unexpected in view of the fact that the blood-barrier is generally less permeable in rats than in primates (3). However, the concentrations in cerebrospinal fluid may not be good indicators of the ability of a drug to cross the blood-brain barrier in deep brain tissue.

The octanol/water partition coefficient for zalcitabine obtained in this study is consistent with previously published data (5). The level of plasma protein binding measured by microdialysis in the present study is higher than that obtained previously by ultrafiltration (5). This difference in results can possibly be explained by the difference in the sensitivities of the methods. For practical purposes the difference between the ultrafiltration data and the present data is not clinically significant.

The pharmacokinetic properties of BEA005 have not been investigated previously. The maximum concentrations of BEA005 both in the brain and in muscle were lower than those of zalcitabine after subcutaneous administration of the same dose, but the times to C_max were similar for both compounds. We also found a difference in the elimination phase of these compounds. The difference between BEA005 and zalcitabine was significant for half-lives in the brain but not those in muscle. The pharmacokinetic parameters CL and V also differed between these cytidine analogs. The abilities of the two cytidine analogs to penetrate the blood-brain barrier were not shown to be significantly different when the AUC ratios between brain and muscle were taken as a measure of this. The absence of a difference between the abilities of the cytidine analogs to pass into the brain may be due to similar affinities for a common transport system. It has previously been found that the functional group at position 3′ on the sugar does not influence the penetration into the cerebrospinal fluid of monkeys (5), despite differences in the lipophilicities of the drugs investigated. At partition coefficients above 10 lipophilicity is known to favor entry into the central nervous system, but this is probably not the case for the cytidine analogs, which have much lower partition coefficients. It seems reasonable to suggest that the transport of cytidine analogs into the brain is carrier mediated and is done in a manner similar to that for thymidine derivatives. Our previous observation for thymidine analogs, that the ability to penetrate to the central nervous system is more closely associated with the nucleoside base structure than with the structure of the sugar moiety, is in agreement with the results of this study (1).