quantitative nmr imaging study of the mechanism of drug release from swelling...

Journal of Controlled Release 68 (2000) 313–333www.elsevier.com/ locate / jconrel

Quantitative NMR imaging study of the mechanism of drugrelease from swelling hydroxypropylmethylcellulose tablets

*C.A. Fyfe , A.I. Blazek-WelshDepartment of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver, BC, V6T Canada

Received 23 September 1999; accepted 17 March 2000

Abstract

NMR imaging has been used to study the release behavior of two model drugs, triflupromazine–HCl and 5-fluorouracil,from swelling hydroxypropylmethylcellulose (HPMC) tablets. Preliminary experiments were performed on equilibriummixtures of drug, polymer and water to determine how properties such as NMR relaxation parameters and self-diffusionwere affected by the drug and polymer concentrations. The tablet swelling was restricted to one dimension and distributions

1 19of the water and model drugs were obtained by H and F imaging, respectively. The HPMC distribution at each time in theswelling process was determined indirectly from its effect on the relaxation parameters of the water and the drugs. In theone-dimensional swelling tablet, distributions of drug and polymer were compared to determine what factors influenced therelease of drug from the swelling tablet. The distributions for triflupromazine–HCl and HPMC paralleled each other and thedrug was only released at the eroding edge of the tablet where the HPMC concentration dropped below 10%. In contrast,5-fluorouracil was released much more rapidly from the tablet and appeared to escape by diffusion from regions as high as30% HPMC. An empirical measure of the rate of tablet edge movement can be obtained from plots of the edge position as afunction of the square root of time. For HPMC, the rate of tablet expansion was determined in this way to be

26 2 21(2.460.8)310 cm s . The self-diffusion of triflupromazine–HCl in equilibrated mixtures of similar composition to the26 2 21eroding tablet edge is |3310 cm s while the self-diffusion coefficient of 5-fluorouracil remained higher than this value

until the HPMC concentration reached about 30%. This comparison of ‘diffusion’ properties may be useful in predicting themechanism of drug release from other swelling hydrophilic matrix systems. 2000 Elsevier Science B.V. All rightsreserved.

Keywords: NMR imaging; Drug release; Hydrophilic matrix tablets; Hydroxypropylmethylcellulose (HPMC)

1. Introduction drug and thus provides a mechanism for controlleddrug release [1].

Many pharmaceutical tablets are based on hydro- Drug release from swelling-controlled deliveryphilic polymers such as methylcellulose and hy- systems has been studied primarily by the USPdroxypropylmethylcellulose (HPMC). When exposed dissolution method where the tablet is suspended in ato water, these polymers form a gel layer around the media bath, or flow-cell, and the total amount oftablet that limits the dissolution and diffusion of the drug released into solution is measured as a function

of time [2]. The relationship between the fraction of*Corresponding author. drug released and time is used to determine whether

0168-3659/00/$ – see front matter 2000 Elsevier Science B.V. All rights reserved.PI I : S0168-3659( 00 )00245-5

314 C.A. Fyfe, A.I. Blazek-Welsh / Journal of Controlled Release 68 (2000) 313 –333

the drug release process follows Fickian or non- clude quantitative imaging studies of the release ofFickian kinetics as determined by the water penetra- two quite different drugs from HPMC tablets totion process. Dissolution tests, however, cannot investigate the details of the mechanism of theirprobe the drug present within the swollen tablet and release during the swelling of the gel. The resultsdo not provide information on the relative impor- obtained from this work are presented in the presenttance of the two mechanisms of drug release from manuscript.swelling-controlled devices, namely diffusionthrough the swollen polymer or, in cases of slowdiffusion, the release of drug at the edges of the 2. Experimentaltablet through erosion.

The swelling processes of hydrogels have been 2.1. Preparation of tabletsmonitored in the past by the measurement of dimen-sional changes and more recently by optical [3–5] HPMC (Dow Methocel K4M) was obtained as aand NMR imaging methods [6–9]. It is, however, gift from UpJohn-Pharmacia. Triflupromazine–HClnot straightforward to obtain quantitative information and 5-fluorouracil were obtained from Aldrich. Tab-from standard NMR imaging protocols. In previous lets were prepared by direct compression (6263work, we introduced a simple geometric arrangement MPa compressional force) of HPMC powder andfor such imaging experiments from which complete drug using a Beta Manesty rotary tablet pressand quantitatively reliable information on the diffu- modified to detect compressional forces. Thesion and swelling processes can be obtained in HPMC–drug mixtures contained 5.30 and 5.25% byreasonable time periods [10]. Although it is not weight of triflupromazine–HCl and 5-fluorouracil,possible to image the polymer directly, its con- respectively, and the tablet parameters for the twocentration can be obtained indirectly from its con- products given in Table 1 indicate that they arecentration-dependent influence on the NMR relaxa- essentially identical. By chance, the tablets containedtion times of the water in its immediate proximity. In the same concentrations of fluorine because the ratiothis way, it is possible to obtain water penetration of fluorines to molecular weight is similar for bothand HPMC concentration profiles as a function of the triflupromazine–HCl and 5-fluorouracil, the formerswelling time. In subsequent work [11], we identified having three fluorines and the latter only one.the presence of air bubbles originating from gastrapped in the tablet matrix during compression as a 2.2. NMR and NMR imaging experimentsmajor potential source of error in all investigations of

1 19hydrogel tablet swelling since they comprise a H and F experiments were carried out using asignificant fraction of the volume occupied by the Bruker MSL 400 spectrometer operating at frequen-swollen gel. A simple protocol was described by cies of 400.13 and 376.45 MHz, respectively.which these effects could be avoided and completely Spin-lattice relaxation time (T ) measurements1

reliable and quantitative HPMC concentration pro- were carried out using a conventional 18082t2908-files could be obtained [11]. acquire sequence [12]. Other experiments were based

These investigations have been continued to in- on the spin–echo sequence shown in Fig. 1a [13].

Table 1Parameters for the HPMC tablets containing the fluorinated model drugs

a b b 19 25Drug in tablet Weight Thickness Diameter Drug F (10 )(mg60.01) (mm60.01) (mm60.01) (mg60.01) (moles60.04)

Triflupromazine–HCl 174 1.30 12.75 9.22 7.115-Fluorouracil 176 1.31 12.75 9.24 7.10

a Weight of the HPMC component, once corrected for moisture content, was 156 mg for the triflupromazine–HCl tablet and 158 mg forthe 5-fluorouracil tablet.

b As measured with digital calipers.

C.A. Fyfe, A.I. Blazek-Welsh / Journal of Controlled Release 68 (2000) 313 –333 315

frequency will change and it will not be refocusedand will not contribute to the echo. In Fig. 1b, thespacing, D, and duration, d, of the two pulses can bevaried while keeping a constant time-to-echo (T ) inE

the experiment and thus a constant amount of T2

relaxation. The echo amplitude can be varied, andthe self-diffusion constant, D, determined by chang-ing D, d, or the gradient strength, G. In the presentwork, the PGSE experiments were carried out usinga Doty Scientific probe, the z-gradient was varied upto 60 G/cm and D obtained from plots of ln (A )G

2versus G .NMR imaging experiments were carried out using

the sequence shown in Fig. 1c. Again, the rf portionof the experiment is the spin–echo sequence withecho formation. However, the complete echo is nowacquired in the presence of a field gradient ofduration 2t, the first gradient of duration t beingapplied to ensure echo formation by keeping thesequence symmetrical around the 1808 pulse. Fouriertransformation of the echo gives an image which isthe projection of the magnetization along the direc-tion of the axis of the gradient. This corresponds to aconcentration profile of an NMR active species alongthis direction. The image is a function of T becauseE

of T relaxation and T from T relaxation.Fig. 1. Pulse sequences used for measurement of (a) spin–spin 2 R 1

A Bruker microscopic imaging accessory was usedrelaxation (T ) times, (b) self-diffusion coefficients using the2

pulsed-gradient spin–echo method and (c) one-dimensional im- with the tablet and water system in a 15 mm O.D.ages. (12.8 mm I.D.) NMR tube held vertically in the rf

coil. The HPMC tablet was placed on a support atAfter the initial 908 pulse, the spins dephase during t, the end of the tube with only one face exposed to the

1mainly due to magnetic field inhomogeneities but are water. For the H images, the side was sealed with arefocused at time t after the 1808 pulse resulting in small amount of fluorinated grease to prevent leak-an echo. There is, however, dephasing due to T age of water between the tablet and the tube wall,2

19relaxation which is not refocused during the 2t time while for the F images, regular grease was used.period. From a series of measurements made at To prevent complications from the release of trappeddifferent values of t, T can be determined. A air, the tablet was first subjected to a vacuum2

modified version of this, the pulsed-gradient spin– treatment in the tube and water added under vacuumecho (PGSE) experiment shown in Fig. 1b was used as described previously [11]. In all the imagingto measure self-diffusion coefficients [14]. The rf experiments, 5 ml of water was added into the tubesportion of the experiment is identical to (a) but in to ensure a sufficient excess of water. For 15 mmaddition, two field gradient pulses are applied as O.D. tubes, this resulted in a water height of |4 cmindicated. As previously, the echo experiment will above the tablet. In general, to ensure quantitation,still refocus the magnetization of those spins whose measurements were made only up to 1.5 cm from theprecession frequencies have remained constant for tablet face.

1the duration of the experiment. However, because an One dimensional H images were obtained exactly19additional inhomogeneous magnetic field is present, as previously described [11]. For the F images, the

if a spin diffuses during the experiment its precession spectral width was increased to 100 kH to includez


more of the region outside the tablet with a res- below as a function of time throughout the tabletolution of 0.020 cm. In addition, the number of scans swellings:was increased to compensate for the lower con-centration of the fluorinated drugs compared to that

Triflupromazine–HCl 1 Spin–echo spectrum of 3-fluoro-4-1of the water imaged in the H imaging experiments. nitrotolueneBecause of the increased number of scans, the T T 52 ms, T 515 s, 80 scans, 20E E R

minrange for the variable-T series was shortened to aE2 Spin–echo spectrum of triflu-maximum value of 64 ms, acquired in the order 2, 4,

promazine–HCl16, 32, 64, 24, 8 and 3 ms, to decrease the timeT 52 ms, T 57 s, 80 scans, 9.5E Rrequired to obtain the set of images. Also, a cali- min

19bration step to quantify the F signal was added to 3 Eight variable-t one-dimensionalE

imagesthe previously described imaging procedure. FurtherT 52–64 ms, T 57 s, 80 scans, 75E Rdiscussion of the imaging experiments is given atminappropriate points in the text.

5-Fluorouracil 1 Spin–echo spectrum of 3-fluoro-4-nitrotoluene2.3. Calibration of the fluorine signalsT 52 ms, T 515 s, 64 scans, 16E R

min1In the case of the H imaging experiments, the 2 Spin–echo spectrum of 5-fluoro-signal intensities can always be calibrated with uracil

T 52 ms, T 520 s, 64 scans, 22respect to that of bulk water. However, those from E R

19 minthe F imaging experiments are in arbitrary units3 Quantitative image of 5-fluorouraciland must be calibrated with respect to that of an

T 52 ms, T 520 s, 64 scans, 22E Rexternal reference sample. 3-Fluoro-4-nitrotoluene min(Aldrich) was chosen because its fluorine signal is 4 Eight variable-t one-dimensionalE

imageswell removed from those of the drugs being investi-T 52–64 ms, T 510 s, 64 scans,E Rgated (|60 ppm) and did not contribute at all to the86 min19F images of the swollen tablet.

19The F T and T relaxation times for 3-fluoro-4-1 2

nitrotoluene in CDCl were |3 s and 350 ms, For both drugs a quantitative spin–echo spectrum of3

respectively, so a spin–echo spectrum acquired with the reference was obtained, followed immediately bya T of 15 s and a T of 2 ms is quantitatively a quantitative spin–echo spectrum of the drug andR E

reliable. A small amount of this compound, 10.7 mg, the relative intensities of the two, combined with thewas weighed into a 5-mm O.D. NMR tube and knowledge of the amount of reference present, useddissolved in CDCl . The 5-mm tube was then to determine the amount of drug. The T and T3 E R

suspended in the larger tube containing the tablet. To values of the spin–echo spectrum and the first imageensure complete detection of the signal intensity (T 52 ms) are identical so the total amount of drugE

from a particular species, the offset of the excitation detectable within the image may now be determinedpulse was adjusted for each experiment in the in a quantitatively reliable manner. The concen-sequence such that the signal from the species of tration of drug at each position in the swelling tablet

19interest was on resonance. The calibration tube was can also be determined since each point in the Fplaced more than 1 cm from the surface of the tablet images represents the signal from a volume of 0.026

3so that no correction needed to be made for the cm .volume it occupied but it was still within the coil for The procedure for the tablets with 5-fluorouracilexcitation. There were no significant differences in was adjusted slightly because the T of the drug in1

images taken with and without the tube in place. dilute solution is about 4 s. If the T in the T -R E

The calibration of the image intensities was car- variation series was chosen to avoid any T relaxa-1

ried out by repeating the sequences of experiments tion effects, i.e. was set at five times the value of T ,1


then the eight images in the series would take 3–4 h Because these NMR experiments detect only theto acquire which was deemed to be too long. An signal from mobile materials, for both drugs, thealternate procedure, where the 2 ms image was signal intensity increases as a function of time duringacquired with a T of 20 s and then the T variation the tablet swelling. Calibration plots of signal in-R E

19was performed with a shorter T , allowed for the tensity versus the F concentration from the refer-R

quantitative determination of the 5-fluorouracil con- ence from 7 to 37 h were linear (least square fits had2centrations and also a measurement of the T dis- r 5 0.997 for both drugs) and yielded linear cali-2

tribution. bration equations when a minor zero time correction

19Fig. 2. Variation of F relaxation times of the triflupromazine–HCl component in HPMC mixtures at equilibrium (a) T values and (b) T1 2

values. The concentrations of the drug are 0.5% (filled circles), 1% (open circles) and 3% (open triangles). The line in (b) is the curvecalculated from Eq. (1).


was applied. These equations were used in all the with the hydrogel matrix. Importantly, both of themdata analysis to calculate the distribution of the each also contain one or more fluorine atoms which make

19 19drug in moles F. possible the use of F NMR spectroscopy andimaging techniques. Since there are no other sourcesof fluorine in the mixtures, these experiments will

3. Results and discussion reflect only the behaviour of the drug molecules andwill be completely independent of any interference

In order to cover a range of diffusion behaviour, from the large excess of (proton bearing) water in1two drugs were chosen that were anticipated to show the system which would affect H NMR studies of

quite different characteristics in their interactions the drugs.

19Fig. 3. Variation of F relaxation times of the 5-fluorouracil component in HPMC mixtures at equilibrium (a) T values and (b) T values.1 2

The concentrations of the drug are 0.5% (filled circles), 1% (open circles) and 3% (open triangles). The line in (b) is the curve calculatedfrom Eq. (2).


Table 2The two drugs chosen were triflupromazine–HCl19Measured F T and T relaxation parameters for the triflup-1 2(1) and 5-fluorouracil (2). Triflupromazine–HCl has

romazine–HCl component of mixtures of HPMC, triflup-a molecular weight three times that of 5-fluorouracil, romazine–HCl and wateris charged and has a long sidechain. It was thus

a b cDrug HPMC T T1 2anticipated that its diffusion would be more hindered(w/w%) (w/w%) (s) (ms)

by interactions with the gel network in various ways.20.50 0 1.2 6.8310A series of different experiments were carried out 20.50 2.14 1.1 2.7310

to characterize the drug–polymer interactions and the 20.50 4.76 1.0 1.2310drug release. For clarity of presentation, these will be 0.50 9.56 0.87 93discussed separately. 0.50 13.6 0.80 63

0.50 20.2 0.75 4321.0 0 1.1 6.031021.0 2.53 1.0 2.831021.0 4.48 0.98 1.5310

1.0 15.2 0.83 641.0 19.9 0.77 491.0 24.2 0.73 36

d,e1.0 30.8 0.70 23.0 (54%) , 8.0 (46%)1.0 39.2 0.67 9.1 (63%), 2.0 (37%)1.0 46.3 0.62 5.5 (50%), 1.02 (50%)1.0 60.0 – 5.9 (21%), 0.75 (79%)

23.0 0 0.88 2.13103.1. Studies of drug–polymer systems at23.0 2.49 0.87 1.6310equilibrium23.0 5.23 0.84 1.1310

2.9 10.2 0.80 763.1.1. NMR relaxation times for water, 3.0 16.0 0.78 58triflupromazine–HCl and 5-fluorouracil in 3.0 21.4 0.76 45

d,e3.0 29.7 0.67 25.0 (54%) 9.2 (46%)equilibrium mixtures3.0 40.3 0.69 13.0 (51%), 1.77 (49%)For both drugs, the effects of the HPMC con-3.0 46.2 0.67 5.0 (58%), 1.18 (42%)

centration and the drug concentration itself on the 3.0 58.4 – 2.4 (28%), 0.74 (72%)19F T and T relaxation times were measured. This1 2 a Corrected for 5.4% moisture content.information is important in order to obtain quantita- b Measured by the inversion-recover method, error65%.ctively reliable imaging data and also because it yields Measured by the spin–echo method, error65%.dinsight into the polymer–drug interactions. The Percentage of the species with the preceding T value.2e Heterogeneous mixtures.results of these experiments are shown in Figs. 2 and

3 and collected in Tables 2 and 3 for triflu-promazine–HCl and 5-fluorouracil, respectively. present and therefore repeat times of 7 and 20 s,

In the case of triflupromazine–HCl there was a respectively, will ensure complete relaxation of all ofslight dependence of T on the concentration of the fluorine nuclei in the system, whatever the HPMC1

drug at very low HPMC concentrations but T for concentration is in their local environments.1

5-fluorouracil and the T values of both drugs were The T values are very strongly dependent on the2 2

generally independent of the drug concentration. In HPMC concentration, decreasing to low values atessentially all cases, there was a substantial depen- high HPMC contents. At HPMC concentrationsdence of the relaxation times on the HPMC con- greater than |30% two T values are observed, due2

centration. In the case of T , this dependence is to the heterogeneous nature of the mixtures at these1

important mainly for the determination of the con- high concentrations [10]. However, even at HPMCditions under which quantitatively reliable data will concentrations of 30%, T is still greater than 2 ms2

be obtained. For both drugs, the largest T values, which means that images collected with T value of1 E

1.2 and 3.6 s for triflupromazine–HCl and 5-fluoro- 2 ms will yield quantitatively reliable values of theuracil, respectively, are seen when there is no HPMC drug concentrations for all but the most concentrated


Table 3 as described previously for the determination of19Measured F T and T relaxation parameters for the 5-fluoro-1 2 HPMC concentration profiles from the effect of the

uracil component of mixtures of HPMC, 5-fluorouracil and waterpolymer on the (proton) images of the water in the

a b cDrug HPMC T T1 2 system [10]. To facilitate such measurements, non-(w/w%) (w/w%) (s) (ms) linear least squares fittings of the data in Figs. 2 and

20.50 0 3.6 3.9310 3 were carried out yielding Eqs. (1) and (2)20.54 10.6 2.3 2.0310

20.01489 T0.50 20.4 1.5 93 2,triflu[HPMC] 5 38.18e21.0 0.909 3.3 3.13102 20.2869 T2,triflu1.0 9.93 2.3 2.0310 1 87.09e (1)21.1 18.0 1.6 1.3310

1.0 30.0 1.06 25 20.1017 T2,5flu[HPMC] 5 20.72e3.0 19.9 1.43 7420.002902 T3.0 31.5 1.01 20 2,5flu1 47.92e 2 16.61 (2)

3.0 39.7 0.82 7.1d,e2.9 46.7 0.73 ¯9–30 (¯13%)

¯2.9 (¯87%) 3.1.2. Self-diffusion coefficients for water,a triflupromazine–HCl and 5-fluorouracil inCorrected for 5.4% moisture content.b equilibrium mixturesMeasured by the inversion-recover method, error65%.c Measured by the spin–echo method, error65%. The self-diffusion constants of water and drugd Percentage of the species with the preceding T value.2 give a good indication of the mobility of the differente Heterogeneous mixtures. components within the swollen polymer matrix.

These were measured by the PGSE technique de-HPMC environments. The HPMC dependence also scribed previously [14] and are presented in Fig. 4

19means that the F images, as well as being quantita- and Table 4.tive measures of the drug distribution itself, can also The self-diffusion constants for the water and forbe used to obtain the HPMC concentrations from the both drugs decrease greatly as the HPMC concen-effect of variation of T in a series of experiments, tration increases, consistent with all three compo-E

Fig. 4. Self-diffusion coefficients of water (open squares), triflupromazine–HCl (open triangles) and 5-fluorouracil (filled circles) in selectedHPMC mixtures. No distinction has been made between mixtures of varying drug concentration.


Table 4 axis of the tube containing the tablet and water as inMeasured self-diffusion coefficients (D) for water and model Fig. 5. The one-dimensional image which resultsdrugs triflupromazine–HCl (triflu) and 5-fluorouracil (5flu) in

then yields the profile of the drug along the axis ofmixtures of HPMC, drug and waterthe tube as indicated. The profile will change with

a bHPMC triflu 5flu D time as the tablet swells and the drug also diffuses25 2 21(w/w, %) (w/w, %) (w/w, %) (10 cm s )within and out of the gel. In addition, since only

0 – – 2.17 mobile material is detected within the tablet, the4.71 – – 1.95 19observation of F signal reflects the diffusion of9.44 – – 1.83

water into the dry tablet. However, data in this18.7 – – 1.4028.5 – – 0.96 region are not considered reliable because of the high38.3 – – 0.51 HPMC concentrations involved and the somewhat

heterogeneous nature of the sample at very low water0 0.5 – 0.4530 1.0 – 0.367 contents.0 3.0 – 0.194 The one-dimensional imaging studies of drug-con-4.61 1.0 – 0.181 taining HPMC tablets are shown in Fig. 6. The4.11 3.0 – 0.131

HPMC tablets, prepared with either triflupromazine–9.66 1.0 – 0.102HCl or 5-fluorouracil, contained essentially the same9.64 3.0 – 0.094

19.3 1.0 – 0.036 weights of the drug and, coincidentally, the same1919.3 3.0 – 0.060 moles of F. Thus, the comparison between the

0 – 0.50 1.01 triflupromazine–HCl and 5-fluorouracil distributions0.909 – 1.0 0.92 in the two systems, Fig. 6, is presented in con-

10.6 – 0.54 0.869.93 – 1.0 0.86

20.4 – 0.50 0.2818.0 – 1.1 0.3230 – 1.0 0.3431.5 – 3.0 0.2737.5 – 1.1 0.1739.7 – 3.0 0.15

a Corrected for 5.4% moisture content.b Measured by the PGSE method, error65 in the last reported

digit.

nents becoming much less mobile, as would beexpected. 5-Fluorouracil is much more mobile thantriflupromazine–HCl, as anticipated, due to the for-mer’s smaller size and the greater number and typeof different interactions with the gel which arepossible for the latter. These data, however, give aquantitative expression to these effects and, as willbe seen later, may be used to predict the releasebehaviour of other systems.

3.2. Imaging studies of the distributions oftriflupromazine–HCl and 5-fluorouracil in the

Fig. 5. Initial state of tablet within the NMR tube is shown withswelling tabletsthe direction of the gradient indicated. Below are representationsof the drug distribution at short and long swelling times. The19The one-dimensional F imaging experiments vertical line indicates the initial position of the water–tablet

were carried out with the z-gradient aligned with the interface.


19Fig. 6. Molar distributions of F in swollen HPMC tablets containing triflupromazine–HCl (filled circles) and 5-fluorouracil (open circles).19In all plots, the vertical axis is F concentration (mol / l) and the horizontal axis is distance (cm). The swelling times in the figure are (a) 1 h,

(b) 4 h, (c) 7 h, (d) 13 h, (e) 19 h, (f) 25 h, (g) 31 h, and (h) 37 h. The vertical line at 0.130 cm indicates the initial position of thewater–tablet interface.


Fig. 6. (continued)

19centration of F rather than drug so that the The total detectable amount of each drug as acomparison between the imaging results for the two function of swelling time is given in Table 5. In both

19drugs will be clearer. these systems, the detectable F in the images


Table 5 examples of which are shown in Fig. 7, were found19Total detectable moles F obtained from the T corrected one-2 to be identical to those obtained from imaging the

dimensional images of HPMC tablets containing triflupromazine–water component, confirming the validity of thisHCl and 5-fluorouracilmethod for the indirect determination of the HPMC

19 25Time Moles F (60.05310 ) concentrations. However, because of the constraints(h) 19

a a imposed by the smaller signal amplitudes in the FTriflu % 5Flu %studies, the experimental scatter was larger and the

1 2.20 31 2.06 29HPMC concentration profiles calculated from imag-4 3.46 49 1.67 24ing the water were used for further comparisons.7 4.18 59 3.86 54

13 5.29 75 5.08 7219 5.81 82 5.77 81 3.3. Investigation of the drug release mechanisms25 6.18 87 6.12 8631 6.42 90 6.30 89

The mechanism of drug release from hydrophilic37 6.52 92 6.35 89matrix tablets can be deduced from a comparison of

a 19Percentage ratio of detected versus known moles of F in the drug and polymer distributions. When the dis-tablet.

tributions of triflupromazine–HCl and HPMC areoverlapped, as in Fig. 8, one can clearly see that the

increased slowly with swelling time. Early on a large majority of the drug is still contained within theportion of the drug within the tablet was immobile polymer tablet. In each of the plots of this figure, theand invisible to the imaging technique. As water HPMC concentrations are given by the scale on the

19penetrated further and further into the tablet, the drug left and the F molarities of the drug are given bydissolved and its high resolution NMR signal grew. the scale on the right. The relative amounts of theHowever, the known concentration of (7.1160.04)3 drug and polymer in the dry tablet were about

25 19 24 1910 moles F within both tablets was not reached 4.56310 mol F/g. The parallel slopes for thefor either system. Even at the longest swelling times, triflupromazine–HCl and HPMC distribution suggest

19only about 90% of the total F signal was detect- that the ratio of triflupromazine–HCl to HPMCable. In the triflupromazine–HCl distributions, it is concentration remains fairly constant and, in fact, thebelieved that the very short T species that are calculated ratios in this region are within 20% of the2

present in HPMC concentrations above ¯30% cause original ratio. The similarity between the drug andsignal loss even in the 2-ms image so that the T polymer distributions suggest that the majority of the2

values calculated for regions of the polymer above drug movement in this regime is the result of the¯30% do not adequately correct for these species. In swelling of the polymer. Drug release only occurredthe 5-fluorouracil system, diffusion of the 5-fluoro- when the polymer concentration dropped belowuracil outside the region of the coil, and hence out of about 10% HPMC. At this point the HPMC is nothe imaging field of view, is the most likely explana- longer in the gel form and is slowly dissolving intion for the missing 10% of signal at very long times. solution, resulting in the erosion of the tablet. Thus,

Fig. 6 clearly shows that the 5-fluorouracil is able the dominant mechanism for triflupromazine–HClto diffuse more freely than the triflupromazine–HCl. release appears to be tablet erosion. The diffusionA proportion of the 5-fluorouracil appeared to have coefficients measured for triflupromazine–HCl in thediffused out of the field of view of the imaging HPMC concentrations of this eroding region range

26 26 2 21experiments as early as 13 h. from 2310 to 4310 cm s .As indicated in the experimental section, series of Similar comparisons for 5-fluorouracil in Fig. 9

experiments were carried out where images were suggest that the drug is diffusing freely from theobtained at different T values. Using these data and tablet from almost the start of the swelling. TheE

Eqs. (1) and (2) it is possible to construct the maximum of the exponentially-shaped concentrationconcentration profiles of the HPMC in the gel using gradient appears to coincide with a polymer con-the general protocol described previously for the centration of ¯30% at all swelling times. Thewater imaging studies [10]. The resulting profiles, measured diffusion coefficient of 5-fluorouracil in


Fig. 7. HPMC distributions calculated from the water images (filled circles) and triflupromazine HCl images (open circles) at swelling timesof (a) 4 h, (b) 13 h, (c) 25 h and (e) 37 h. The vertical line indicates the original position of the water–tablet interface. The vertical axes areHPMC (%) and the horizontal axes are distance (cm).


19Fig. 8. Molar F distributions of triflupromazine–HCl (open circles) overlapped with HPMC distributions (filled circles) calculated from19the water images. In all plots, the left vertical axis is HPMC concentration in weight percent, the right vertical axis is F concentration in

moles per liter and the horizontal axis is distance in cm. The swelling times in the figure are (a) 1 h, (b) 4 h, (c) 7 h, (d) 13 h, (e) 19 h, (f) 25h, (g) 31 h and (h) 37 h. The position of the 5% HPMC concentration region is indicated by the filled arrow. The horizontal arrows indicatethe appropriate axis for each curve.


Fig. 8. (continued)

26 2 2130% HPMC was 3.4310 cm s for the 1% range observed for triflupromazine–HCl in 0–10%26 2 21drug mixture and 2.7310 cm s for the 3% HPMC, suggesting that a diffusion coefficient of

26 2 21drug mixture. This is the same diffusion coefficient about 3310 cm s is required for drugs to


19Fig. 9. Molar F distributions of 5-fluorouracil (open circles) overlapped with HPMC distributions (filled circles) calculated from the water19images. In all plots, the left vertical axis is HPMC concentration (w/w, %), the right vertical axis is F concentration (mol / l) and the

horizontal axis is distance (cm). The swelling times in the figure are (a) 1 h, (b) 4 h, (c) 7 h, (d) 13 h, (e) 19 h, (f) 25 h, (g) 31 h, and (h) 37h. The position of the 5% HPMC concentration region is indicated by the filled arrow. The horizontal arrows indicate the appropriate axisfor each curve.


Fig. 9. (continued)

escape the swelling tablet and that the rate of the An estimate of the rate of tablet expansion can beswelling must be comparable to, or less than, that obtained by tracking the position of the tabletvalue. thickness as a function of swelling time. The edge of


Table 6 5% HPMC chosen as an intermediate concentrationDistance of various concentration regions in the HPMC dis- between the two extremes. The distances of the ¯0,tributions obtained from the water imaging experiments

5 and 10% concentration regions as a function ofTime Distance (60.01 cm) time are presented in Table 6. These distances(h) exhibited linear relationships with the square root of

¯0% HPMC 5% HPMC 10% HPMCtime, as shown in Fig. 10 resulting in different slopes

a0 0.13 – – for each concentration region. In order to compare1 0.25 0.20 0.19

the swelling of the tablet with the diffusion co-4 0.36 0.31 0.28efficients of the drugs, apparent rate constants for7 0.40 0.37 0.33

13 0.52 0.46 0.40 tablet expansion were determined by squaring the26 2 2119 0.60 0.52 0.46 slopes, resulting in 3.16310 cm s for 0%

26 2 2125 0.66 0.58 0.51 region, 2.34310 cm s for the 5% region and31 0.73 0.63 0.54 26 2 211.59310 cm s for the 10% region, with an37 0.79 0.67 0.57

26 2 21average of (2.460.8)310 cm s . These appar-a Only the 0% region should coincide with the tablet thickness ent expansion rates are simple empirical measures of

at zero time.the movement of the tablet front. The previous datasuggest that drug will not be released from the

the tablet can be defined by a particular HPMC HPMC tablet until the diffusion coefficient of the26 2 21concentration from the HPMC distributions. Three drug was approximately 3310 cm s . The

different concentration regions were considered as ‘apparent expansion’ rates of the tablet edge arerepresenting the ‘edge of the tablet’: ¯0% HPMC within the same range as the diffusion coefficientequivalent to the tablet edge measured by bulk required for drug release suggesting that the relation-methods, 10% HPMC discussed previously [10] as ship between the diffusion coefficients of the drug inthe transition concentration for gel formation, and a polymer and the apparent expansion rate of the

Fig. 10. Distance of various regions in the swelling HPMC tablet as a function of the square root of the swelling time. The three sets of dataare (a) |0% (filled squares), (b) 5% (open triangles) and (c) 10% (filled circles). The linear least squares fit to each set of data resulted in: (a)

1 / 2 2 1 / 2 2 1 / 2slope 0.1066 cm h , intercept 0.134 cm and r 0.999, (b) slope 0.09170 cm h , intercept 0.121 cm and r 0.999 and (c) 0.07570 cm h ,2intercept 0.124 cm and r 0.995.


polymer, as determined by plots such as Fig. 9, could ratio of the square of the slopes indicates that the ratebe used in other systems to determine drug release of 5-fluorouracil release from these HPMC tabletsbehaviour. was about five times faster than the rate of tri-

Thus, from the measurement of self diffusion flupromazine–HCl release.constants for drugs at different HPMC concentrationsas described earlier, it should be possible to predicttheir release behaviour. These measurements would 4. Conclusions

19not be restricted to F containing drugs and most19importantly would need only standard high resolu- F NMR imaging of model drugs has been used

tion NMR equipment such as is available in many to monitor the drug distributions within swellingpharmaceutical research laboratories. HPMC tablets. The HPMC distributions calculated

19The intermediate 5% HPMC region was chosen as from F variable-T experiments on both drugsE

the dividing line between the inside and outside of agreed well with the HPMC distributions calculatedthe tablet with the latter considered to contain the from the water experiments. A comparison of thedrug released from the device at each swelling time. distributions of drug and polymer showed that theThe concentrations of drug in the two regions of the majority of triflupromazine–HCl remained within thesystem are given in Table 7. The amount of drug swollen tablet even at long swelling times. Inreleased from the triflupromazine–HCl tablets in- contrast, the 5-fluorouracil diffused within the tabletcreased slowly with time. The amount of 5-fluoro- system at early times and more of this drug wasuracil released also increased with time and was released. The condition for drug release from thegreater than the triflupromazine–HCl released at all tablet appeared to be that the diffusion coefficient oftimes. After a short period of swelling, the 5-fluoro- the drug be greater than the expansion rate of the

26 2 21uracil had diffused out of the imaging field of view tablet, estimated as (2.460.8)310 cm s . Forwhich made the determination of this drug’s release triflupromazine–HCl, this condition was met only atdifficult. Therefore, at 37 h, it was assumed that the the edge of the swelling tablet where low HPMCamount of 5-fluorouracil released from the tablet was concentrations resulted in tablet erosion. The diffu-the total concentration of drug minus the amount still sion coefficient of 5-fluorouracil in 30% HPMC wasremaining inside the tablet. For both of the drugs, the large enough to satisfy the condition for drug releasefraction of drug released was a linear function of the and this drug escaped the tablet through diffusion atsquare-root of time indicating that drug release from this HPMC concentration. The fraction of drugthe swelling HPMC tablet was Fickian (Fig. 11). The released for both drugs was linear with the square-

Table 719Drug inside and outside the HPMC tablets as determined from the F imaging experiments; the edge of the swelling tablet, defined as 5%

HPMC, is determined from the HPMC distributions obtained from the water imaging experiments19 25Time Tablet Moles F (60.05310 )

(h) edge(cm) Triflupromazine–HCl 5-Fluorouracil

Inside Outside Inside Outsidea a a a1 0.20 1.57 0.63 0.75 1.30a a a a4 0.31 2.92 0.53 0.46 1.21

7 0.37 3.46 0.72 2.13 1.7313 0.46 4.34 0.94 2.95 2.1319 0.52 4.72 1.09 3.42 2.3525 0.58 5.02 1.17 3.68 2.4431 0.63 5.18 1.24 3.81 2.49

b37 0.67 5.14 1.39 3.87 2.48 (3.24)a Deviations at early times are the result of the higher relative noise level.b 19Amount released if the 10% missing from the maximum moles F is assumed to be outside the imaging field of view.


pany for the gift of the HPMC used in this study andDr. R. Miller and R. Oates of the Department ofPharmaceutical Sciences, UBC, for their help in theuse of a rotary tablet press.

References

[1] D.A. Alderman, A review of cellulose ethers in hydrophilicmatrices for oral-controlled release dosage forms, Int. J.Pharm. Technol. Prod. Manuf. 5 (3) (1984) 1–9.

[2] US Pharmacopeial Convention, Rockville, MD, USA. TheUS Pharmacopeia National Formulary, 1995. Dissolution,

711..[3] P. Gao, R.H. Meury, Swelling of hydroxypropyl methyl-

cellulose matrix tablets. 1. Characterization of swelling usinga novel optical imaging method, J. Pharm. Sci. 85 (7) (1996)725–731.

[4] L.S. Cutts, S. Hibberd, J. Adler, C.D. Melia, Characterisingdrug release processes within controlled release dosageforms using the confocal laser scanning microscope, J.Control. Rel. 42 (1996) 115–124.

[5] C.D. Melia, P. Marshal, P. Stark, S. Cunningham, A.Kinahan, J. Devane, Investigating in vitro drug releasemechanisms inside dosage forms monitoring liquid ingressinto hpmc hydrophilic matrices using confocal microscopy,in: D. Young, J.G. Devane, J. Butler (Eds.), In Vivo–in VitroCorrelations. Advances in Experimental Medicine and Biolo-gy, Volume 23, Plenum Press, New York, 1997, p. 129.

[6] R. Bowtell, J.C. Sharp, A. Peters, P. Mansfield, A.R. Rajabi-Siahboomi, M.C. Davies, C.D. Melia, NMR microscopy ofFig. 11. Fraction of drug released as a function of the square-roothydrating hydrophilic matrix pharmaceutical tablets, Magn.of the swelling time. The two plots are (a) fraction triflup-Res. Imag. 12 (2) (1994) 361–364.romazine–HCl released and (b) fraction 5-fluorouracil released.

[7] A.R. Rajabi-Siahboomi, R.W. Bowtell, P. Mansfield, A.The linear least squares fit to each set of data resulted in: (a)1 / 2 2 1 / 2 Henderson, M.C. Davies, C.D. Melia, Structure and behavior0.02573 h , intercept 0.03644 and r 0.983 and (b) 0.06152 h ,

2 in hydrophilic matrix sustained release dosage forms: 2.intercept 0.07555 and r 0.990.NMR-imaging studies of dimensional changes in the gellayer and core of HPMC tablets undergoing hydration, J.Control. Rel. 31 (1994) 121–128.

root of time indicating that the drug release was [8] A.R. Rajabi-Siahboomi, R.W. Bowtell, P. Mansfield, M.C.Fickian. Davies, C.D. Melia, Structure and behavior in hydrophilic

matrix sustained release dosage forms: 4. Studies of watermobility and diffusion coefficients in the gel layer of HPMCtablets using NMR imaging, Pharm. Res. 13 (3) (1996)Acknowledgements376–380.

[9] M. Ashraf, V.L. Iuorno, D. Coffin-Beach, C.A. Evans, L.L.The authors acknowledge the financial assistance Augsburger, A novel nuclear magnetic resonance (NMR)

of the NSERC of Canada. A.I.B. acknowledges the imaging method for measuring the water front penetrationrate in hydrophilic polymer matrix capsule plugs and its roleaward of an NSERC PGS Fellowship and an I.W.in drug release, Pharm. Res. 11 (5) (1994) 733–737.Killam Predoctoral Fellowship. The authors ack-

[10] C.A. Fyfe, A.I. Blazek, Investigation of hydrogel formationnowledge the financial support of this project by from hydroxypropylmethylcellulose (HPMC) by NMR spec-Glaxo-Wellcome and helpful discussions with Dr. B. troscopy and NMR imaging techniques, Macromolecules 30Fahie. They also thank the UpJohn-Pharmacia com- (20) (1997) 6230–6237.


[11] C.A. Fyfe, A.I. Blazek, Complications in investigations of [13] H.Y. Carr, E.M. Purcell, Effect of diffusion on free preces-the swelling of hydrogel matrices due to the presence of sion in nuclear magnetic resonance experiments, Phys. Rev.trapped gas, J. Control. Rel. 52 (1998) 221–225. 94 (3) (1954) 630–638.

[12] R.L.Vold, J.S. Waugh, M.P. Klein, D.E. Phelps, Measurement [14] E.O. Stejskal, J.E. Tanner, Spin diffusion measurements:of spin relaxation in complex systems, J. Chem. Phys. 48 spin echoes in the presence of a time-dependent field(1968) 3831–3832. gradient, J. Chem. Phys. 42 (1965) 288–292.

quantitative nmr imaging study of the mechanism of drug release from swelling...

Documents