Estimation of In Vivo Percutaneous Absorption of Emedastine from Bile Excretion Data Using a Deconvolution Method

Drug Metab. Pharmacokinet. 20 (5): 331–336 (2005).

Regular Article Estimation of In Vivo Percutaneous Absorption of Emedastine from Bile Excretion Data Using a Deconvolution Method Shoichi HARADA1, Fumiyoshi YAMASHITA2 and Mitsuru HASHIDA2 1Clinical 2Department

Project Management, Nippon Organon K.K., Osaka, Japan of Drug Delivery Research, Graduate School of Pharmaceutical Sciences, Kyoto University, Kyoto, Japan

Full text of this paper is available at http://www.jstage.jst.go.jp/browse/dmpk

Summary: In vivo percutaneous absorption of emedastine difumarate was investigated in rats and compared with rat skin in vitro. Since emedastine entering the systemic circulation is mostly excreted in bile, we ˆrst came up with the method of collecting bile with a minimal skin incision. In vivo skin permeation of the drug was estimated from biliary excretion data by deconvolution analysis. Prior to applying deconvolution analysis, it was conˆrmed that biliary excretion of emedastine was linear against its dose. When the in vivo permeation proˆle estimated by deconvolution was compared with the in vitro proˆle, the lag time for permeation was signiˆcantly shorter in vivo than in vitro, whereas the skin permeability coe‹cient was almost the same. If we presume a two-layer diŠusion model, then this ˆnding may primarily be due to the shorter diŠusion length of the dermis.

Key words: emedastine; bile excretion; deconvolution method; percutaneous absorption; in vivo permeation.5–7) Previous comparative in vitro W in vivo studies failed to provide exact quantitative correlations, but some positive correlations were observed.5,6) We previously analyzed skin permeation of three drugs, varying in their lipophilicity based on a two-layer in vivo diŠusion model, and suggested that the in vitro W diŠerence in their skin permeation might be associated with the diŠerence in the diŠusion length of a viable layer and the degree of hydration of the stratum corneum.8) Thus, in order to gain further insight into the relationship between pharmacological activity and skin permeability, it is necessary to evaluate the in vivo skin permeation of emedastine. Deconvolution analysis is a useful method which estimates the rate of absorption by assuming linear systemic pharmacokinetics of the drug.9–13) Therefore, in vivo proˆles computed through deconvolution can be directly compared with in vitro permeation proˆles. We previously proposed that urinary excretion data may be more appropriate than blood concentration data in evaluating percutaneous absorption in terms of the acceptable number of samples and the concentration of drug in the samples.14) However, as previously reported by Sakai et al.,15) after oral administration of 14Cemedastine difumarate, most of the dose is rapidly

Introduction Emedastine difumarate, a potent anti-histaminic agent, is a promising anti-allergy agent for the treatment of dermatitis, urticaria, etc.1) We have previously demonstrated that emedastine was readily absorbed when administered cutaneously.2) We also found that emedastine permeation was strongly dependant upon vehicle type, exhibiting a V-shaped bilinear relationship between the quasi-steady state ‰ux of emedastine and the dielectric constant of solvents.3) Our most recent work showed that the antihistamine eŠect of emedastine applied cutaneously varied greatly depending upon the vehicle, due to the diŠerence in skin permeability.4) When the antihistamine eŠect of emedastine was plotted against the calculated in vitro transdermal ‰ux, and not the concentration applied, their relationship was sigmoidal and common, regardless of the vehicles used. We proposed that the in vitro permeability of emedastine would be a good measure of its pharmacological eŠect.4) While we have proposed in vitro permeability of emedastine would be a good measure of its pharmacological eŠect,4) a number of transdermal studies have demonstrated an in vitro W in vivo diŠerence in skin

Received; April 26, 2005, Accepted; August 2, 2005 To whom correspondence should be addressed : Shoichi HARADA, Clinical Project Management, Nippon Organon K.K. Dojima Avanza 14F, 1-6-20 Dojima, Kita-ku, Osaka 530-0003, Japan. Tel. ＋81-6-6347-9788, Fax. ＋81-6-6347-9795, E-mail address: shoichi.harada＠organon.jp

331

332

Fig. 1. 14 C

Shoichi HARADA et al.

Chemical structure of emedastine difumarate. *labeled with

excreted into bile: 24 h after ingestion, biliary and urinary excretion of radioactivity were 89z and 11z, respectively. Therefore, it seems to be more appropriate to use biliary excretion data to evaluate absorption. In the present study, we collected bile samples after cutaneous administration of emedastine difumarate, using a novel surgical technique, and analyzed the bile excretion proˆles of emedastine using a deconvolution method. Finally, the results in the in vivo experiments were compared with in vitro results. Materials and Methods 14

C-labeled emedastine difumarate (1.90 GBq W mmol) was synthesized by Daiichi Pure Chemicals Co., Ltd (Fig. 1). Its purity as measured by TLC was 98z or higher. The other reagents used were of analytical grade. In vitro skin permeation: Abdominal hairs were removed with a hair clipper in 7- to 8-week-old male Wistar rats under ether anesthesia. Following an incision of the carotid artery, the abdominal skin was excised and the adipose tissue adhering to the dermis was carefully removed. The skin was punched out into a 3 cm diameter disk and mounted on a ‰ow-through type diŠusion cell (exposed area: 3.14 cm2 ).16) The receiver chamber was perfused with pH 7.4 phosphate buŠered saline (PBS) containing streptomycin sulfate (50 mg W L) and penicillin G potassium (30 mg W L) at a rate of 6 mL W h. The epidermis was treated with 1 mL PBS for 6 h. The donor ‰uid was then replaced with 1 mL of PBS containing 10.0 mg W mL emedastine difumarate. The receptor ‰uid was collected every 60 minutes for 6 h. The diŠusion cell was kept in a thermostat at 379C throughout the experiment. In vitro metabolism: One gram of skin tissue excised from rats was homogenized in 5 mL of pH 7.4 PBS on ice. After centrifugation of the homogenate at 10,000 g for 30 min, the supernatant was taken for the in vitro metabolism study. The metabolism experiment was started by mixing 25 mL PBS containing emedastine difumarate with 5 mL of the homogenate supernatant. In the control group, the test solution was mixed with 5 mL PBS. The initial concentration of emedastine difumarate was set at 0.05 mM. One milliliter aliquots of the mixture were taken at 0, 2, 4, 6, and 24 h. The samples were immediately deproteinized

with an equal volume of methanol and subjected to centrifugation at 3,000 g followed by the HPLC assay. HPLC method: HPLC was performed using an HPLC Series 1050 (Hewlett Packard, Wilmington, DE) connected with an UV detector (wavelength: 280 nm). A reverse phase Inertsil ODS-2 column (5 mm, 5 ID×150 mm, GL Science, Tokyo, Japan) served as the column. The mobile phase was composed of a 1:1 mixture of acetonitrile and 0.025M phosphate buŠer (pH 2.4) containing 0.25z sodium lauryl sulfate. The ‰ow rate of the mobile phase was set at 1.2 mL W min. The calibration curve was prepared from 5 reference solutions of emedastine difumarate (1–100 mg W mL) (r2 ＝0.9998). The detection limit was 0.5 mg W mL. The reproducibility for the 1 mg W mL reference solution was 2.7z. Intravenous administration: Male Wistar rats, weighing about 200 g, received an intraperitoneal dose of 1 g W kg urethane. The duodenum was pulled out through a 3-cm incision in the right lateral abdomen and a polyethylene tube (0.28 mm ID, 0.61 mm ID; Dural) was inserted into the bile duct. The intestine was then returned to the abdominal cavity and the incision was kg sutured. Physiological saline containing 20 kBq W 14 C-labeled emedastine difumarate with 0.1 or 10 mg W kg non-radioactive compound was rapidly injected into the femoral vein of the rat. The spontaneously excreted bile was collected into a 25 mL glass vial every 15 min and 700 mL Soluene-350 (Packard Instrument, IL) was added. The mixture was agitated, incubated at 509C overnight, and neutralized with 200 mL of 3N HCl. After adding 5 mL of liquid scintillation medium (Clear-Sol, Nacalai Tesque, Kyoto, Japan) to the lysate, radioactivity was measured using a liquid scintillation counter (LSC-5000, Beckman, Tokyo, Japan). In vivo percutaneous absorption: Abdominal hairs were removed with a hair clipper in 7- to 8-week-old male Wistar rats (body weight about 200 g) under anesthesia with 1 g W kg intraperitoneal urethane. The bile duct of the rat was cannulated in the same manner as in the intravenous injection experiments. A cylindrical glass cell with an inner diameter of 2 cm was attached to the abdominal skin using surgical glue (Aron Alpha A, Sankyo Co., Tokyo, Japan). One milliliter PBS was placed into the glass cell to stabilize the skin. Six hours later, the PBS solution was replaced mL 14C-emedastine with 1 mL PBS containing 20 kBq W difumarate and 10.0 mg W mL cold compound. The bile was collected and treated in the same manner as in the intravenous injection experiments. At the end of the experiment, the drug solution was collected by washing oŠ with approximately 30 mL of water. The skin tissue was then excised and solubilized with Soluene-350 (Packard Instrument, IL). The radioactivity associated with bile, donor solution, and skin samples was

Estimation of In Vivo Percutaneous Absorption of Emedastine

333

Table 1. Stability of emedastine in phosphate-buŠer saline (pH 7.4) containing rat skin homogenate at 379C Time(h) 2

4

Homogenate 100.1±1.32 100.5±1.64 Control

—

a)

—

6

24

99.6±0.782

99.0±0.617

100.2±0.856 100.7±0.434

The stability test was studied at initial concentration of 0.05 mM in the v) rat skin homogenate at 379C. Values are the presence of 3.3z (w W mean±S.D. (z of initial concentration) of three experiments. a) not tested

measured using a liquid scintillation counter. Data analysis: The amount of emedastine absorbed percutaneously was estimated from bile excretion data using a deconvolution method similar to a previously reported method.17) As the ˆrst step, the bile excretion rate-time proˆle following intravenous injection was approximated to the following bi-exponential equation:

dXb W dt＝A*e－a t＋B*e－bt

Fig. 2. Biliary excretion of radioactivity following intravenous injection of 14C-emedastine difumarate with 0.1 (open symbol) and 10 (closed symbol) mg W kg cold emedastine difumarate in rats. Each point represents the mean±S.D. value of tree or four animals.

(Eq. 1)

where dXb W dt indicates the bile excretion rate (z of dose W h). The coe‹cients A, B, a, and b were obtained by a non-linear regression method with a Gauss-Newton algorithm implemented in MULTI.18) The amount of emedastine absorbed into the systemic circulation was then calculated using a deconvolution algorithm17) by performing rectangular approximation of bile excretion rate-time proˆles following cutaneous administration. The skin permeation rates were calculated from the slope of the linear portion of cumulative permeation amount-time proˆles. The lag time for permeation was the intercept of the regression line to the x-axis. Results

In vitro metabolism of emedastine: Emedastine difumarate was incubated in PBS containing rat skin homogenate supernatant. As summarized in Table 1, emedastine was not metabolized at all, even when incubated for 24 h. This suggests that the metabolism of the drug during transport across the skin is negligible. Biliary excretion of 14C-emedastine after intravenous injection: Figure 2 shows the biliary excretion ratetime proˆles of radioactivity after intravenous injection of 14C-labeled emedastine difumarate in rats. The radioactive compound was administered with two diŠerent doses (0.1 and 10 mg W kg) of cold emedastine difumarate. As shown in Fig. 2, the biliary excretion of 14 C-radioactivity did not diŠer between the two doses. Cumulative excretion of radioactivity into bile within 4 h was 69 and 65z of the dose for 0.1 and 10 mg W kg, respectively. When a biliary excretion rate-time proˆle after intravenous injection of 14C-emedastine difumarate was

Fig. 3. Biliary excretion of radioactivity following cutaneous administration of 14C-emedastine difumarate in rats. Each point represents the mean±S.D. value of four experiments.

analyzed by a nonlinear curve ˆtting procedure, it was approximated to the following bi-exponential equation:

dXb W dt＝130*exp(－2.10*t)＋12.0*exp(－0.40*t) (Eq. 2) where dXb W dt indicates the excretion rate into bile (z of dose W h), and t denotes time (h). Estimation of in vivo percutaneous absorption and comparison with in vitro absorption: To evaluate percutaneous absorption of a drug, the skin incision should be minimized. Therefore, bile duct cannulation was performed with a short (3 cm) incision of the lateral abdominal skin and muscle wall. Figure 3 shows biliary

334

Shoichi HARADA et al. Table 3. Amounts of drug recovered at the end of 4-h in vivo absorption experiments Recovery(z applied)a) Donor

Skin

Bile

Absorbedb)

Totalc)

91.3±8.49 1.22±0.252 0.213±0.134 1.12±0.213 93.7±7.86 a) Data are expressed as the mean±SD of four experiments. b) Values were calculated using a deconvolution method. c) Values are the sum of the amounts of emedastine in the donor, skin, and absorbed.

Fig. 4. In vivo (closed symbol) and in vitro (open symbol) permeation through rat skin. In vivo absorption proˆles were estimated from biliary excretion data using a deconvolution method. Each point represents the mean±S.D. value of at least three experiments.

Table 2. Permeability coe‹cient and lag time for permeation of emedastine difumaratea)

In vitro In vivo

Permeability coe‹cient ×103 (cm W h)

Lag time (h)

0.980±0.473 1.21±0.317

1.91±0.428b) 0.994±0.389

a) Permeability coe‹cient and lag time were estimated from the linear portion of permeation proˆles (Fig. 4). The data represent the mean± SD of four experiments. b) Signiˆcantly diŠerent from in vivo (pº0.05).

excretion of radioactivity following cutaneous administration of 14C-emedastine difumarate in rats. The excretion rate of radioactivity into bile increased with h at 4 h. The time, achieving about 0.1z of the dose W cumulative excretion into bile until 4 h was 0.21z of the dose. Since the biliary excretion proˆle re‰ects all ADME processes, a deconvolution analysis was made to estimate the absorption rate of 14C-emedastine difumarate. The in vivo absorption proˆle of 14Cemedastine difumarate was estimated from biliary excretion data following cutaneous administration and intravenous injection (Eq. 2), and compared with its in vitro absorption through excised rat skin (Fig. 4). The amount absorbed in vivo was slightly greater than that absorbed in vitro. Table 2 summarizes the permeability coe‹cient and lag time for permeation, which were calculated from in vitro and in vivo permeation proˆles. The in vitro and in vivo permeability coe‹cients were similar. However, the lag time for permeation was signiˆcantly shorter in vivo. Thus, the greater in vivo permeation seen in Fig. 4 would be principally due to a shorter lag time for permeation. At the end of the experiment, the amounts of

emedastine difumarate remaining in the donor ‰uid, skin, and receptor ‰uid were measured. Table 3 shows the results of this measurement as well as the cumulative excretion calculated by the deconvolution method. When emedastine difumarate was administered to skin, most of the emedastine remained in the donor ‰uid (91z of the dose) and skin (1.2z of the dose). The amount of the drug systemically absorbed, as calculated using the deconvolution method, was about 1.1z of the dose. At the end of the experiment, the emedastine concentration in rat serum was lower than the detection mL). limit (about 50 ng W Deconvolution analysis requires linearity of systemic pharmacokinetics. In intravenous injection experiments, the biliary excretion of emedastine difumarate was linear, at least within a dose range of 0.1–10 mg W kg (Fig. 2). In Fig. 2, the maximum excretion rate for the h, while the rate of higher dose was roughly 2,000 mg W excretion at 4 h for the lower dose was roughly 0.5 mg W h. Since the rate of biliary excretion after cutaneous administration was approximately 10 mg W h (0.1z of h), it was assumed to fall into the range of linear dose W pharmacokinetics. Discussion As mentioned before, several investigations have demonstrated an in vitro W in vivo diŠerence in the skin permeation of drugs. Although we previously suggested that the in vitro transdermal ‰ux of emedastine might be a good measure of its pharmacological activity, further studies of the in vitro W in vivo diŠerence are required. The present study explored a method for evaluation of in vivo percutaneous absorption of emedastine difumarate. We have already demonstrated that in vivo percutaneous absorption of three drugs (mannitol, acyclovir and butylparaben) can be evaluated from urinary excretion data by a deconvolution method.8) However, as previously reported by Sakai et al., most of the dose was rapidly excreted in bile after oral administration of emedastine difumarate.15) In the present study, we came up with the developed the method of collecting bile samples with a minimal skin incision. It should be noticed that the blood level of the drug after

Estimation of In Vivo Percutaneous Absorption of Emedastine

cutaneous administration is too low to be detected due to the strong barrier properties of skin. Although the non-radioactive compound is generally more practical to handle, radioactive emedastine difumarate was used for the in vivo percutaneous absorption study because of the limitations of detection. In fact, Awata et al.19) reported that no intact emedastine was detected in the bile after oral administration and, moreover, four diŠerent metabolites of emedastine were found. It should be noted that the use of total radioactivity data associated with the intact drug and its metabolites is theoretically acceptable in performing a deconvolution analysis. To apply a deconvolution method to estimate the absorption rate of a drug, the following two conditions must be met. Firstly, the relationship between biliary excretion and dose should be linear, since convolution calculation is theoretically the summation of the responses to impulse inputs. When 14C-emedastine difumarate was injected intravenously with 0.1 or 10 mg W kg cold compounds, no diŠerence in biliary excretion proˆle between the two doses was observed (Fig. 2). Since the level of biliary excretion after cutaneous application of emedastine was intermediate between the levels found after two diŠerent dose injections (0.1 kg), the linearity of systemic pharmacokiand 10 mg W netics would be assured in this study. The second point to be conˆrmed in deconvolution analysis is that no metabolism occurs during the absorption processes. Since the metabolic fate of emedastine after entering the systemic circulation was similar to that after intravenous injection, the form that enters the systemic circulation after cutaneous administration should be substantially the same as the form injected intravenously, i.e., an intact drug. We investigated the metabolic activity of rat skin on this basis (Table 1). Since emedastine was hardly metabolized in the presence of rat skin homogenate, the second condition for deconvolution analysis was also satisˆed. The in vivo permeation proˆle of emedastine difumarate was estimated from its biliary excretion proˆle by a deconvolution method and compared with the in vitro permeation proˆle in excised rat skin. Although the in vitro and in vivo permeability coe‹cients were similar, the lag time for permeation was signiˆcantly shorter in vivo than in vitro (Fig. 4 and Table 2). This result can be explained by assuming that the skin barrier consists of two diŠerent layers, namely the stratum corneum and the underlying layer. Based on a two-layer diŠusion model, the steady-state permeability (Papp) and the lag time for permeation (LT) can be expressed as,

Papp＝ LT＝ Lsc Ksc Dsc

Ø

335

1 Lsc Ld ＋ Ksc Dsc K d Dd

»

Lsc2 L2 Ld ＋ d ＋ 6Dsc 2D d L d Dd Lsc Ld ＋ Ksc Dsc K d Dd

(Eq. 3)

Ø

Lsc2 L2 ＋ d 2Dsc 6D d

»

(Eq. 4) where K, D, and L are partition coe‹cient, diŠusion coe‹cient, and thickness, respectively, and subscripts sc and d are the stratum corneum and the underlying layer, respectively. When permeation across the stratum corneum is rateLsc9Kd Dd W Ld), these equations might limiting (Ksc Dsc W be approximated with:

Ksc Dsc Lsc

(Eq. 5)

Lsc2 L2 ＋ d 6Dsc 2D d

(Eq. 6)

Papp＝ LT＝

In (Eq. 6), the diŠusion coe‹cient in the stratum corneum is much smaller than in the underlying aqueous layer. However, the second term of (Eq. 6) cannot be neglected because the square term of thickness of the underlying layer (Ld2) is large enough to cancel out the diŠerence in diŠusion coe‹cient. In other words, the lag time for permeation could be aŠected by the thickness of the underlying layer. Under in vivo conditions, since the compounds are removed in the upper dermis, which contains capillary blood vessels, it is possible that the in vivo diŠusion length in the dermis is shorter than in the in vitro condition. Thus, the in vivo W in vitro diŠerence in the lag time can be explained, even though there was no diŠerence in permeability coe‹cient. However, it in vitro cannot still be ruled out that the in vivo W diŠerence might be due to the diŠerence in method of measurement between in vitro and in vivo (that is, nonradioactive compound and radioactive one, respectively). In a previous study,4) we evaluated the anti-histaminic eŠect of emedastine in rats. The concentration- and time-dependent pharmacological eŠect of emedastine in various formulations was strongly correlated with the rate of permeation across excised rat skin. Based on these observations, we concluded that in vitro transdermal ‰ux is a good measure of the anti-histaminic eŠect of emedastine. On the other hand, the present study demonstrated that the lag time for permeation was approximately one hour shorter in vivo. It is possible that that this diŠerence in lag time coincided with a delay for expression of pharmacological activity. However, as long as the in vitro and in vivo permeability

336

Shoichi HARADA et al.

coe‹cient is similar, in vitro transdermal ‰ux would undoubtably be a good measure of pharmacological activity. References 1) 2)

3)

4)

5)

6)

7)

8)

9)

Kanebo Ltd., Japan Patent 58–79983 (1983) [Chem. Abstr., 99, 139971 (1983)]. Harada, S., Takahashi, Y. and Nakagawa, H.: Transdermal administration of emedastine, Biol. Pharm. Bull., 16: 884–888 (1993). Harada, S., Takahashi, Y., Nakagawa, H., Yamashita, F. and Hashida, M.,: EŠect of vehicle properties on skin penetration of emedastine, Biol. Pharm. Bull., 23: 1224–1228 (2000). Harada, S., Nakada, Y., Yamashita, F. and Hashida, M.,: Biopharmaceutical considerations on antihistamine eŠects of topically administered emedastine. J. Pharm. Sci., 94: 17–24 (2005). Bronaugh, R. L. and Franz, T. J.: Vehicle eŠects on percutaneous absorption: in vivo and in vitro comparisons with human skin. Bri. J. Dermatol., 115: 1–11 (1986). Bronaugh, R. L. and Maibach, H. I.: Percutaneous absorption of nitroaromatic compounds: in vivo and in vitro studies in the human and monkey. J. Invest. Dermatol., 84: 180–183 (1985). Wester, R. C., Hui, X., Hartway, T., Maibach, H. I., Bell, K., Schell, M. J., Northington, D. J., Strong, P. and Culver, B. D.: In vivo percutaneous absorption of boric acid, borax, and disodium octaborate tetrahydrate in humans compared to in vitro absorption in human skin from inˆnite and ˆnite doses. Toxicol. Sci., 45: 42–51 (1998). Yamashita, F., Bando, H., Koyama, Y., Kitagawa, S., Takakura, Y. and Hashida M.: In vitro and in vivo analysis of skin penetration enhancement based on a two-layer diŠusion model with polar and nonpolar routes in the stratum corneum. Pharm. Res., 11: 185–191 (1994). Kiwada, H., Hayashi, M., Fuwa, T., Awazu, S. and Hanano, M.: The pharmacokinetic study on the fate of 8-hydroxyquinoline in rat. Chem. Pharm. Bull., 25: 1566–1573, (1977).

10)

11)

12)

13)

14)

15)

16)

17)

18)

19)

Vaughan, D. P. and Dennis, M.: Mathematical basis of point-area deconvolution method for determining in vivo input functions. J. Pharm. Sci., 67: 663–665 (1978). Cutler, D. J.: Numerical deconvolution by least squares: use of polynomials to represent the input function. J. Pharmacokin. Biopharm., 6: 243–263 (1978). Veng-Pedersen, P.: An algorithm and computer program for deconvolution in linear pharmacokinetics. J. Pharmacokin. Biopharm., 8: 463–481 (1980). Iga, K., Ogawa, Y., Yashiki, T. and Shimamoto, T.: Estimation of drug absorption rates using a deconvolution method with nonequal sampling times. J. Pharmacokin. Biopharm., 14: 213–225 (1986). Yamashita, F., Koyama, Y., Sezaki, H. and Hashida, M.: Estimation of a concentration proˆle of acyclovir in the skin after topical administration. Int. J. Pharm., 89: 199–206 (1993). Sakai, T., Takahashi, H., Hamada, T., Awata, N. and Watanabe, J.: The biological fate of 1-(2-ethoxyethyl)-2(4-methyl-1-homopiperazinyl) benzimidazole difumarate (KB-2413) I Absorption and excretion after oral administration to rats and guinea Pigs, Xenobio. Metab. Dispos., 2: 123–131 (1987). Yamashita, F., Yoshioka, T., Koyama Y., Okamoto, H., Sezaki, H. and Hashida, M.: Analysis of skin penetration enhancement based on a two-layer skin diŠusion model with polar and nonpolar routes in the stratum corneum: Dose-dependent eŠect of 1-geranylazacycloheptan-2-one on drugs with diŠerent lipophilicities. Biol. Pharm. Bull., 16: 690–697 (1993). Kiwada, H., Morita, K., Hashida, M., Awazu, S. and Hanano, M.: A new numerical calculation method for deconvolution in linear compartment analysis of pharmacokinetics. Chem. Pharm. Bull., 25: 1312–1318 (1997). Yamaoka, K., Tanigawara, Y., Nakagawa, T. and Uno, T.: A pharmacokinetic analysis program (MULUTI) for microcomputer. J. Pharm. Dyn., 4: 879–885 (1981). Awata, N., Takahashi, H., Sakai, T. and Hamada, T.: Identiˆcation of the metabolites of an antiallergic agent, 1-(2-ethoxyethyl)-2-(hexahydro-4-methyl-1H-1,4diazepine-1-yl)-1H-benzimidazole difumarate (KG-2413), in rats. Yakugaku Zasshi, 109: 318–328 (1989).