Development and validation of an in vitro–in vivo correlation for buspirone hydrochloride extended release tablets

Journal of Controlled Release 88 (2003) 147–157 www.elsevier.com / locate / jconrel

Development and validation of an in vitro–in vivo correlation for buspirone hydrochloride extended release tablets Sevgi Takka*, Adel Sakr, Arthur Goldberg 1 Industrial Pharmacy Program, University of Cincinnati, Medical Center, Cincinnati, OH, USA Received 10 September 2002; accepted 11 December 2002

Abstract The aim of this study was to develop an in-vitro–in-vivo correlation (IVIVC) for two buspirone hydrochloride extended release formulations and to compare their plasma concentrations over time with the commercially available immediate release (IR) tablets. In vitro release rate data were obtained for each formulation using the USP Apparatus 2, paddle stirrer at 50 and 100 rpm in 0.1 M HCl and pH 6.8 phosphate buffer. A three-way crossover study in 18 healthy subjects studied a 30 mg ‘‘Fast’’ (12 h) and 30 mg ‘‘Slow’’ (24 h) formulation of buspirone hydrochloride given once a day, and 2315 mg immediate release tablets dosed at a 12 h interval. The similarity factor ( f2 ) was used to analyze the dissolution data. A linear correlation model was developed using percent absorbed data and percent dissolved data from the two formulations. Predicted buspirone hydrochloride concentrations were obtained by use of a curve fitting equation for the immediate release data to determine the volume of distribution and fraction absorbed constants. Prediction errors were estimated for Cmax and area under the curve (AUC) to determine the validity of the correlation. pH 6.8 at 50 rpm was found to be the most discriminating dissolution method. Linear regression analyses of the mean percentage of dose absorbed versus the mean in vitro release resulted in a significant correlation (r 2 .0.95) for the two formulations. An average percent prediction error for Cmax was 20.16%, but was 16.1%, for the AUCs of the two formulations.  2002 Elsevier Science B.V. All rights reserved. Keywords: Buspirone hydrochloride; In-vitro–in-vivo correlation; Extended release tablet

1. Introduction Establishing a correlation between the in-vivo

*Corresponding author. Present address: University of Gazi, Faculty of Pharmacy, Pharmaceutical Technology Department, Ankara 06330, Turkey. Tel.: 190-312-212-2107; fax: 190-312212-7958. E-mail address: [email protected] (S. Takka). 1 Current address: Pharmaceutical Development Group, Inc., 624 Sand Hill Circle, Menlo Park, CA 94025, USA.

plasma concentration profile and the in-vitro dissolution profile of an extended release (ER) formulation has been of great interest for a number of years. Extended release (ER) of drugs in the gastrointestinal (GI) tract following oral administration is the intended rate-limiting factor in the absorption process. It is therefore desirable to use in-vitro data to predict in vivo bioavailability parameters for the rational development and evaluation process for extended release dosage forms [1,2]. The ultimate goal of an in-vitro–in-vivo correla-

0168-3659 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved. doi:10.1016 / S0168-3659(02)00490-X

148

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

tion (IVIVC) should be to establish a meaningful relationship between in-vivo behavior of a dosage form and in-vitro performance of the same dosage form, which would allow in-vitro data to be used as a surrogate for in-vivo behavior. A meaningful IVIVC for extended release dosage forms would be of benefit as a surrogate for bioequivalence studies which might typically be required with scale up or minor post-approval changes (SUPACs) in formulation equipment, manufacturing process or in the manufacturing site. A meaningful IVIVC could lead to improved product quality and decreased regulatory burden [3–5]. It is well known that in-vitro dissolution testing is a powerful and useful method for determining product quality. The utility of in-vitro dissolution as a surrogate for in-vivo bioavailability is very attractive and has been demonstrated for several products. Furthermore, to utilize this dissolution test as a surrogate for bioequivalence, the IVIVC must be predictive of in-vivo performance of the product. Levels A, B, C, and multiple level C correlation has been described in the US Food and Drug Administration (FDA) IVIVC guidance. The most useful of these is a level A correlation, which is described as a point-to-point correlation in which the in vivo percent absorbed curve is compared to in vitro percent dissolved curve. Generally these correlations are linear and are considered most informative and very useful from a regulatory viewpoint. The FDA guidance describes the methods of evaluation of prediction error internally and / or externally. Internal validation refers to how well IVIVC model describes the data used to develop the correlation. External validation determines how well the IVIVC model describes data that was not used in the development of the model [4]. Numerous IVIVC studies of sustained or extended release formulations have been previously reported [6–11], however, there is no any extended release buspirone hydrochloride formulations in the US or European markets. According to the Biopharmaceutics classification system, buspirone hydrochloride can be classified as a ‘‘Class 1’’ drug, i.e., high solubility and permeability. In addition, it is a highly variable drug, exhibiting a very high first pass metabolism and only about 4% of an orally administered dose will reach the systemic circulation

unchanged after oral administration [12,13]. Therefore, the purpose of this study was to develop an IVIVC for a novel hydrophilic matrix extended release buspirone hydrochloride tablets. The validity of the correlation was established through the external predictability approach, by using the data from one study to predict the plasma concentration of a similar dose form, with a different rate of release.

2. Materials and methods

2.1. Materials Buspirone hydrochloride was purchased from Brantford Chemicals (Canada). Methocel  K100M and K100 LV were purchased from Dow Chemical Company (Midland, MI, USA). Methacrylic acid copolymer (Eudragit  L 10055) was a gift from Rohm America (Somerset, NJ, USA). Silicified microcrystalline cellulose (Prosolv  ) was purchased from Penwest (Patterson, NY, USA), and magnesium stearate was purchased from Mallinckrodt (St. Louis, MO, USA).

2.2. Formulations Extended release formulations of buspirone hydrochloride were developed using hydroxypropyl methylcellulose (HPMC) as one of the release rate controlling excipients, and Eudragit L100-55 as the other controlled release polymer, and included silicified microcrystalline cellulose as filler, and magnesium stearate as lubricant. The formulations were designed to release buspirone hydrochloride at two different rates referred to as ‘‘Slow’’ and ‘‘Fast’’. The high-viscosity HPMC (Methocel K100M) and the low-viscosity HPMC (Methocel K100LV) are used for slow and fast release, respectively.

2.3. Dissolution testing The dissolution behavior of buspirone hydrochloride was continuously recorded using a fully automated dissolution apparatus. The release characteristics of the formulations were determined using USP Apparatus II, at 50 and 100 rpm, in 0.1 M HCl or pH

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

6.8 phosphate buffer maintained at 37 8C. Dissolution tests were performed on six tablets and the amount of drug released was analyzed spectrophotometrically at a wavelength of 238 nm. Dissolution samples were collected at the following times: 0, 0.5, 1.0, 1.5, 2.0, 3.0, 4.0, 6.0, 8.0, 10, 12 and 24 h.

2.4. Bioavailability study The bioavailability study has been previously reported [14]. Briefly, this was an open-label, fasting, single dose, three-treatment crossover study using normal healthy volunteers. Subjects provided informed consent to participate, and all studies were approved by the institutional review boards of the clinical sites conducting the studies. Eighteen male, non-smoking volunteers were enrolled in the study and received two extended release, once-per-day, formulations (slow and fast) of buspirone hydrochloride (30 mg) in a randomized fashion. In addition to the extended release formulations, an immediate release (2315 mg) of buspirone hydrochloride (BUSPAR ) was also administered. The order of treatment administration was randomized in three sequences (ABC, BCA, CAB) in blocks of three. Blood samples were obtained at 22 time points from pre-dose (0 h) until 36 h post-dose. A washout period of 1 week was allowed between dose administrations. Subjects fasted for 12 h prior to the morning drug administration when the extended and immediate release products were administered, and for 4 h prior to the evening drug administration of the immediate release product. The plasma samples were stored at 220 8C until assayed.

2.5. Assay method for buspirone hydrochloride An analytical method for the determination of buspirone and 1-(2-pyrimidyl) piperazine (1-PP) in human heparinized plasma was developed and validated using liquid chromatography–tandem mass spectrometry (LC–MS–MS). The method determined concentrations of buspirone using a calibration range of 0.05 to 10.0 ng / ml. The accuracy of the assay for buspirone (as determined from the calibration standards and control samples) ranged from 97.8 to 102% and 97.1 to 100%, respectively.

149

2.6. In vitro dissolution data analysis The dissolution profiles for each formulation were determined by plotting the cumulative percent of buspirone hydrochloride dissolved at various time points. The in-vitro drug release profiles of the two ER dosage forms were compared using the similarity factor, f2 , presented in the following equation [15]: f2 5 50 log

HF O

1 n 11] F 2 Std 2 n t 51 s t

G

20.5

3 100

J (1)

where Ft and St are the percent dissolved at each time point for the fast product and the slow product, respectively.

2.7. In vivo data analysis The buspirone hydrochloride concentration–time data were evaluated by analysis of variance using SAS  version 6.12, GGLM procedure and an F-test to determine statistically significant differences (a 5 0.05) by Pharmakinetics Laboratories. The measured plasma concentrations were used to calculate the area under the plasma concentration–time profile from time zero to the last concentration time point (AUC ( 0 – t ) ). The AUC (0 – t ) was determined by the trapezoidal method. AUC (0 – `) was determined by the following equation: C(t ) AUC ( 0 – ` ) 5 AUC ( 0 – t ) 1 ] ke

(2)

k e was estimated by fitting the logarithm of the concentrations versus time to a straight line over the observed exponential decline. The Wagner–Nelson method [16] was used to calculate the percentage of the buspirone hydrochloride dose absorbed: F(t ) 5 C(t ) 1 k e AUC ( 0 – t )

(3)

where F(t ) is the amount absorbed. The percent absorbed is determined by dividing the amount absorbed at any time by the plateau value, k e AUC (0 – `) and multiplying this ratio by 100:

F

Cstd 1 k e AUC s0 – td % dose absorbed 5 ]]]]] k e AUC s0 – `d

G

3 100 (4)

150

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

2.6. In-vitro–in-vivo correlation

each formulation. The percent prediction errors for Cmax and AUC were calculated as follows:

The data generated in the bioavailability study were used to develop the IVIVC. The percent of drug dissolved was determined using the aforementioned dissolution testing method and the fraction of drug absorbed was determined using the method of Wagner–Nelson [16]. Linear regression analysis was used to examine the relationship between percent of drug dissolved and percent of drug absorbed. The percent of drug un-absorbed was calculated from the percent absorbed. The slope of the best-fit line for the semi-log treatment of this data was taken as the first order rate constant for absorption. The dissolution rate constants were determined from % released vs. the square root of time. Linear regression analysis was applied to the in-vitro–in-vivo correlation plots and coefficient of determination (r 2 ), slope and intercept values were calculated.

2.7. Internal validation of the IVIVC The internal predictability of the IVIVC was examined by using the mean in-vitro dissolution data and mean in-vivo pharmacokinetics of the extended matrix tablets. Briefly, the correlation of the mean in-vitro dissolution rate constants was correlated to the mean absorption rate constants for the two extended release dosage forms (see Fig. 4). These two data points, along with the zero–zero intercept were used to calculate the expected absorption rate constants (i.e., where absorption rate constants5 [slope]3dissolution rate constant1[intercept]). The prediction of the plasma buspirone hydrochloride concentration was accomplished using the following curve fitting equation: ka 2k t 2k t y 5 Const. 3 (Dose) 3 ]]se e 2 e a d ka 2 ke

(5)

where, y5predicted plasma concentration (ng / ml); Const.5the constant representing F /Vd , where F is the fraction absorbed, and Vd is the volume of distribution; k a : absorption rate constant; k e : overall elimination rate constant. The de-convolution was accomplished on a spread-sheet in Excel. To further assess the predictability and the validity of the correlations, we determined the observed and IVIVC model-predicted Cmax and AUC values for

F

G

3 100

(6)

F

G 3 100

(7)

Cmaxsobsd 2 Cmax (pred) % PE C max 5 ]]]]]]] Cmaxsobsd

AUCsobsd 2 AUC(pred) % PEAUC 5 ]]]]]]] AUCsobsd

where Cmax (obs) and Cmax (pred) are the observed and IVIVC model-predicted maximum plasma concentrations, respectively; and AUC(obs) and AUC(pred) are the observed and IVIVC model-predicted AUC for the plasma concentration profiles, respectively.

2.8. External validation of the IVIVC The external validation was accomplished by reformulating the extended release dosage form to a release rate between the ‘‘Fast’’ and the ‘‘Slow’’ rates, selected to provide a Cmax of the re-formulated product equivalent to the Cmax obtained from the IR tablets, and to re-test the re-formulated product against the IR tablets in another bioequivalence test in human subjects.

3. Results and discussion

3.1. In vitro studies Dissolution plots of the cumulative percent buspirone hydrochloride release from the slow and fast formulations are presented in Fig. 1A and B. The effect of HPMC type on drug release was evaluated with two different dissolution solvents, 0.1 M HCl and pH 6.8 phosphate buffer at rotation speeds of 50 and 100 rpm. It is observed that the high-molecularweight (high viscosity) polymer has a slower dissolution rate than the dosage form with the lowermolecular-weight (lower viscosity) polymer in both pH media. The release of buspirone hydrochloride from the slow and fast formulations were found to be almost indistinguishable from each other when the dissolution is measured in 0.1 M HCl based on high solubility of drug in acidic media, but f2 values were 42.2 and 47.7 at 50 and 100 rpm, respectively. However, at pH 6.8, the differences between the

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

151

Fig. 1. (A) Cumulative buspirone hydrochloride release versus time profile for ‘‘Slow’’ and ‘‘Fast’’ extended release tablets using (a) pH 6.8, 50 rpm, (b) 0.1 M HCl, 50 rpm, (c) pH 6.8, 100 rpm, (d) 0.1 M HCl, 100 rpm. (B) Cumulative buspirone hydrochloride release versus square root of time profile for ‘‘Slow’’ and ‘‘Fast’’ extended release tablets using (a) pH 6.8, 50 rpm, (b) 0.1 M HCl, 50 rpm, (c) pH 6.8, 100 rpm, (d) 0.1 M HCl, 100 rpm.

formulations were more evident. Weakly basic buspirone hydrochloride has a lower solubility in pH 6.8 phosphate buffer than in 0.1 M HCl. The calculated similarity factors ( f2 ) confirmed the conclusion (Table 1). Eddington et al. [10] reported that it is

imperative to utilize a dissolution methodology that discriminates between formulations and mimics the in-vivo release profile in the process of developing an IVIVC. Accordingly, pH 6.8 phosphate buffer at both 50 and 100 rpm were found to be the more

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

152

Fig. 1. (B) (continued) Table 1 f2 similarity factors for buspirone hydrochloride extended release dosage forms in various dissolution conditions pH

Conditions

Formulation

f2

0.1 M HCl 0.1 M HCl

50 rpm 100 rpm

Fast versus slow Fast versus slow

42.2 47.7

pH 6.8 pH 6.8

50 rpm 100 rpm

Fast versus slow Fast versus slow

24.3 29.7

discriminating dissolution media in our study and 50 rpm in phosphate buffer was then used in the IVIVC model development.

3.2. In vivo studies The mean pharmacokinetic parameters are summarized in Table 2, and the mean plasma buspirone hydrochloride concentration vs. time profiles after

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

153

Table 2 Mean pharmacokinetic parameters of buspirone hydrochloride Formulation

Cmax (ng / l)

T max (h)

AUC (0–`) (ng h / ml)

Immediate release Slow formulation Fast formulation

1.6562.1 1.3761.37 1.7661.16

0.67 6 6

12.9615.1 22.2623.6 21.0622.9

each formulation and the immediate release formulation are presented in Fig. 2. There are discernible differences in the plasma level concentrations between the three dosage forms (‘‘Slow’’, ‘‘Fast’’ and IR tablets). It was also found that the rank order of release observed in the dissolution testing was also apparent in the plasma buspirone hydrochloride concentration profiles with a mean Cmax of 1.37 and 1.76 ng / l for the slow and fast releasing formulations. However, the same rank order was not observed in the AUC ` (Table 2). There is not a significant or noticeable difference in the AUC from the slowest releasing dosage form compared to the fast releasing dosage form, showing that the extent of absorption of buspirone was the same despite the differences in release rates between the two dosage forms. The AUC of buspirone was much higher from the extended release forms than from the IR tablets, due to changes of buspirone metabolism, probably due to the change in location of buspirone absorption in the GI tract.

3.3. IVIVC correlation development A level A in-vitro–in-vivo correlation was investigated using the percent dissolved vs. the percent absorbed data for both the slow and fast formulations, using both 0.1 M HCl and pH 6.8 phosphate buffer dissolution media at both 50 and 100 rpm. A good linear regression relationship was observed between the dissolution testing using pH 6.8 phosphate buffer at 50 rpm and the percents absorbed for the combined data of the two dosage forms (Fig. 3A; correlation coefficient50.9492). Another good linear regression relationship was observed between the dissolution testing using 0.1 M HCl as the dissolution media at 50 rpm, and the percents absorbed for the combined data of the two dosage forms (Fig. 3B; correlation coefficient50.9483). It is also observed that the in-vivo absorption rate constant (k a ) correlates well with the pH 6.8 phosphate buffer in-vitro dissolution rate constant (k diss ), exhibiting a correlation coefficient of 0.9353 (Fig. 4, Table 3). This was a better correlation than was obtained using the dissolution rates in 0.1 M HCl, and therefore, pH 6.8 phosphate buffer was selected as the dissolution media of choice.

3.4. Internal validation

Fig. 2. Mean busprione hydrochloride plasma concentration versus time profile after IR, slow and fast release formulations.

The dosage forms are designed to be the ‘‘slow step’’ in the absorption process; therefore it should follow that the rise in the percent of the amount absorbed should mimic the release of drug from the dosage form. Since the release of buspirone from the dosage form follows the receding boundary mechanism of release (see Fig. 1B), it should follow that the increase in the cumulative AUC should mimic this release, and be linear with respect to the square root of time, and it does (Fig. 5). The slopes of the in-vivo data were determined to be about 37.1 and 31.0 for the ‘‘Fast’’ and ‘‘Slow’’ dosage forms,

154

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

Fig. 4. Plot of in vitro dissolution rate (k diss ) versus in vivo absorption rate (k a ) constants (The zero–zero point is theoretical). Table 3 Dissolution rate (k diss ) and absorption rate (k a ) constants for the slow and fast formulations Formulation

k diss

ka

Slow Fast

22.0 37.9

0.134 0.377

centrations when the dissolution rates are used to calculate an absorption rate constant. The validity of correlations was also assessed by determining how well the IVIVC models could

Fig. 3. IVIVC model linear regression plots of % absorbed vs. % dissolved for the slow and fast tablets: (A) pH 6.8, 50 rpm and (B) 0.1 M HCl, 50 rpm.

corresponding to the in-vitro release rates of 37.9 and 21.9 for the corresponding dosage forms (Fig. 1B). These are good indications that the mechanism and rates of release in-vivo mimic mechanism and rates of release in-vitro. Fig. 6 illustrates the observed and predicted buspirone hydrochloride plasma concentrations for the ‘‘Fast’’ (Fig. 6A), and ‘‘Slow’’ (Fig. 6B) formulations. It is observed that a good correlation is found between the actual and the predicted plasma con-

Fig. 5. Plot of percent cumulative AUC versus square root of time for buspirone hydrochloride extended release formulations.

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

155

Table 4 Prediction errors (%) associated with Cmax and AUC

Fig. 6. Observed and predicted buspirone hydrochloride plasma concentration for the (A) ‘‘Fast’’ and (B) ‘‘Slow’’ releasing formulation using the IVIVC model.

predict the rate and extent of buspirone hydrochloride absorption as characterized by Cmax and AUC. Table 4 presents the percent errors estimated for the difference between the observed and predicted Cmax

Formulation

Cmax

AUC

Slow Fast

2.34 22.67

14.6 17.6

Average

20.16

16.1

and AUC values for the IVIVC model. The Cmax prediction errors for the slow (2.34%) and fast (22.67%) formulations were both found to be very close to the observed mean values. It is found that the observed mean of AUC values was 14.6 and 17.6% for ‘‘Slow’’ and ‘‘Fast’’ formulations, respectively. The FDA guidance [4] on IVIVC states that an average absolute percent prediction error of #10% for Cmax and AUC establishes the predictability of the IVIVC. In addition, the percent prediction error for each formulation should not exceed 15%. In the present study, the predicted AUC value of fast formulation is 17.6% greater than the observed AUC values. The failure of the IVIVC to predict the extent of buspirone hydrochloride absorption may be explained by first pass metabolism. Since the buspirone is released from the IR tablets at a much faster rate than from either extended release dosage form, there is much more buspirone within the dosage form further down the GI tract than there is from the IR tablets. If there is gut wall metabolism high up in the GI tract, more buspirone will be metabolized from the IR tablets than from the extended release dosage forms. This is confirmed by examining the extent of formation of the major metabolite (1-PP) of buspirone. The AUC of the 1-PP was higher from the IR tablets than from the extended release dosage forms [14]. Even though there was a substantial difference in release rate between the ‘‘Fast’’ and ‘‘Slow’’ releasing dosage forms, there was little or no difference in the AUC of either buspirone or 1-PP between the two dosage forms.

3.5. External validation Buspirone ER tablets were reformulated with a release rate between the ‘‘Fast’’ (12 h) and ‘‘Slow’’

156

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157

Acknowledgements This research was supported by American Pharmaceutical International, Inc. (API), Cincinnati, OH, USA.

References

Fig. 7. Plot of plasma concentration versus time of predicted formulation with the actual ‘‘Fast’’ and ‘‘Slow’’ formulations.

(24 h) formulations, and the IVIVC for the fast and slow formulation was used to predict the plasma concentrations of the new formulation. Fig. 7 shows the plasma concentration–time profile predicted for the new formulation with the actual ‘‘Fast’’ and ‘‘Slow’’ formulations. The predicted curve of the new formulation is similar to the faster and the slower releasing products. The actual maximum of the average plasma concentration at steady state was determined to be 1.60 ng / ml, and very close to the maximum of the average plasma concentration of the IR tablet (1.61 ng / ml) [17].

Conclusions The significant correlations between the in-vitro and the in-vivo parameters reported here indicate that the IVIVC was excellent for predicting Cmax , however, it was not acceptable for predicting AUC because of unusual AUC data (AUC of extended release. .AUC of immediate release due to first pass metabolism). It is also observed that the prediction errors of AUC (14.6 and 17.6% for ‘‘Slow’’ and ‘‘Fast’’ formulations, respectively) are in excellent agreement between the two dosage forms.

[1] Z. Hussein, M. Friedman, Release and absorption characteristics of novel theophylline sustained release formulations: in vitro–in vivo correlation, Pharm. Res. 7 (11) (1990) 1167–1171. [2] N. Sirisuth, L.L. Augsburger, N.D. Eddington, Development and validation of a non-linear IVIVC model for a diltiazem extended release formulation, Biopharm. Drug Disp. 23 (2002) 1–8. [3] R.J. Rackley, Examples of in vitro–in vivo relationships with a diverse range of quality, in: D. Young, J.G. Devane, J. Butler (Eds.), In Vitro–In Vivo Correlations, Advances in Experimental Medicine and Biology, Vol. 423, Plenum Press, New York, 1997, pp. 53–65. [4] US Department of Health, Food and Drug Administration, Guidance for the Industry: Extended Release Solid Oral Dosage Forms: Development, Evaluation and Application of In Vitro / In Vivo Correlations, US Department of Health, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), September 1997. [5] US Department of Health, Food and Drug Administration, Modified Release Solid Oral Dosage Form Guidance: ScaleUp and Postapproval Changes: Chemistry, Manufacturing and Controls, In Vitro Dissolution Testing and In Vivo Bioequivalance Documentation, US Department of Health, Food and Drug Administration, Center for Drug Evaluation and Research (CDER), September 1997. [6] G. Balan, P. Timmins, D.S. Greene, P.H. Marathe, In vitro–in vivo correlation (IVIVC) models for metformin after administration of modified-release (MR) oral dosage forms to healthy human volunteers, J. Pharm. Sci. 90 (8) (2001) 1176–1185. [7] J.T. Dalton, A.B. Straughn, D.A. Dickason, G.P. Grandolfi, Predictive ability of level A in vitro–in vivo correlation for ringcap controlled-release acetaminophen tablets, Pharm. Res. 18 (12) (2001) 1729–1734. [8] H. Mahayni, G.S. Rekhi, R.S. Uppoor, P. Marroum, A.S. Hussain, L.L. Augsburger, N.D. Eddington, Evaluation of ‘‘External’’ predictability of an in vitro–in vivo correlation for an extended-release formulation containing metoprolol tartrate, J. Pharm. Sci. 89 (10) (2000) 1354–1361. [9] P. Veng-Pedersen, J.V.S. Gobburu, M.C. Meyer, A.B. Straughn, Carbamazepine level A in vitro–in vivo correlation (IVIVC): a scaled convolution based predictive approach, Biopharm. Drug Disp. 21 (2000) 1–6. [10] N.D. Eddington, P. Marroum, R. Uppoor, A. Hussain, L. Augsburger, Development and internal validation of an in

S. Takka et al. / Journal of Controlled Release 88 (2003) 147–157 vitro–in vivo correlation for a hydrophilic metoprolol tartrate extended release tablet formulation, Pharm. Res. 15 (3) (1998) 466–473. [11] M.N. Roshdy, R.L. Schnaare, E.T. Sugita, S. Zietz, J.B. Schwartz, An adjusted pharmacokinetic equation for prediction drug levels in vivo based on in vitro square root of time release kinetics, Pharm. Dev. Techol. 7 (2) (2002) 203–213. [12] R.F. Mayol, B.S. Adamson, R.E. Gammans, Pharmacokinetics and disposition of 14C-buspirone HCl after intravenous and oral dosing in man, Clin. Pharmacol. Ther. 37 (1985) 210–215. [13] H. Goldberg, Buspirone hydrochloride: a unique new anxiolytic agent. Pharmacokinetics, clinical pharmacology, abuse potential and clinical efficacy, Pharmacotherapy 4 (6) (1984) 315–324.

157

[14] A. Sakr, M. Andheria, Pharmacokinetics of buspirone extended release tablets: a single dose study, J. Clin. Pharmacol. 41 (2001) 1–7. [15] US Department of Health, Food and Drug Administration, Dissolution Testing of Immediate Release Solid Oral Dosage Forms, August 1997. [16] J.G. Wagner, Biopharmaceutics and Relevant Pharmacokinetics, Drug Intelligance Publishers, Illinois, 1971. [17] A. Sakr, M. Andheria, A comparative multidose pharmacokinetic study of buspirone extended release tablets with a reference immediate release product, J. Clin. Pharmacol. 41 (2001) 886–894.