Pharmacokinetics of the Antiviral Agent β-l-3′-Fluoro-2′,3′-Didehydro-2′,3′-Dideoxycytidine in Rhesus Monkeys

Abstract

β-l-3′-Fluoro-2′,3′-didehydro-2′,3′-dideoxycytidine (l-3′-Fd4C) is a potent and selective antiretroviral nucleoside with activity against lamivudine-resistant human immunodeficiency virus type 1 (HIV-1) and hepatitis B virus (HBV) in vitro. The pharmacokinetics of l-3′-Fd4C were characterized in three rhesus monkeys given single intravenous and oral doses. A two-compartment open model was fitted to the plasma and urine data. Plasma concentrations declined in a biexponential fashion with an average beta half-life of 12.45 h and central and steady-state volumes of distribution of 0.43 and 1.90 liters/kg, respectively. The average systemic and renal clearance values were 0.23 and 0.08 liters/kg, respectively, and the apparent mean terminal half-life of the oral dose was 12.5 h. The serum concentrations exceeded the 90% effective concentration value for lamivudine-resistant and wild-type HIV-1 after oral administrations. A large variation was observed in the oral bioavailability, which ranged from 15 to 31%. To determine whether the bioavailability may be improved by using a basic buffer solution, the oral dose was repeated to the same animals in a sodium bicarbonate solution. The bioavailability of l-3′-Fd4C administered with sodium bicarbonate was not significantly different from the bioavailability when the oral dose was administered in the absence of buffer (P = 0.49), suggesting that further development of this compound may warrant other approaches, such as development of a prodrug to improve its oral absorption.

The need to overcome drug-resistant mutations in viruses (7, 8, 14, 20, 22) and adverse effects during long-term human immunodeficiency (HIV) and hepatitis B virus (HBV) therapies have led to the development of several novel antiviral nucleoside agents (1, 2, 4, 6, 9, 11, 17, 21, 23, 24, 27, 28). Pharmacokinetic studies with relevant animal models are critical to determine important pharmacokinetic parameters that are needed to develop rational starting dose regimens for clinical trials in humans.

β-l-3′-Fluoro-2′,3′-didehydro-2′,3′-dideoxycytidine (l-3′-Fd4C) isa potent inhibitor of viral replication in the chronically HBV-infected HepAD79 cell line (90% effective concentration [EC₉₀] of <3.1 μM) containing the rtM204V mutation and against wild-type HIV type 1 (HIV-1)-infected human peripheral blood mononuclear cells (EC₉₀ of <0.22 μM) (5, 6, 13, 30). The d-analogues in the series of nucleosides related to L-3′-Fd4C were mostly inactive against HIV-1 (EC₉₀ of >10 μM). However, it has been previously demonstrated that nucleosides containing modifications at the 2′ and 3′ positions of the sugar moiety can have potent activity against resistant mutant viruses (4, 5, 13, 15, 30). l-3′-Fd4C was essentially nontoxic to various cells and had no mitochondrial toxicity in HepG2 cells at concentrations up to 100 μM; it was nontoxic to mice when administered intraperitoneally in a dosage of 100 mg/kg of body weight for 6 days.

The objective of this study was to determine the single dose pharmacokinetics of l-3′-Fd4C in rhesus monkeys following intravenous ([i.v.] in sterile saline) and oral administration. The nucleoside was administered orally both with and without bicarbonate buffer for comparison, since d-D4FC (Reverset), developed earlier in our laboratory, is a 5-fluorine-containing cytidine analogue that is unstable at gastric pH. The bioavailability of d-D4FC in monkeys was improved when administered with sodium bicarbonate buffer (16). However, l-3′-Fd4C is known to be stable for at least 1 h at neutral, low-acidic and high-alkaline conditions at room temperature (5, 25). Nevertheless, we hypothesized that the use of bicarbonate may increase its oral bioavailability.

METHODS AND MATERIALS

Chemicals

The synthesis of l-3′-Fd4C (molecular weight, 227.19) (Fig. 1) and the internal standard 3TC (lamivudine) have been described else where (5, 13, 30). The chemical purity of each compound was determined by high-performance liquid chromatographic (HPLC) and spectral analysis to be greater than 98%. Acetonitrile (HPLC grade) and all other chemicals (analytical grade) were obtained from Fisher Scientific (Fair Lawn, N.J.).

FIG. 1.

Chemical structure of β-l-3′-fluoro-2′,3′-didehydro-2′,3′-dideoxycytidine.

Pharmacokinetic studies in monkey.

Three female rhesus monkeys (Macaca mulatta) weighing from 5.1 to 6.5 kg were used for the pharmacokinetic studies. The animals were housed at the Yerkes Regional Primate Research Center at Emory University, which is fully accredited by the American Association for Accreditation of Laboratory Animal Care, in accordance with guidelines established by the Animal Welfare Act and the Guide for the Care and Use of Laboratory Animals (18) of the National Institutes of Health. Monkeys were administered a single dose of 25 mg/kg of body weight of l-3′-Fd4C i.v. in 10 ml of sterile saline. After a washout period of at least 4 weeks, monkeys were given oral doses (p.o.) of 25 mg/kg of body weight of l-3′-Fd4C, administered by gastric intubation, in a total volume of 10 ml of water, followed by a further 3 ml of water. Saturated sodium bicarbonate (10 ml) was administered by nasogastric tube to each monkey for the bicarbonate oral administration (o.b.) prior to treatment with l-3′-Fd4C. Animals were maintained on their backs on a heating pad, covered with a blanket and under anesthesia for 4 h after dosing with a mixture of ketamine HCl (60 mg) and tiletamine HCl plus zolazepam HCl (Telazol; 20 mg) intramuscularly, and monitored for alertness. Additional anesthetics (30 to 60 mg of ketamine HCl) were administered as necessary. This method of anesthesia was used since it was considered less likely to cause a pharmacokinetic interaction than volatile anesthetics (31). Blood samples were taken at 0, 0.25, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 12, and 24 h after drug administration through the femoral vein. Cerebrospinal fluid (CSF) samples were taken from all treated monkeys at 1 h after drug administration by cisternal or lumbar tap with a 22-gauge needle. The monkeys were catheterized for urine collection. Urine samples were collected at 0 to 0.25, 0.25 to 0.5, 0.5 to 1, 1 to 1.5, 1.5 to 2, 2 to 3, 3 to 4, 4 to 6, and 6 to 8 h after dosing. Plasma, CSF, and urine samples were kept on ice and then frozen at −70°C until used for analyses.

HPLC analysis of l-3′-Fd4C in monkey plasma, urine, and CSF samples.

HPLC analysis was performed by using a Hitachi HPLC system (Tokyo, Japan) equipped with a model L-7100 pump, a model L-7420 detector, and a model L-7250 autosampler with a Phenomenex Columbus C₁₈ reverse-phase column (4.6 by 250 mm; 5-μm particle diameter). The mobile phase was isocratic 6% acetonitrile and 94% water. The flow rate was maintained at 0.7 ml/min. l-3′-Fd4C and the internal standard 3TC were detected at a wavelength of 270 nm and eluted from the column at 12.5 and 14 min, respectively.

Preparation of standards.

Standard solutions of l-3′-Fd4C were prepared in deionized water. Calibration plots for l-3′-Fd4C in plasma were prepared by adding standard solutions to the plasma of an untreated monkey at concentrations ranging from 0.1 to 100 μg/ml. Standard curves for the analysis of urine and CSF samples were prepared by using urine and deionized water, respectively, over the same concentration range. The intraday accuracy and precision of the assay methodologies for plasma were determined by assaying six samples per concentration level on the same day. For interday assay accuracy and precision determinations, samples were analyzed on three separate days. Precision was reported as relative standard deviations, and accuracy was calculated by comparing measured concentrations to the known values. The intraday and interday relative standard deviations were less than 9%, and the intraday and interday accuracies were greater than 91%. The standard curves were linear over a concentration range of 0.1 to 100 μg/ml (r² = 0.999). Under these conditions, l-3′-Fd4C was baseline separated in monkey plasma, urine, and CSF samples.

Extraction procedure.

For plasma samples, 2.0 μl of 3TC (0.5 mg/ml) was added as an internal standard to a 50-μl plasma sample in a microcentrifuge tube. Acetonitrile (250 μl) was added before vortexing and centrifuged at 7,800 × g for 5 min. The supernatant was evaporated to dryness, and the samples were reconstituted with 100 μl of 5% acetonitrile in water and injected onto the HPLC column. The HPLC method was validated according to the Food and Drug Administration's Guidelines for Industry Bioanalytic Method Validation protocol (26). The extraction recovery of l-3′-Fd4C was calculated by comparing the mean peak area for six extracted plasma samples with that of the three standard samples containing the same amount of nucleoside. Low (0.5 μg/ml), medium (5 μg/ml), and high (50 μg/ml) concentration levels were investigated. The percent extraction recovery was calculated as follows: 100 × A_extracted/A_standard, where A_extracted is the peak area of extracted drug from biological fluid and A_standard is the peak area of same amount of drug without extraction. The extraction recovery of l-3′-Fd4C was greater than 89%.

For CSF samples, 2 μl of internal standard (3TC; 0.5 μg/ml) was added to 50 μl of CSF samples and 48 μl of water, and 50 μl of sample was injected onto the HPLC column for analysis. Urine samples were diluted 20- to 50-fold, and 4 μl of the same internal standard was added to 200 μl of diluted urine, and the mixture was injected onto the HPLC for analysis.

Pharmacokinetic analysis.

The areas under the concentration-time curve after oral dose (AUC_oral), maximum plasma concentration (C_max), time of maximum concentration (T_max), bioavailability, and apparent terminal phase half-life (t_1/2 from 8 to 24 h) were compared for each animal following a 25 mg/kg of body weight dose of l-3′-Fd4C in the presence and absence of bicarbonate buffer by using noncompartmental analysis software (Model 200, WinNonlin, version 4.1, 2003; Pharsight Corp., Carey, N.C.), assuming extravascular administration. A paired t test of the nontransformed and log-transformed data showed no significant difference. Therefore, a two-compartment open pharmacokinetic model was fitted to the i.v. and oral (no bicarbonate) plasma data from each monkey by using a nonlinear regression curve-fitting program (WinNonlin, version 4.1, 2003; Pharsight Corp.). Fitting the model on the i.v. and oral data simultaneously enhances the fit of the intercompartment disposition parameters (k₁₂ and k₂₁, first-order disposition rate constants between compartment 1 and compartment 2), which may otherwise be obscured by the oral absorption, and avoids reporting two sets of disposition rate constants for the same animal. However, this approach assumes that the intercompartment disposition rate constants remain consistent once the drug reaches the circulation (4, 10). Initial estimates for model parameters were obtained by using parameters from similar compounds tested in rhesus monkeys (12). A weighting factor of 1/(predicted value)^1.5 was used for fitting the plasma data (Fig. 2). Since the accumulation of l-3′-Fd4C in the urine had not plateaued at 8 h, the fraction excreted in the urine and renal clearance (CL_renal) could not be calculated directly. The CL_renal value was derived from a model fitted to the urine excretion data by using the pharmacokinetic parameters derived from the plasma data. The following equation was used: dX_u/dt = CL_renal × C_p, where dX_u is the rate of renal accumulation of compound during time interval dt, and C_p is the predicted concentration of the drug in the plasma based on the plasma data coinciding with the urine collection interval (4). No weighting factor was used for the urine data.

FIG. 2.

Simultaneous fit of i.v. (•) and oral (▪) l-3′-Fd4C. The symbols and vertical lines represent observed average concentrations ± standard deviations, while the dashed lines represent model predictions. Oral doses in the presence of bicarbonate (□) are shown but not modeled (see text). l-3′-Fd4C doses were equal (25 mg/kg). To convert micrograms per milliliter to micromolar concentrations, multiply by 4.4.

Other pharmacokinetic parameters listed in Table 1 were calculated by using standard two-compartment equations. For the area under the plasma concentration-time curve following i.v. dose (AUC_iv) and the AUC_oral the equations were as follows: AUC_iv = dose/V · (α − k₂₁)/(α − β)/α + dose/V · (k₂₁ − β)/(α − β)/β and AUC_oral = F · AUC_iv, where V is the distribution volume of the central compartments, α and β are biexponential elimination rate constants (more rapid and less rapid, respectively) from plasma, and F is the fraction of oral dose absorbed. Other calculations were made as follows: for the terminal phase half-life, t_1/2β = ln (2)/β; for systemic clearance, CL = dose/AUC_iv; for the mean residence time after i.v. dosing, MRT_iv = [(A/α²) + (B/β²)]/[(A/α) + (B/β)], where A = dose/V · (α − k₂₁)/(α − β) and B = dose/V · (k₂₁ − β)/(α − β); for the mean residence time after oral dosing, MRT_oral = MRT_iv + 1/K_a, where K_a is the absorption rate constant; for the steady-state distribution volume, V_ss = CL · MRT_iv; and for the mean absorption time, MAT = 1/K_a (10, 12).

TABLE 1.

Pharmacokinetic parameters of L-3′-Fd4C (25 mg/kg) after simultaneous two-compartment model fits to intravenous and oral administrations in three rhesus monkeys^a

Parameter	Value for:
	Mk-1	Mk-2	Mk-3	Mean (± SD)	Pooled
CLsystemic (liters · h−1 · kg−1)	0.26	0.18	0.26	0.23 (0.05)	0.25
V (liters · kg−1)	0.47	0.44	0.38	0.43 (0.04)	0.42
k21 (h−1)	0.08	0.08	0.15	0.10 (0.04)	0.13
α (h−1)	0.97	0.67	1.04	0.90 (0.20)	0.93
β (h−1)	0.05	0.05	0.10	0.07 (0.03)	0.08
t1/2β (h)	14.74	15.40	7.22	12.75 (4.55)	8.56
Vss (liters · kg−1)	2.66	1.86	1.18	1.90 (0.74)	1.36
k10 (h−1)	0.56	0.40	0.69	0.55 (0.14)	0.60
k12 (h−1)	0.38	0.24	0.30	0.31 (0.07)	0.28
AUCiv (μg · h · ml−1)	96.23	140.87	95.86	110.98 (25.88)	99.65
MRTiv (h)	10.26	10.46	4.51	8.41 (3.37)	5.41
CLrenal (liters · h−1 · kg−1)	0.09	0.06	0.08	0.08 (0.01)	0.08
rc iv serum	0.97	0.97	0.99		0.99
rc iv urine	0.99	0.99	0.99		0.99
Ka (h−1)	0.85	0.48	0.61	0.65 (0.19)	0.62
Tlag (h)	0.84	0.20	0.54	0.53 (0.32)	0.58
AUCoral (μg · h · ml−1)	21.66	21.12	29.42	24.07 (4.65)	22.36
MRToral (h)	11.43	12.52	6.16	10.04 (3.40)	7.02
F(%)	23.00	15.00	31.00	23.00 (8.00)	22.00
Tmax (h)	1.5	2	1.5	1.67 (0.29)
Cmax (μg · ml−1)	4.20	3.14	5.84	4.4 (1.40)
rc oral serum	0.99	0.92	0.98		0.99
rc oral urine	0.99	1.00	0.99		0.99

^aMonkeys are identified as Mk-1, Mk-2, and Mk-3. r^c, the correlation between the predicted values using the model and the observed values; k₁₀, the elimination rate constant for the central compartment.

RESULTS

Pharmacokinetic studies.

The paired t test of nontransformed and log-transformed data of the parameters C_max, T_max, AUC_oral, bioavailability, and t_1/2 (8 to 24 h) obtained from noncompartmental analysis failed to detect significant differences in oral doses, in the presence and absence of bicarbonate (two-tailed P values of ≥0.45). Therefore, the two-compartment pharmacokinetic model was fitted for individual monkeys by using the i.v. and oral doses, without bicarbonate buffer data. The results are summarized in Table 1, and the pooled fit is depicted in Fig. 2.

The serum concentrations of l-3′-Fd4C at 24 h were below the limit of detection (LOD) for the i.v. doses but were above the LOD for the oral doses. The average percentages recovered in the urine within 8 h of dosing were 48.9, 11.2, and 13.5%, respectively, for the i.v., p.o., and o.b. administrations. The average concentration of l-3′-Fd4C in cerebrospinal fluid at 1 h following i.v. administration was 1.72 μM, which exceeds the EC₉₀ against HIV-1 by eightfold (0.22 μM). However, the concentrations of the compound in CSF after 1 h of p.o. and o.b. administrations were below the limit of detection (0.1 μg/ml or 0.4 μM).

DISCUSSION

l-3′-Fd4C is a nucleoside with potent and selective dual activity against HIV and HBV activities in vitro (5, 6, 30). This nucleoside is structurally similar to l-(2,3-dideoxy-2-fluoro-β-l-glyceropent-2-enofuranosyl)cytosine (l-2′-Fd4C) and d-D4FC (Reverset) (4, 16, 17). However, d-D4FC has a fluorine moiety at the 5 position of the pyrimidine base, while l-2′-Fd4C and l-3′-Fd4C are fluorinated in the sugar ring (2′ and 3′ positions). Since d-D4FC is unstable at gastric pH and has improved bioavailability in bicarbonate buffer, we compared the oral bioavailability of l-3′-Fd4C with and without bicarbonate. However, we failed to determine significant differences in C_max, T_max, AUC_0-∞ and t_1/2 (8 to 24 h) values (P ≥ 0.05) using only three animals. Furthermore, the two-compartment open model described the single 25 mg/kg i.v. and oral doses of each animal (r > 0.92; Table 1), and the data of the pooled model appear to describe both oral formulations, even though the data from the bicarbonate formulation were not used for model fitting (Fig. 2). Since the bioavailability (F %) of l-3′-Fd4C was not improved in the presence of sodium bicarbonate, other formulation or prodrug approaches should be considered to improve the oral absorption.

The C_max values observed for l-3′-Fd4C (10 to 26 μM) were at least comparable to d-D4FC (33.4 μM) and l-2′-Fd4C (23.1 μM), considering that 25 mg/kg of body weight of l-2′-Fd4C was administered, while 33 mg/kg was given in l-2′-Fd4C and d-D4FC. T_max values were also similar (1 to 4 h) (4, 16). The oral dose AUC and K_a values of l-3′-Fd4C (93 to 130 μM · h and 0.64 h⁻¹, respectively) and of d-D4FC (172 μM · h/ml and 0.64 h⁻¹, respectively) were comparable. Although l-3′-Fd4C was not observed in the CSF within 1 h of oral dosing, it was detected in the CSF 1 h after i.v. administration. However, it may have slow absorption through the blood-brain barrier, similar to d-D4FC, which could not be found after 2 h of oral dosing in two monkeys but was detected in the 3-h CSF samples (16). The oral bioavailability of l-3′-Fd4C (15 to 31%) was comparable to that of l-2′-Fd4C (18.9 to 20.1%). The total systemic clearance was 0.23 liter · h⁻¹ · kg⁻¹, which is similar to that of l-2′-Fd4C (0.25 liter · h⁻¹ · kg⁻¹), and the renal clearance was 0.08 liter · h⁻¹ · kg⁻¹. The elimination rate constant from the central compartment (k₁₀) of l-3′-Fd4C (0.55 h⁻¹) was also comparable to that of l-2′-Fd4C (0.58 h⁻¹) and d-D4FC (0.63 h⁻¹). However, the t_1/2β value of l-3′-Fd4C (12.45 h) was longer than that of l-2′-Fd4C (5.02 h) and d-D4FC (3.57 h). The t_1/2β values for zidovudine, DDC (2′,3′-dideoxycytidine), D4T (3′-deoxy-2′,3′-didehydrothymidine), and DDI (2′,3′-dideoxyinosine) range from 0.81 to 1.82 h in rhesus monkeys (3, 19, 21). The active forms of antiretroviral nucleoside agents are the intracellular nucleoside triphosphates, which generally have slower elimination rates than the parent nucleoside (29). The apparent terminal phase half-life for the oral doses (p.o. and o.b.) of l-3′-Fd4C were >12 h, suggesting that l-3′-Fd4C may have potential for further development as a once daily antiretroviral agent.

The antiviral activity of l-3′-Fd4C against 3TC-resistant virus (EC₉₀ value of <3.1 μM, or 0.7 μg/ml) and against wild-type virus (EC₉₀ value of <0.22 μM, or 0.05 μg/ml) was previously reported (5, 6, 25, 30). The concentration of l-3′-Fd4C in the plasma of rhesus monkeys remained higher than the EC₉₀ value against 3TC-resistant virus after 8 h for i.v. and 12 h for the oral doses and against wild-type HIV-1, 12 h after the i.v. and oral doses. The pharmacokinetic profile of l-3′-Fd4C suggests that this compound may have potential as an antiretroviral drug. Moreover, l-3′-Fd4C was found to be nontoxic in many cells and in the mitochondria of HepG2 cells at concentrations up to 100 μM, and in mice given up to 100 mg/kg of body weight intraperitoneal doses for 6 days (5, 25, 30). Therefore, l-3′-Fd4C could be a useful candidate for individuals coinfected with HBV and HIV.