Radioimmunoassay of a New Angiotensin-Converting Enzyme Inhibitor (Perindopril) in Human Plasma and Urine: Advantages of Coupling Anion-Exchange Column Chromatography with Radioimmunoassay L. DOUCET*~, B. DE VEYRAC*,M. DELAAGE*, H. CAILLA*, C. BERNHEIM*, M. DEVISSAGUET~

AND

Received May 15, 1989, from 'lmmunotech S. A. Case 915, 13288 Marseille Cedex 9, and the *lnstifut de Recherches lnternationalesServier, Accepted for publication November 1, 1989. 22, rue Gamier, 92200 Neuilly Sur Seine, France.

isolated by Immunotech (France). Preparation of the ImmunogenSince FT is not immunogenic, a hapten:protein conjugate was prepared by coupling PT to bovine serum albumin (BSA) using carbodiimideas the activator. Then, 10 mg of PI' and 0.1 mg of tritiated PT (CEA, specific activity >40 Ci/mmol; 20 pL) were added to 2 mL of a 10 mg/mL BSA solution(Boehringer)in 50 mM MES buffer, pH 5.5 (MES:2-morpholinoethaneaulfonic;Sigma). Then 10 mg of l-ethyl-3-3-dimethylaminopropyl carbodiimide* HCl (Sigma) was added in three portions, one every 20 min. The conjugatewas then purified by chromatography on a Pharmacia G 25 PD 10 column and eluted with a lo-' M NaCl solution (Merck).The concentration of BSA was defined by UV absorbance at 280 nm, as P'T does not absorb at this wavelength. The concentration of PT was determined according to its specific radioactivity. The molar ratio hapten:BSA was then calculated to be 5.6; the coupling yield, determined by measurement of radioactivity, was 7%. Immunization and Bleeding-The PT:BSA conjugate (250 was emulsified in Freund's complete adjuvant (Difco) and injected subcutaneously into rabbits. Four booster injections were administered a t three-week intervals. After the final injection, the rabbits were bled and the serum was collected for testing the antibody content. Aliquots of serum were stored at -20 "C. Perindopril' [P(2S,3aS,7&)- 1-1[ 1-ethoxy-S-buty1amino)Preparation of the Radioligand-Since PT cannot be directly (S)-propionyll-octahydroindole-2-carboxylicacidl i s a new labeled with iodine, a tyrosyl derivative was fist prepared by angiotensin-converting enzyme (ACE) inhibitor.2 It is a procoupling glycyltyrosinamide (GTNH,) with PT by the carboxyl group drug that has t o be converted in vivo into the active metabof the side chain. Then 5 mg of PT and 0.1 mg of tritiated PT (20 IJ1) olite perindoprilat [PT;(2S,3aS,7uS)-1-{2[1-(S) carboxywere added to 1 mL of 2 x lo-' M GTNH2 solution (Bachem) in 50 butylaminol(S)-propionyl}octahydroindole-2-carboxylic mM MES buffer (pH 5.5). The GTNH2 was purified beforehand by acid]. chromatography on an SP Sephadex column (Pharmacia). Then 10 In order to quantitate these compounds in biological sammg of 1-ethyl-3-3-dimethylaminopropyl carbodiimide * HCl were ples during pharmacokinetic studies, a radioimmunoassay added in three portions every 20 min. The reaction mixture was then chromatographed on an SP 25 Sephadex column, equilibrated with was developed which allowed the assay o f PT as i s and of P M sodium phosphate buffer (pH 8; Merck), and eluted with a after quantitative conversion i n t o PT by alkaline hydrolysis. lo-' to 4 x lo-' M NaCl gradient. An additional purification wan The first results led u s t o suspect a third immunoreactive carried out by fast protein liquid chromatograph (FPLC) on a Mono compound, which was subsequently isolated by chromatogQ column (Pharmacia) equilibrated with 2 x 10- M triethanolamine raphy and identified as a glucuronide of PT (PT-G1.3 buffer (pH 7.6);the elution was performed with a lo-' to 1M NaCl In t h i s report, the characteristics of the RIA method are gradient. The molar ratio, determined as described in the preparation described and i t s validity for the pharmacokinetic studies of of the immunogen, was 1.05 and made iodination of the tyrosyl P and i t s metabolites are assessed. In addition, i t s usefulness derivative possible. The method of Hunter and Greenwood with i s illustrated by a n example. chloramine T was used.6 First, 1 mCi of ['2611NaI (Amersham) was added to 50 pL, (2.68 nmol) of n G T N H , and 10 pL of 1 mg/mL chloramine T in 5 x M phosphate buffer (pH 7.2;Merck). The Experimental Section reaction was stopped after 1 min by adding 20 pL of a 1 mg/mL Materials4-Perindopril [P; (2S,3aS,7aS)-l-l[l-ethoxy-S-butyl- aqueous sodium metabisulfite solution (Merck) and 500 pL of 0.1 M amino)-(S)-propionyl]-octahydroindole-2-carboxylicacid], its active phosphate buffer (pH 4; Merck). The lZ6I[PT:GTHN,] was isolated from the reaction medium by (2S,3aS,7aS)-1-{2[1-(S) carboxybutylmetabolite perindoprilat [PT, chromatography on a Sephadex G25 superfine column (50 x 1 cm, amino] (S)-propionyl}octahydroindole-2-carboxylicacid], other Pharmacia) and eluted with 0.1 M acetate buffer (pH 4). The first metabolites of P which are the lactam forms BMD 4-1 "41eluted monoiodinated derivative was immediately diluted to give a carbethoxybutyl)-1.4-dioxo-( 1.2a)-perhydroindolopyrazineI,BMD 5200 000 cmp/mL solution in 0.1 M Tris-HC1 buffer (pH 8.5, Merck) 1 [N-(l-carboxybutyl)-1.4-dioxo-(l.2a)-perhydroindolopyrazinel, and containing BSA (1 g/L). The 4-mL aliquots were stored a t -20 "c. BMD 6 1 (isomer of BMD 5-11, and another inhibitor of ACE [l; Under these conditions, stability is maintained for 3 months. (2S,3aS,7aS)-1-{2[1-(S)-carboxy-l-methyl-3 butylaminol (S)Plasma Specimens-Two healthy male subjects were treated with propionyl)octahydroindole-2-carboxylicacid] were provided by Instia single dose of 8 mg (4 x 2-mg tablets) of P. Blood samples were tut de Recherches Servier (France). The glucuronide of PT (PT-G)was Abstract Perindopril(P) is a prodrug whose active metabolite perindoprilat (PT) is an antihypertensive agent which acts by inhibition of angiotensin-converting enzyme (ACE). Anti-PT antiserum was produced in a rabbit immunized against PT that was covalent1 linked to bovine serum albumin. The radioligand is an iodinated ("I) derivative of PT-glycyltyrosinamide. Both the drug (PT) and the prodrug (P) are assayed in the same sample; PT is assayed as is and P is assayed after quantitative alkaline hydrolysis into PT. Certain data obtained from such assays suggest the Occurrence in plasma and urine of a third immunoreactive component. A chromatographicfractionation of samples allowed us to isolate a new immunoreactive metabolite which was further identifiedas a glucuronide of PT (PT-G). Therefore, the whole assay was carried out as follows: biological samples were fractionated by stepwise chromatography on a anion-exchange resin (the first fraction contained P, the second contained PT, and the third contained PT-G); and RIA was performed on fractions 2 and 3 as is, and on fraction 1 after alkaline hydrolysis. Performancesand assessments of this method are presented together with an example of a pharmacokinetic profile.

B

0022-~9/90/08OO-074 1$0 1.0010 0 1990, American PharmaceuticalAssociation

Journal of PharmaceuticalSciences / 741 Vol. 79, No. 8, August 1990

taken before and 1.5,5, and 8 h after drug administration. The urine was collected before and during the time intervals 0-4 h and 4-8 h after drug administration. Samples were provided by Dr. Duchier (ThBrapharm, 19 rue de la Tour,75016 Paris, France). Assay Procedure-PerindopriI (P) and PT were assayed in the same sample; PT was assayed as is, and P was assayed after alkaline hydrolysis to PT.This was achieved by incubating 100 p L of the sample with 50 & of 1 M NaOH (Merck) for 1 h a t 37°C. Hydrolysis was stopped by addition of 50 CJ. of 1 M HCI (Merck), and the hydrolysate was assayed. The concentration obtained is the sum of P and PT.A 50-pL aliquot of the same sample was assayed as described to determine the concentration of PT.The concentration of P was calculated by subtraction of the concentration of PT from the "total" concentration obtained by RIA after hydrolysis. Standards were prepared by serial dilutions of a stock solution (lo-' M)of PT in 0.1 M Tris HCl buffer (pH 8.5; 1 glL BSA).The concentrations used were 0,0.039,0.078,0.16,0.31,0.63,1.3,2.5,5, and 10 nM. These solutions were ueed for the assay without hydrolysis, and diluted to 1:2 in 0.5 M NaCl to be at the same ionic strength (0.25 M NaC1) as that induced by the hydrolysis for the assay. The assay tubes for the standard curve and the unknown samples were set up according to the following protocol: all samples were assayed in duplicate; 50 & of radioligand and 100 & of antiserum at the working dilution were added to 5-mL tubes containing the sample or standard (50 4);aRer 22 h of incubation at 4 "C, 100 pL of blank plasma and 300 pL of 25%polyethylene glycol 6000 (PEG) were added to all tubes except the total; the tubes were shaken for 5 min and centrifuged at 3600 x g for 15 min at +4 "C; the supernatants were discarded; the radioactivity present in the pellet and in the total were counted (Packard A5412 gamma spectrometer).

Results and Discussion Titer of the Antiserum-The antiserum titer is expressed as the dilution of antiserum binding 50% of the tracer.6.7 The titer was determined to be 630.However, the working dilution for the RIA was 1:400in order to compensate for the fall in the ratio bound radioactivity:total radioactivity (B:T) caused by the concentration of NaCl of the incubation medium. Affinity of the Antiserum-The affinity constant (Kd) for PT (4 x lo-" M) and the antibody site concentration (4 x lo-' M) in the antiserum were calculated from the scatchard plot.8 Specificity of the Antiserum-The parent drug (PI, its known metabolites PT, BMD 4-1, BMD 5-1, BMD 6-1, PT-G, and the other ACE inhibitor (1) were tested and their IC, values were compared with that of PT. The percent cross reactivity was defined as the ratio of the concentration of PT to that of the tested compound ( x 100) in the incubation medium for a 50% displacement of the radioligand (ICso). Results are given in Table I. The antiserum has the same best affinity for PT and PT-G. The esterification of the carboxylate at the end of the lateral chain decreases the recognition by more than two orders of magnitude. The lactam forms (BMD) are poorly recognized. Cross reactivity with the other active ACE inhibitor (1) was 2.9%)its structure being close to that of PT (see structures). Table CSpeclflclty of the Antl-Perlndoprllet Antlberum'

Compound PT

PT-G P BMD 4-1 BMD 5 1 BMD &l 1

IC5os M

Cross Reactivity, Yo

1 x 10-10 1 x 10-10 3 x 10-8 3.5 x 10-5 2.2 x 10-5 8.7 x 10-7 3.5 x 10-9

100 100 0.3 2.9 x 10-4 4.6 x 10-4 1.2 x 10-2 2.9

a The compounds were tested and their lCm values were compared with that of PT; the cross reactivity percentage is defined as the ratio of the IC, value of PT to that of the tested compound ( x 100).

742 I Journal of Pharmaceutical Sciences Vol. 79, No. 8, August 7990

These results indicate that the epitope recognized by the antiserum includes the lateral chain on the nitrogen position. Characteristics of the RIA-Kinetics of the Formation of the Antibody:Antigen Complex-Incubations were performed a t t4 "C for variable times up to 24 h. The equilibrium was reached after 19 h, but for the convenience of the routine assay, an incubation time of 22 h was chosen. Alkaline Hydrolysis of Ferindopril (P)-To asseas the efficacy and duration of P hydrolysis, this compound was added to human plasma (final Concentration 9.1 nM) and then hydrolyzed in 0.33 M NaOH a t 37 "C for periods of time ranging from 30 min to 24 h. The hydrolysates were then diluted and assayed by RIA. The eventual breakdown of PT during hydrolysis was also assayed under the same conditions by adding this compound to human plasma in a h a 1 concentration of 11.8 nM. Finally, spontaneous hydrolysis of P was revealed in a 25 nM plasma solution or buffer solution of this compound that wae left to stand a t ambient temperature for periods of time up to 24 h. Aliquots were then sampled and assayed by RIA. Hydrolysis of P in 0.33 M NaOH was complete in 30 min at 37°C)and was not followed by breakdown of PT for up to 24 h. In buffer solution, required for the formation of the antigen:antibody complex, there was no spontaneous hydrolysis of P. In plasma, hydrolysis due to esterase activity did not exceed 6.4% after 24 h. Detection Limitstandard solutions of concentrations of PT ranging from 0 to 3.14 x lo-'' M were assayed 40 times. The amount in the tube that caused a 20% dlsplacement of the radioligand was stated to be the limit of detection; this was

determined to be 0.04nM(0.014ng/mL).Taking into account the dilution factors, the limit of detectionwas 0.32 nM (0.109 ng/mL) for urine samples and 1.6 nM (0.545 ng/mL) for plasma samples (samples were always diluted eight times in the assay tubes by addition of antiserum, radioligand, and NaC1, and plasma samples were diluted five more times in buffer to avoid interference in the antigen:antibody binding due to the protein). Reproducibility-To establish intra-assay precision, 12 standard concentrations ranging from 0 to 1 x M were assayed 40 times by the same person using the same diluted solutions of antiserum and radioligand. The coefficient of variation was 510.8% of the B:T on the exploitable part of the curve (B:T > 10%;Table 11).To establish interassay precision, standard concentrations ranging from 0 to 1 x M were assayed nine times by different persons using different solutions of antiserum and radioligand. The coefficient of variation was 5 10.4% of the B:T on the exploitable part of the curve (B:T > 10%;Table 111). Precision of the RIA-During the intra-assay reproducibility test, plasma samples containing both P and PT were diluted serially and assayed with or without hydrolysis. A linear regression was established between observed concentrations and theoretical concentrations. The coefficients of regression ranged from 0.992 to 0.999 for nonhydrolyzed samples (Table IV) and from 0.990 to 0.999 for hydrolyzed samples (Table V). Characterization of a New Metabolite (PT-GkSince it was possible to determine the concentrations of both P and PT in biological samples by radioimmunoassay with and without hydrolysis, samples were first analyzed this way. However, some results suggested the presence of a third immunoreactive compound. Therefore, in a first approach, we defined chromatographic conditions allowing a good separation between PT and P. A linear gradient of formic acid (50 mL of lo-* M HCOOH, and 50 mL of 0.5 M HCOOH) on a n anion-exchange resin (DOWEX)provides the elution of P and PT with a quantitative recovery. Then, a human plasma sample (subject 1, time 5 h; provided by Dr. Duchier) was chromatographed and eluted under the same conditions. All eluted fractions were assayed by RIA before and after alkaline hydrolysis. Two peaks were detected before hydrolysis; one was identified as PT and one was a new peak that eluted later. After hydrolysis, a third peak, corresponding to P, appeared (Figure 1). Therefore, only one additional immunoreactive compound, named PT-G, was detected. Therefore, it was demonstrated that alkaline hydrolysis converted PT-G into PT. Another human plasma sample containing the three compounds was hydrolyzed and then chromatographed under the same conditions. Only one peak was detected by RIA before and after hydrolysis at the Table ICReproducibillty of the Standard Curve: intra-assay Test’

Concentration of PT, M

B:T (mean f SD;

n = 40)

cv, %

0 9.82. 1.96*10-” 3.92.lo-‘’ 7.86* 1 OW’’ 1.57* lo-’’ 3.14.lo-’’ 6.29* lo-‘’ 1.257 1 O-g 2.515 10-9 5.030* lo-’ 1.006.lo-’

50.5f 0.4’ 50.2f 0.4 49.4f 0.3 48.7? 0.3 47.8t 0.4 44.7 f 0.3 40.5 f 0.3 32.1 t 0.4 21.4 f 0.3 12.0f 0.2 6.7f 0.2 3.9f 0.1

4.8 5.6 4.5 4.5 5.2 4.7 5.4 7.5 9.4 10.8 16.4 23.1

-

a Twelve standard concentrations ranging from 0 to 1 x 1 0-’ M were assayed 40 times. n = 34.

’

Table llCReproduciblllty of the Standard Curve: Interassay Test

Concentration of PT, M

B:T (mean t SD;

0 9.82* 3.92* lo-’’ 1.57* lo-’’ 6.29* lo-’’ 2.515 * 1 O-’ 1.006-lo-’

55.8f 1.1 54.6 ? 1.3 53.9 f 1.2 51.2f 0.9 39.3 f 1.2 16.4t 0.6 4.8t 0.7

(n

cv, %

= 9)

5.7 7.0 6.9 5.3 9.2 10.4 41.7

a Seven standard concentrations ranging from 0 to 1 x 1 0-’ M were assayed nine times.

Table IV-Preclslon of the RIA before Hydrolysls of Serial Dilutions of Plasma Samples

Sample 1

2

3

4

5

Dilution of the Sample

PT Concentration, nM

Theoretical

Assayed

-

108.3 48.1 20.1

100 200 400

54.2 27.1

50 100 200 400

30.8 15.4 7.7

200 400 600 1200

133.9 89.2 44.6

20 40 100 200

30 60 100 250

-

13.9 5.6 2.8

27.9 16.7 6.7

61.6 33.0 14.4 6.8 267.7 119.8 88.9 39.5 27.8 16.8 5.8 3.0 55.7 25.4 17.1 5.3

retention time of PT.For routine purposes, a new procedure was proposed. Plasma samples were first fractionated by stepwise chromatography, and then the three fractions were collected and assayed separately. Chromatographic Separation-Dowex AG 1 x 2 resin (200-400 mesh, chloride form, Biorad) was transformed to a formiate form by washing with 1M formic acid (Merck) and equilibrated with 0.2 M formic acid. This was then introduced into a Pharmacia C 10120 column to a height of 9 cm. The chromatography of the samples, plasma or urine, was fully automated. A 500-pL aliquot of the unknown sample was applied on the resin and eluted in a stepwise manner with 31 mL of 0.2 M formic acid and 20 mL of 0.5 M formic acid using a pump (Seres) adjusted to have a flow rate of 28 mM.1. The eluate was collected in three fractions of 11, 20, and 20 mL containing, respectively, P in the first fraction, PT in the second fraction, and PT-G in the third fraction. After elution was completed,the resin was washed with 10mL of 1M formic acid and regenerated with 15mL of 0.2 M formic acid. A 2-mL aliquot of each fraction was freeze dried (Usifroid SMH 15 freezedryer) and taken up in either 500 pL of distilled water for the first fraction or 0.05 M Tris HC1buffe~0.25M NaC1:0.5 g/L BSA for the second and third fractions. Therefore, dilution factors were introduced in comparison with the initial sample; the dilutions were 5.5 for the first fraction and 10 for the second and third fractions. All samples (fractions and stanJournal of Pharmaceutical Sciences I 743 Vol. 79, No. 8, August 1990

Table V-Preclrlon of the RIA after Hydrolysis of Serlal Dllutlons of Plasma S a m p b

50

Dilution of the Sample

Sample 6

"Total" PT Concentration, nM Theoretical Assayed

-

100 200

156.7 78.4 39.2

400 800 7

-

10 20

26.4 13.2 6.6

40 80 5 8

-

10 20

9

x v

52.7 26.4 13.4 7.3 36.9

80

108.7

160 320 640

54.3

135.0 80.4 38.4 21.5

v

106.0 60.5 30.8 16.3

0 i' 0

-

26.5 13.3

160

I

0.01

0.1

0.1 8

1.84

1

18.4

(nM)

10

184

(pg/tube)

Concentration Flgure 2-The RIA standard curve for perindoprilat (PT).

-

53.0

30

lot 01

40 80 40

40 80

.

20.

19.9 10.7 5.0 2.5 217.3

27.2 13.6

40

313.4 183.1 86.2 47.8

18.5 9.2 4.6 2.3

20

10

n

300h

n

I t

t L

i -

r

a,

0 t 0

0

0,3 12

0

24

36

48

10 2 0 30 4 0 50 GO 70 80 90 100

Fraction number

Time ( h ) Flgure %Plasma concentration versus time (48 h) profile for P (W), PT (0),and PT-G (0).

.-'0 4 0 c

0

30

c \

- 20'

!i

Q

10

0

10 20 30 4 0 50 GO 70 80 90 100

Fraction n u m b e r Flgum l-chromatography of a plasma sample on an anion-exchange

resin, eluted with a formic acid gradient. Radioimmunoassay was performed on each fraction before (top) and after (bottom)hydrolysis.

dards) were assayed at the same ionic strength (0.25M NaCl). The standards were prepared as previously described in the Experimental Section. A typical standard curve is shown in Figure 3. The first fraction containing P was converted by alkaline hydrolysis into PT, and the second and the third 744 I Journal of Phannaceutical Sciences Vol. 79, No. 8, August 1990

fractions were assayed without any additional treatment. The assay tubes were set up according to the protocol previously described in the Experimental Section. Reproducibility of the Chromatography-Three untreated urine samples were spiked with known concentrations of P and PT. The final concentrations were 5000 nM for each compound in sample 1,500nM for each compound in sample 2,and 10 nM for P and 5 nM for PT in sample 3. These samples were c h m a b graphed six times on six Merent columns and then assayed at three serial dilutions. The results showed coefficients of variation of 15.6% for P and 22.2% for PT in sample 3. Theae coefficienta were 516.6% for P and 513.6% for PT in the two other samples when they were sdliciently diluted (B:T> 10%). Recovery of the Chromatography-Analyses were performed in two different ways. On the one hand, the plasma specimen was spiked with known concentrations of P (ranging from 0 to 300 nM) and of PT (ranging from 2 to 30nM).On the other hand, plasma samples were provided by D. Duchier. Each time, the results of the assay of the chromatographic fractions were compared with those of the complete samples before (sample) and afier (hydrolyzed sample) hydrolysis. The chromatographic yield was therefore evaluated by the recov-

ery of the compound. In the spiked plasma samples, the mean recovery of PT (fraction blsample) was 92.82 8.6%.The mean recovery of P + PT (fraction a + bhydrolyzed sample) was 102 2 9.5%. In the plasma samples of treated subjects, the mean recovery of PT + FT-G (fractions b + dsample) was 92.6 I 4.1%. The mean recovery of P + PT + FT-G(fractions a + b + dhydrolyzed sample) was 108.5 2 4.2%. The results show quantitative recovery of the chromatography both for the spiked plasma and the plasma of the treated subjects, at both high and low concentrations. Perindopril Pharmacokinetice-Five healthy male volunteerswere treated with 4 mg of P.Blood was then collected before administration and at 20 intervals over a period of 120 h aRer administration. Plasma was separated shortly after collection. The mean plasma concentration versus time profiles of P, PT, and W-G9 are shown in Figure 3. The parent drug (PI is rapidly converted to PT, the major immunoreactive component, which is quanitatable at 48 h post dose. The profile of the metabolite PT-G appears to mimic that of the parent drug, P.

Conclusions Coupling chromatography with radioimmunoassay provides a useful method to determine the pharmacokinetics of

P in humans. This method permits the specific and selective quantitation of the prodrug (PI, the drug (I")and , the glucuronide (PT-GI, all of which are immunoreactive.

References and Notes 1. Vincent, M.;Remond, G.; Portevin, B.; Serkiz, B.; Laubie, M . Tetmhedron Lett. 1982,23,16,1677-1680. 2. Laubie, M.; Schiavi, P.; Vincent, M.; Schmitt, H. J. Cardiovasc. P h a r m o l . 1984,6,10761082. 3. Garnier, M. H.; Moc uard M. T.; Dabe, J. F.; Bertrand, M.; Luijten, W.; Bromet, N.;%rislkin, L.; Resplandy, G.; Devissaguet, M. Xenobiotica, in press. S-9780-1), the lactam 4. Perindopril ( PS-9490-1) erindoprilat (PT; metabolites BMD 4-1, BhfD 5-1, and BMD 6-1, and com und 1 (S-11171-1) are proprietary compounds of Institut de Recgrches Servier (France). 5. Hunter, W. M.; Greenwood, F. C. Nature 1962,194,495496. 6. Chard, T. In Introduction to R a d i o i n m u m s a y s and Related Technics; Elsevier: Amsterdam, 1978. 7. Jeffrey, C.; Travis, P. D. Fundamentals of RIA; Radioassay Publishers-Division of Scientific News Letters; 1979. 8. Scatchard, G.Ann. N.Y. Acad. Sci. 1949,51, 660-672. 9. Bmmet, N.; Leli&vre,E.; Vidal, D.; Grislain, L.; Devissaguet, M.; Jochemsen,. R: Ammoury, N.; Devissaguet, J. Ph. Jourde de Z'HTA, Paris (Prance), 15-16 December 1988, poster no. 94.

Journal of Pharmaceutical Sciences I 745 Vol. 79, No. 8, August 1990