Modeling the heterogeneous intestinal absorption of propiverine extended-release

European Journal of Pharmaceutical Sciences 76 (2015) 133–137

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences journal homepage: www.elsevier.com/locate/ejps

Modeling the heterogeneous intestinal absorption of propiverine extended-release Michael Weiss a,b,⇑, Pakawadee Sermsappasuk b, Werner Siegmund c a

Department of Pharmacology, Martin Luther University Halle-Wittenberg, Germany Faculty of Pharmaceutical Sciences, Naresuan University, Phitsanulok, Thailand c Division of Clinical Pharmacology, Institute of Pharmacology, Ernst Moritz Arndt University, Greifswald, Germany b

a r t i c l e

i n f o

Article history: Received 12 January 2015 Received in revised form 8 April 2015 Accepted 10 May 2015 Available online 11 May 2015 Keywords: Heterogenous absorption Propiverine Bioavailability Dissolution time

a b s t r a c t Propiverine is a widely used antimuscarinic drug with bioavailability that is limited by intestinal first-pass extraction. To study the apparent heterogeneity in intestinal first-pass extraction, we performed a population analysis of oral concentration–time data measured after administration of an extended-release formulation of propiverine in ten healthy subjects. Using an inverse Gaussian function as input model, the assumption that the systemically available fraction increases as a sigmoidal function of time considerably improved the fit. The step-like increase in this fraction at time t = 3.7 h predicted by the model suggests that propiverine is predominantly absorbed in colon. A nearly perfect correlation was found between the estimates of bioavailability and mean dissolution time. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction Propiverine is a widely used muscarine receptor blocking drug for the treatment of patients suffering from hyperactive bladder syndrome. From a pharmacokinetic point of view, it is an interesting and unique property of propiverine that the extended-release formulation (ER) has a higher bioavailability than immediate-release tablets (May et al., 2008). The authors suggested a lower first-pass metabolism (CYP3A4) and efflux transport (P-glycoprotein and MRP2) of propiverine in the distal small intestine and colon. This was supported by a positive correlation between bioavailability and mean absorption time. However, no modeling of the absorption process was carried out. Therefore, the objective of the present study was first to reanalyze the ER data using a parametric absorption model. The selection of the inverse Gaussian function as input model (Weiss, 1996; Wang et al., 2008) was was also motivated by the fact that in a pilot study this model well described the in vitro dissolution data of propiverine ER. Second, we tried to improve the model by incorporating a systemically available fraction (absorption and intestinal first-pass availability) that increases with intestinal transit time. Despite its simplicity, the novel input model considerably improved the fit and the results are in line with the reported expression of relevant CYP enzymes and transporters

⇑ Corresponding author at: Department of Pharmacology, Martin Luther University Halle-Wittenberg, D-06097 Halle (Saale), Germany. E-mail address: [email protected] (M. Weiss). http://dx.doi.org/10.1016/j.ejps.2015.05.010 0928-0987/Ó 2015 Elsevier B.V. All rights reserved.

along the gastrointestinal tract (Berggren et al., 2007; Zimmermann et al., 2005). The model may be used as a first indicator of regional differences in intestinal drug uptake.

2. Material and methods 2.1. Data and study protocol The experimental protocol of the bioavailability study in healthy volunteers has been described before (May et al., 2008). Here we refer only to the data obtained after intravenous injection (IV) and administration of extended-release capsules (propiverine ER containing 45 mg propiverine hydrochloride; MictonormÒ, APOGEPHA, Dresden, Germany). In brief, in a randomized, controlled, crossover study with a washout period of at least 7 days, ten healthy German white subjects (six males, four females; age 22–26 years) received 15 mg propiverine hydrochloride by IV bolus injection (5 min) and 10, 15, 30 and 45 mg propiverine ER swallowed using 200 ml table water. Blood samples were taken before the IV injections and 0.5, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 15, 18, 24, 30, 36, 48 and 72 h thereafter. For propiverine ER, the sampling times were before administration and after 1, 2, 4, 6, 8, 10, 12, 15, 18, 24, 30, 36, 48, 72 and 96 h. The serum concentrations of propiverine were measured using liquid chromatography–tandem mass spectrometry (LC–MS/MS) after solid phase extraction with a lower limit of quantification of 0.78 ng/ml and a within-day precision <10%.

134

M. Weiss et al. / European Journal of Pharmaceutical Sciences 76 (2015) 133–137

Vd

D fIG(t)

Plasma concentration

Available fraction

Dissolution rate (1/h)

CLd

Fin(t)

Time (h)

Vc CL

Time (h)

Dissolution

Time (h)

Absorption/Intestinal availability

Disposition

Fig. 1. Two-compartment disposition model with an input function consisting of an in vivo dissolution rate (inverse Gaussian density) and a fraction absorbed as function of time (IG-Fin(t) model).

release from propiverine ER is the rate-limiting step of the input process. We used the inverse Gaussian density function, fIG(t), to model the dissolution time distribution of slow release formulations in vivo and in vitro (Wang et al., 2008; Weiss, 1996)

Table 1 Population mean values and interindividual variability (% coefficient of variation in brackets) estimated from data obtained following intravenous administration. Parameter

Values (% CV)

Description

CL (l/h) Vc (l) CLd (l/h) Vp (l)

12.9 89.2 10.1 79.3

Clearance Central volume Distribution clearance Peripheral volume

(35) (31) (28) (15)

f IG ðtÞ ¼

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi MDT 2pRD2 t3

" exp 

ðt  MDTÞ2

# ð1Þ

2RD2 MDT t

MDT and RD2 denote the mean and the relative dispersion (normalized variance) of dissolution time, respectively. The input function is then given by (IG-F model)

2.2. Pharmacokinetic model

IðtÞ ¼ D F f IG ðtÞ

The plasma concentration–time curve of a drug after oral administration is the result of two independent processes, drug input into and elimination from the systemic circulation. To define the input function I(t), it was assumed that the rate of propiverine

ð2Þ

The factor F is the bioavailability of the orally administered dose D. Based on a two-compartment model for propiverine disposition kinetics, the drug amounts in the central (xc) and peripheral compartments (xp) are described by the differential equations:

3

-3

102

-2

7 6 5 4 3

Standardized residual

Observed concentration (ng/ml)

2

2

101 7 6 5 4 3 2

-1

0

1

2

100 7

3 78

10

0

2

3

4 5 6 78

10

1

2

3

4 5 6 78

Predicted concentration (ng/ml)

10

2

2

3

1

3

5

7

9

11

13

15

17

Predicted concentration (ng/ml)

Fig. 2. Plots of individual predicted vs. observed concentrations and standardized residuals vs. individual predictions for the fit of the 15 mg propiverine IV dose data in 10 subjects obtained with a two-compartment model. The solid lines presents the line of identity and the zero reference line, respectively.

135

M. Weiss et al. / European Journal of Pharmaceutical Sciences 76 (2015) 133–137 Table 2 Estimated population mean values and interindividual variability (% coefficient of variation in brackets) of model parameters. Parameter

Values (% CV)

AIC MDT (h) RD2 ts (h) s F0 F (%) a

Description

IG-F-model

IG-Fin(t)-model

109 13.2 (16) 1.25 (49)

1447 8.63 (33) 2.24 (16) 3.72 (1) 0.112 (18) 0.070 (94) 58.8a (25)

58.5 (26)

Akaike inform. criterion Mean dissolution time Relative dispersion of DT Inflection point of Fin(t) Steepness of Fin(t) at ts Baseline of Fin(t) Bioavailability

Derived parameter: Eq. (7).

dxc =dt ¼ IðtÞ  ðCL þ CLd Þxc =V c þ CLd xp =V p

ð3Þ

dxp =dt ¼ CLd xc =V c  CLd xp =V p

ð4Þ

CL, CLd, Vc and Vp denote clearance, distribution clearance, volume of the central and peripheral compartment, respectively. First, the IV data were fitted instantaneous input in Eqs. (2) and (3). Second, because of the system linearity, i.e., dose-proportional increase in AUC and Cmax (May et al., 2008), the data of all four doses of propiverine ER were fitted simultaneously using Eq. (2) as input function. Finally, it was tested whether the incorporation of fraction Fin(t) that increases with time would improve the fit, i.e., F in Eq. (2) was replaced by Fin(t) (IG-Fin(t) model)

IðtÞ ¼ D F in ðtÞ f IG ðtÞ

ð5Þ

where Fin(t) can be attributed to absorption and intestinal availability (fraction that escapes first pass gut wall metabolism). That means, assuming that the release rate from propiverine ER during intestinal transit can be described by the IG model [fIG(t)], the systemically available fraction increases with intestinal transit time. A Boltzmann function was chosen as an empirical model for the time course of the bioavailability across the gut wall

1  F0 F in ðtÞ ¼ F 0 þ 1 þ expðts stÞ

ð6Þ

Where F0 denotes the baseline (lower asymptote), ts is the inflection point of the sigmoid function and s determines the steepness (slope) of the curve. A graph of the model structure is shown in Fig. 1. The bioavailability was calculated as

F¼

Z

1

F in ðtÞf IG ðtÞdt

ð7Þ

0

In order to avoid ambiguities it should be noted that the fraction Fin(t) is independent of the dose normalized rate of propiverine release, fIG(t). 2.3. Data analysis A population approach (nonlinear mixed-effects modeling) with maximum likelihood estimation via the EM algorithm implemented in the software ADAPT 5 (D’Argenio et al., 2009) was used for data analysis. The MLEM program provides estimates of the population mean and inter-subject variability as well as of the individual subject parameters (conditional means). We assumed log-normally distributed model parameters and that the measurement error has a standard deviation that is a linear function of the measured quantity:

VARi ¼ ½r0 þ r1 Cðt i Þ2

ð8Þ

After fitting the IV data, the individual parameter estimates (CL, CLd, Vc and Vp) were held fixed in fitting the oral data (ER tablets). Thereby, the four data sets (obtained with the 10, 15, 30 and 45 mg doses) for each subject were analyzed simultaneously to estimate the parameters MDT, RD2 and F of the IG-F model (Eqs. (1) and (2)) or MDT, RD2, F0, ts and s of the IG-Fin(t) model (Eqs. (5) and (6)), respectively. MAPLE 8 (Waterloo Maple, Waterloo, ON, Canada) was used to numerically evaluate the integral in Eq. (7). The model fit to the data was assessed by the goodness-of-fit plot, standardized residuals versus predicted concentration, and individual fits. Model discrimination was done by Akaike’s information criterion. A population simulation without noise of the fIG(t) (Eq. (1)) and Fin(t) (Eq. (6)) was performed with the simulation module of ADAPT 5 using the estimated parameters. 3. Results Table 1 shows the disposition parameters obtained with the two-compartment model. Use of a three compartment model did not improve the fit (increase in AIC from 1047 to 1054). The goodness of fit and standardized residual plots are shown in Fig. 2. The additive and proportional components of the residual error model where r0 = 0.58 and r1 = 0.11, respectively.

3

6 5 4 3

Standardized residual

Observed concentration (ng/ml)

10 2

2

10 1 6 5 4 3 2

1

-1

10 0 6

-3 6 78

10 0

2

3

4 5 6 78

10 1

2

3

Predicted concentration (ng/ml)

4 5 6 78

10 2

0

20

40

60

80

100

Predicted concentration (ng/ml)

Fig. 3. Plots of individual predicted vs. observed concentrations and standardized residuals vs. individual predictions for the simultaneous fit of the 10, 15, 30 and 45 mg propiverine ER dose data in 10 subjects obtained with the IG-Fin(t) model. The solid lines presents the line of identity and the zero reference line, respectively.

136

M. Weiss et al. / European Journal of Pharmaceutical Sciences 76 (2015) 133–137

45 mg doses) are shown in Fig. 3. It is obvious that the model underestimated some of the highest observed concentrations. Using the model, an excellent correlation between bioavailability and the mean dissolution time MDT (Fig. 4) can be demonstrated. Plots of individual model fits to the data in two representative subjects are depicted in Fig. 5. Graphs of the model predictions of fIG(t) and Fin(t) determining the underlying input rate (Eq. (5)) are shown in Fig. 6 for the subject where bioavailability was closest to the median value. A characteristic result is the step-like increase in Fin(t) at time ts = 3.72 ± 0.05 h.

0.8

Bioavailability

0.7

0.6

0.5

0.4

4. Discussion In reanalyzing the data, we overcame some limitations of the originally applied numerical integration (non-compartmental) approach (May et al., 2008), where results highly depend on the sampling schedule and data noise; furthermore, concentration–time profiles cannot be predicted, and estimates of between-subject variability are not available. The analysis started with the IG-F input model that has proved useful in describing the dissolution time distribution of slow release formulations in vivo and in vitro (Wang et al., 2008; Lotsch et al., 1999; Weiss, 1996; Brvar et al., 2014). The extended model IG-Fin(t) accounted for an increase in systemically available fraction with intestinal transit time. Although the analyses based on the IG-F model and the IG-Fin(t) model led to very similar estimates of bioavailability (F) and a high correlation between F and MDT (Fig. 4), the IG-Fin(t) model considerably improved the fit as demonstrated by the large reduction in AIC (DAIC = 1338). The modeling implies that the systemically available fraction increases at time

0.3

6

8

10

12

14

Mean dissoluon me (h) Fig. 4. Correlation between individual estimates of bioavailability and mean dissolution time (R2 = 0.977, P < 0.0001).

The results of the population analysis using the IG-F and IG-Fin(t) model are given in Table 2 (population means and interindividual variability). We present details of the analysis for the novel model where the fraction absorbed increased with time (IG-Fin(t)) since the reduction in AIC from 109 to 1447 indicates that it described the data much better than the conventional IG-F model. The additive component of the residual error model (r0) tended toward zero while the proportional component ranged from 0.29 to 0.42 for the four doses. The goodness of fit and standardized residual plots for all propiverine ER data (10, 15, 30 and

100 120 80

100

80

60

60 40

Propiverine concentration (ng/ml)

40 20 20

0

0 0

20

40

60

80

100

0

20

40

60

80

100

0

20

40

60

80

100

60 100

80 40 60

40

20

20

0

0 0

20

40

60

80

100

Time (h) Fig. 5. Fits obtained with the IG-Fin(t) model for the four propiverine ER doses of 10 (s), 15 (h), 30 (j) and 45 (d) mg in four representative subjects.

M. Weiss et al. / European Journal of Pharmaceutical Sciences 76 (2015) 133–137

A

Normalized dissolution rate (1/h)

0.25

0.20

0.15

0.10

0.05

0.00 0

Available fraction, Fin(t)

1.0

5

10

15

20

137

first-order absorption process (Wang et al., 2008) did not improve the fit, but in fact increased the AIC value (DAIC = 11). The present analysis also shows the positive correlation of bioavailability with MDT (Fig. 4) much more pronounced (R2 = 0.977 compared to R2 = 0.262 in (May et al., 2008)). Note that, for the sake of simplicity, the contribution of hepatic extraction has not been explicitly defined in the IG-Fin(t) model. However, this does not change the overall picture because the hepatic extraction of propiverine is only about 10%. Despite the improvement in fit obtained with the IG-Fin(t) model, this simple empirical model is no substitute for more elaborate, physiologically based models of drug absorption (Bergstrand et al., 2009; Hénin et al., 2012; Kostewicz et al., 2013). As any model it does not provide a proof of something; rather it offers a possible explanation for the increase in bioavailability with the increase in mean dissolution time. 5. Conclusion

B

In summary, based on a parametric dissolution model and a systemically available fraction that changes with time, we developed a parsimonious model for ER formulations that accounts for heterogeneity in intestinal drug uptake. It remains to be seen whether similar results will be obtained with this model for other drugs with large differences in regional absorption.

0.8

0.6

0.4

References 0.2

0.0 0

2

4

6

Time (h) Fig. 6. Simulated time courses (mean ± SD) of in vivo dissolution rate, fIG(t), (A) and systemically available fraction, Fin(t), (B). The population simulation was based on the means and variances of model parameters (Table 1).

t = 3.72 h after intake (Fig. 6B). There is a striking coincidence with the mean transit time for tablets through the gastrointestinal tract from the stomach to the colon of 3.75 h (Bergstrand et al., 2009) and the orocecal transit time of 3.83 h (Corá et al., 2006). This suggests that propiverine ER is systemically absorbed in the colon, whereas the Fin across the gut wall seems to be very low in entire small intestine. This is in principle in accordance with the finding of low expression of CYP3A4, P-glycoprotein and MRP2 in the distal small intestine and colon (Berggren et al., 2007; Zimmermann et al., 2005). Although the Fin(t) model (Eq. (6)) allowed for a more continuous increase, a nearly step-like change in Fin at time ts was observed (Fig. 6B). That the bioavailability values estimated with both models were quite similar can be explained by the fact that F, by definition, is an model-independent parameter that is robust to small changes in the C(t) profile (with minor influence on the area under the curve). It is important to note that the mean dissolution time in vivo of 8.63 h obtained from the IG-Fin(t) model is in better agreement with mean dissolution time in vitro of 6.96 h (data on file, APOGEPHA, Dresden, Germany), than that of the IG-F model (13.2 h). This also indicates that the release of propiverine from ER tablet is the rate limit step of the input process, and confirms the underlying assumption that fIG(t) in Eq. (5) represents the in vivo dissolution rate. It should be mentioned that an extension of the IG-F model by adding a

Berggren, S., Gall, C., Wollnitz, N., Ekelund, M., Karlbom, U., Hoogstraate, J., Schrenk, D., Lennernäs, H., 2007. Gene and protein expression of P-glycoprotein, MRP1, MRP2, and CYP3A4 in the small and large human intestine. Mol. Pharm. 4, 252– 257. Bergstrand, M., Söderlind, E., Weitschies, W., Karlsson, M.O., 2009. Mechanistic modeling of a magnetic marker monitoring study linking gastrointestinal tablet transit, in vivo drug release, and pharmacokinetics. Clin. Pharmacol. Ther. 86, 77–83. Brvar, N., Mateovic-Rojnik, T., Grabnar, I., 2014. Population pharmacokinetic modelling of tramadol using inverse Gaussian function for the assessment of drug absorption from prolonged and immediate release formulations. Int. J. Pharm. 473, 170–178. Corá, L.A., Romeiro, F.G., Américo, M.F., Oliveira, R.B., Baffa, O., Stelzer, M., de Arruda Miranda, J.R., 2006. Gastrointestinal transit and disintegration of enteric coated magnetic tablets assessed by ac biosusceptometry. Eur. J. Pharm. Sci. 27, 1–8. D’Argenio, D., Schumitzky, A., Wang, X., 2009. ADAPT 5 User’s Guide: Pharmacokinetic/Pharmacodynamic Systems Analysis Software. Biomedical Simulations Resource, Los Angeles, CA. Hénin, E., Bergstrand, M., Standing, J.F., Karlsson, M.O., 2012. A mechanism-based approach for absorption modeling: the Gastro-Intestinal Transit Time (GITT) model. AAPS J. 14, 155–163. Kostewicz, E.S., Aarons, L., Bergstrand, M., Bolger, M.B., Galetin, A., Hatley, O., Jamei, M., Lloyd, R., Pepin, X., Rostami-Hodjegan, A., 2013. PBPK models for the prediction of in vivo performance of oral dosage forms. Eur. J. Pharm. Sci. 57, 300–321. Lotsch, J., Weiss, M., Ahne, G., Kobal, G., Geisslinger, G., 1999. Pharmacokinetic modeling of M6G formation after oral administration of morphine in healthy volunteers. Anesthesiology 90, 1026–1038. May, K., Giessmann, T., Wegner, D., Oertel, R., Modess, C., Oswald, S., Braeter, M., Siegmund, W., 2008. Oral absorption of propiverine solution and of the immediate and extended release dosage forms: influence of regioselective intestinal elimination. Eur. J. Clin. Pharmacol. 64, 1085–1092. Wang, J., Weiss, M., D’Argenio, D.Z., 2008. A note on population analysis of dissolution-absorption models using the inverse Gaussian function. J. Clin. Pharmacol. 48, 719–725. Weiss, M., 1996. A novel extravascular input function for the assessment of drug absorption in bioavailability studies. Pharm. Res. 13, 1547–1553. Zimmermann, C., Gutmann, H., Hruz, P., Gutzwiller, J.-P., Beglinger, C., Drewe, J., 2005. Mapping of multidrug resistance gene 1 and multidrug resistanceassociated protein isoform 1 to 5 mRNA expression along the human intestinal tract. Drug Met. Disp. 33, 219–224.