Table I-
`
`Ar
`
`Enoxacir
`
`Lomeflox
`
`buN‘
`
`5 M
`
`iloxacir
`Nalidixic
`
`Norfloxar
`
`Ofloxacir
`
`Pefloxac
`
`Pipemidi
`
`determii
`had beet
`pore Kog
`column.
`For at
`(BIP-l, é
`violet sp
`strong z
`Shimadz
`boric ac:
`chosen 2
`between
`serum a]
`eter was
`and 5, rt
`For ar.
`equippet
`and a re
`Waters .
`acetonit:
`for NFL
`the excit
`420 nml
`and the
`(Shimad
`were ca
`containi
`in the u]
`of calibr;
`curves v
`94.0—10:
`<5.3%.
`
`The I
`binding
`differer
`ug/mL)
`therape
`subject:
`
`
`
`Serum Protein Binding cf Lomefloxacin
`and Its Related Quinolones
`
`, a New Antimicrobial Agent 3
`
`EllCHl OKEZAKI", TETSUYA TERASAKI’F, MASATO NAKAMUFiAi
`AND AKIRA Tsutui'x
`
`Received December 16, 1987, from the *Central Research Laboratory, Hokuriku Seiyaku Co., Ltd., Katsu
`of Pharmaceutical Sciences, Kanazawa University, Kanazawa 920, Japan.
`yama, Fukui, Japan, and the *Faculty
`Accepted for publication November 22, 1988.
`
`
`Abstract 3 The serum protein binding of lomefloxacin (LFLX), a new
`quinolone (pyridonecarboxyiic acid), and its related analogues was
`studied by an ultrafillration technique. The extent of binding of quinolones
`was independent of the concentration of quinolones below 100 pg/mL in
`rat serum; but, above this concentration, the binding decreased with
`increased drug concentration in the case of nalidixic acid and analogue
`3. The extent of binding in rat serum differed widely among the
`quinolones examined [i.e., from 15% (nortloxacin) to 84% (nalidixic acid)
`at concentrations of O.4-—10.0 pg/mL]. Lomefloxacin was bound to serum
`proteins to the extent of 28.1, 20.1, and 20.6% in the sera of rats, dogs,
`and humans, respectively. The binding of nalidixic acid with rat serum
`albumin, which was very similar to that in rat serum, was concentration
`dependent. Some quinolone derivatives with a piperazinyl group or a
`relatively large-sized substituent at the 7-position exhibited a percentage
`unbound of ~70—80%, while some derivatives with small-sized substi-
`tuents gave a low percentage unbound of 20—30%. This suggests that
`there is a steric effect of the substituents at the 7-position of quinolones
`on their binding characteristics with serum proteins. The results of the
`present study indicate that quinolones bind mainly with albumin among
`serum proteins and that the remarkable difference of the extent of binding
`of quinolone analogues is related to the size of the substituent at the
`7-position of the molecule, possibly due to its steric effect.
`
`volume of the drug. In particular, if ft,i is constant, tissue
`distribution will depend on fp, as verified in rats for
`quinolones.10 Moreover, serum protein binding is important
`for the antimicrobial activity and toxicological response,
`which are associated with unbound drug concentrations in
`serum. With respect to quinolones, however, there appears to
`be no report on the structure~serum protein binding relation-
`ship or the mechanism of serum protein binding.
`The present study describes the binding characteristics of
`LFLX and its related analogues with serum for a wide
`concentration range.
`
`Serum protein binding of antimicrobial agents is one of the
`important factors that determine their pharmacokinetic and
`antibacterial behaviors in vivo. Recently, several quinolones
`(pyridonecarboxylic acids) with wide antibacterial spectra
`against gram-negative and gram-positive organisms have
`been developed. In these antimicrobial agents, pharmacoki-
`netic deficiencies of the first generation drugs,
`including
`nalidixic acid, such as high extent of serum protein binding,
`poor tissue distribution, and extensive metabolism, have been
`improved.1 Lomefloxacin (LFLX) is a newly developed fluor-
`inated quinolone derivative characterized by the presence of
`a methyl group at the 3-position of piperazine moiety (Table
`I).
`
`It has been reported that the serum protein binding of
`enoxacin and pefloxacin, which have bulky substituent
`groups at the 7-position of the quinolone ring, was rela-
`tively low (20—35%)2'3 compared with that of miloxacin
`(86%),4 which has a small substituent group. To confirm the
`substituent effect on the extent of serum protein binding,
`we synthesized various quinolone analogues (e.g., 1—5,
`Table I).
`Lomefloxacin has a broad spectrum of activity covering
`both gram-positive and gram-negative organisms.5 In hu-
`mans, LFLX is almost completely absorbed after an oral dose,
`eliminated predominantly by renal excretion, and metabo-
`lized only to a small extent.6 The new quinolones, enoxacin,2
`norfloxacin,7 ofloxacin,8 pefloxacin,3 and LFLX,9 distribute
`well into tissues. Generally speaking, the unbound fraction of
`a drug in serum (fp), as well as the unbound fraction in tissues
`(fl), are important factors in determining the distribution
`
`504/ Journal of Pharmaceutical Sciences
`Vol. 78, No. 6, June 1989
`
`Experimental Section
`Materials—Nalidixic acid (NA), ofloxacin (OFLX), and [14CJOFLX
`(57 ,uCi/mg) were kindly supplied from Daiichi Pharmaceutical,
`Tokyo, Japan. Norfloxacin (NFLX),
`lomefioxacin (LFLX),
`[14C]lomefl0xacin ([14C]LFLX; 9.29 ”Ci/mg), analogues 1—5, and
`pipemidic acid (PPA) were synthesized in Central Research Labora-
`tory, Hokuriku Seiyaku Co., Fukui, Japan. The structures of the
`quinolones used in this study are shown in Table I. Rat serum
`albumin (RSA, Fraction V) was purchased from Sigma Chemical (St.
`Louis, MO). All other reagents were commercially available and of
`analytical grade.
`Serum and Rat Serum Albumin Samples—Blood was obtained
`from rats, dogs, and healthy human volunteers, to whom no drugs or
`anticoagulants were given, and centrifuged at 3000 rpm for 10 min
`to obtain serum samples. Sera samples thus obtained were pooled and
`stored at —20 0C until use in binding experiments. The BSA was
`dissolved in Krebs-Ringer bicarbonate bufler (pH 7.4)11 to produce an
`albumin concentration of 3.95 g/100 mL.
`Drug Solutions—An exactly weighed quinolone was dissolved
`with 0.3 M NaOH and the solution was diluted with distilled water.
`Then, 0.3 M HCl was added to obtain a drug solution with a pH value
`of 7—8.
`
`Binding Experiments—Serum protein binding was determined
`by an ultrafiltration method. Unless otherwise mentioned, 300 [.LL
`of the drug solution was added to 6 mL of the pooled serum or RSA
`solution to obtain an exact drug concentration ranging from 0.1 to
`400 ,ug/mL. In the case of LFLX and OFLX, the drug solution
`contained the corresponding 14C-labeled compounds. The mixture
`was incubated for 30 min at 37 °C and then an aliquot of 1 mL was
`ultrafiltered using a micropartition system (MPS-l, Amicon, Lex-
`ington, MA) with a membrane filter (YMT-membrane, Amicon) at
`2500 rpm for 20 min at 37 °C. The ultrafiltrate was assayed by
`counting radioactivity or by HPLC, as described later. The con-
`centration of quinolone bound to proteins was calculated by
`subtracting the concentrations of the drug in the ultrafiltrate from
`the known total concentration in serum. The percent of total
`volume collected as filtrate was ~30—40%. Since the nonspecific
`adsorption of the drug to the membrane was small (<6%), no
`correction for nonspecific binding was made.
`Analytical Procedures—The concentrations of LFLX and OFLX
`in the ultraflltrate were determined by liquid scintillation counting
`by adding 100 ML of filtrate to 16 mL of liquid scintillation fluid
`(ASC—II, Aloka, Tokyo, Japan), and counting the radioactivity in a
`spectrometer (LSC-700, Aloka).
`The concentration of the other quinolones in the ultrafiltrate was
`
`0022-3549/89/0600-0504$01 . 00/0
`© 1989, American Pharmaceutical Association
`
`ALCON 2236
`Apotex Inc. v. Alcon Pharmaceuticals, Ltd.
`Case |PR2013-00012
`
`

`

` Table I—Ouinolone Analogues
`
`R2
`
`
`Analogue
`
`Enoxacin (ENX)
`Lomefloxacin (LFLX)
`
`X,
`X2
`R2
`R3
`Fl4
`
`/—\
`\_.Jm
`—-N
`NH
`—N
`NH
`\—<
`CH3
`
`N
`C
`
`C
`C
`
`CzH5
`02H5
`
`-
`
`——
`F
`
`F
`F
`
`OH
`F
`CZH5
`C
`C
`1
`—N(CH3)CH2CH20H
`F
`02H5
`C
`C
`2
`—NH-CH20H20H
`F
`02H5
`C
`C
`3
`—CH2N(CZH5)2
`—
`CZH5
`C
`CH
`4
`——OCZH5
`--
`CZH5
`C
`CH
`5
`——OCH20—
`—
`OCH3
`CH
`CH
`Miloxacin (MLX)
`CH3/—_\
`——
`——
`CZH5
`CH
`N
`Nalidixic acid (NA)
`\_/
`—N
`NH
`F
`—
`C2H5
`C
`CH
`Norfloxacin (NFLX)
`m
`—N
`N—CHS
`F
`——
`—CH(CH3)CH20
`C
`CH
`Ofloxacin (OFLX)
`m
`.
`\_/
`—N
`N—CH3
`F
`—
`CQH5
`C
`CH
`Pefloxacm (PFLX)
`m
`\_J
`—
`——N
`NH
`——
`02H5
`N
`N
`Pipemidic acid (PPA)
`
`using experimental animals, which is estimated to be >100
`ug/mL at a subacute dose of 1 g/kg in rats from previous
`pharmacokinetic data}?
`The percentages of unbound quinolones with rat serum
`proteins are listed in Table II. For example, the percentages
`ofunbound LFLX were 67.1 i 0.83, 66.6 i 0.59, 69.6 i 2.32,
`73.9 t 0.81, 71.4 i 1.16, 74.8 i 1.83, 74.6 i 0.59, and 77.5 i
`0.85 (mean : SEM, n = 5) at the total LFLX concentrations
`of 0.1, 0.4, 1, 4, 10, 40, 100, and 400 ug/mL, respectively.
`All the quinolones tested showed linear behavior in their
`serum protein binding in the therapeutic serum concentra-
`tion range. On the other hand, the serum protein bindings of
`NA and 3 were nonlinear in the toxic concentration range,
`while those of other quinolones were still linear, as shown in
`Figure 1. These results suggest that nonlinearity of serum
`protein binding should be considered in acute and subacute
`toxic studies of some quinolones using experimental animals,
`and raise a possibility that the quinolones to be developed in
`the future may exhibit nonlinear pharmacokinetics due to
`their nonlinear serum protein binding in a therapeutic con-
`centration range. Thus, possible concentration dependency of
`serum protein binding of quinolones is a matter of clinical
`importance.
`The extent of binding of clinically usable quinolones in rat
`serum was in the order NA > LFLX > OFLX > PPA > NFLX.
`As expected from the data listed in Table II, quinolone
`derivatives with a piperazinyl group (NFLX, LFLX, OFLX,
`PPA) or a relatively large-sized substituent (4) at the 7-
`position exhibit a percentage unbound of ~70—80%. On the
`contrary, the other derivatives, with small-sized substituents
`at the 7-position, such as NA, 1, 2,. 3, and 5, gave a low
`percentage unbound of 20430172, at 5 ug/mL. The low extent of
`binding of these quinolones with serum proteins may be
`attributed to the steric hindrance by the substituents at the
`7—position of the molecule.
`Nalidixic acid (NA) exhibited nonlinear binding not only
`with rat serum but also with RSA, although the percentage
`
`determined by HPLC assay. A portion (50 ab) of the ultrafiltrate that
`had been passed through a membrane filter (0.45 pm, Nihon Milli-
`pore Kogyo, Yonezawa, Japan) was injected onto an HPLC analytical
`column.
`For analysis of NA and analogues 1—5, a solvent delivery system
`(BIP-l, Japan Spectroscopic, Tokyo, Japan), equipped with an ultra-
`violet spectrophotometer (UVIDEC-IOOV, Japan Spectroscopic) and a
`strong anion exchange column (2.1 mm X 50 cm, Zipak SAX,
`Shimadzu, Kyoto, Japan), was used. The mobile phase was aqueous
`boric acid solution of pH 9.0. The concentration of boric acid was
`chosen arbitrarily from 0.02 to 0.05 M so that a good separation
`between the blank peak obtained for the ultrafiltrate of the pooled
`serum and each drug peak could be established. The spectrophotom-
`eter was set at 258, 277, 284, 279, 252, and 270 nm for NA, 1, 2, 3, 4,
`and 5, respectively.
`For analysis of NFLX and PPA, a solvent delivery system (BIP-l),
`equipped with a spectrofluorometer (FF-110, Japan Spectroscopic)
`and a reversed-phase column (3.9 mm X 30 cm, u-Bondapak Cw,
`Waters Associates, Milford, MA), was used. The mobile phase was
`acetonitrile:0.05 M citric acidrl M ammonium acetate (1628321 v/v%
`for NFLX; 1318611 v/v% for PPA). The spectrofluorometer was set at
`the excitationiemission wavelength of 330:420 nm for NFLX and 340:
`420 nm for PPA. For both HPLC systems, the flow rate was 2 mL/min
`and the peak areas were recorded with a Chromatopac C-RBA
`(Shimadzu). The unknown concentrations in the ultrafiltrate samples
`were calculated by comparing the peak areas for the samples
`containing the known concentrations of the standard drug dissolved
`in the ultrafiltrate of the pooled serum. For this calculation, two sets
`of calibration curves were prepared: 005—10 and 10—100 ug/mL. Both
`curves were shown to be linear. The mean recoveries obtained were
`94.0—103.3% for all compounds, and the coefficients of variation were
`<5.3%.
`
`Results and Discussion
`
`The present study was carried out to give insight into the
`binding characteristics of several quinolones with sera of
`different animal species. The concentration range (01—400
`ug/mL) of quinolones employed in this study covered both the
`therapeutic serum concentrations (1—~20 ug/mL) in human
`subjects and the serum concentrations in toxicological studies
`
`F
`F
`F
`F
`F
`
`Journal of Pharmaceutical Sciences/ 505
`Vol. 78, No. 6, June 1989
`
`'aculty
`
`, tissue
`ate for
`portant
`sponse,
`ions in
`wars to
`alation—
`
`stics of
`a wide
`
`3]OFLX
`:eutical,
`LFLX),
`—5, and
`Labora-
`s of the
`. serum
`ical (St.
`a and of
`
`btained
`lrugs or
`10 min
`:led and
`3A was
`duce an
`
`ssolved
`l water.
`I-I value
`
`rmined
`300 [4L
`or RSA
`1 0.1 to
`olution
`iixture
`nL was
`n, Lex-
`con) at
`yed by
`re con-
`
`ted by
`:e from
`f total
`:pecific
`%), no
`
`OFLX
`unting
`n fluid
`by in a
`
`te was
`
`7 1 . 00/0
`)ciation
`
`

`

`. Nakz
`mart
`Y. 2t
`moth
`. Nag‘
`29(S
`Oka;
`Tach
`. Nag:
`T.; T
`. Oke:
`Tsuj:
`. ShaV
`32, 4
`. Oke:
`
`
`
`v
`
`V
`
`1’77
`i O
`
`V
`
`O
`
`a
`
`,
`
`A
`
`'
`
`I
`
`U
`
`_
`D
`.nu‘:J
`Ifi—l—.__L___l___1
`O.
`100
`200
`300
`400
`Total concentration (pg/ml)
`
`A01 100
`E
`U
`
`80
`
`60
`
`4
`
`20
`
`O
`
`E
`3
`
`8g
`
`“5
`
`C
`g
`DO:
`
`W
`
`90
`
`Percentageofunbounddrug
`
`CD 0
`
`|—_1__;_1—1
`O
`100
`200
`300
`400
`Total concentration (ug/ml )
`Figure 1—Effect of concentration of quinolones on the extent of binding
`
`with rat serum albumin (:1) and rat serum (other symbols). Panel A
`presents nalidixic acid (ELI), norfloxacin (V), lomefloxacin (O), and 3
`(A). Panel B presents ofloxacin (A) and pipemidic acid (0). Each point
`represents the mean : SEM (n = 3—5).
`
`ance, respectively. Therefore, it is suggested that the difference
`in CLR (per body weight) of LFLX can be attributed to the
`species differences of R or CLint,s’ because of no significant
`difference in fp among animal species.
`In conclusion, quinolones were bound mainly with albumin
`among serum proteins, and the difference of the extent of
`serum protein binding of quinolone analogues is related to the
`size of the substituent at the 7-p0sition of the molecule,
`possibly due to its steric effect.
`
`PPN"
`
`References and Notes
`. Shimada, J. Microbiology 1986, 222—225.
`Nakamura, S.; Kurobe, N.; Kashimoto, S.; Ohue, T.; Shimizu, S.
`Chemotherapy (Tokyo) 1984, 32(8-3), 86—94.
`Montay, G.; Goueflbn, Y.; Roquet, F. Antimicrob. Agents Che-
`mother. 1984, 25, 463—472.
`Izawa, A.; Kisaki, Y.; Irie, K.; Eda, Y.; Kornatsu, T.; Namiki, H.;
`Mizutani, T.; Nagate, T.; Kangouri, K.; Ohmura, S. Chemother‘
`apy (Tokyo) 1978, 26(S—4), 48—59.
`.
`5. Hirose, T.; Okezaki, E.; Kato, H.; Ito, Y.; Inoue, M.; Mitsuhashi,
`S. Antimicrob. Agents Chemother. 1987, 31, 854—859.
`
`Table ll—Percentage Unbound‘ for Quinolones in the Sera of the
`Flat, Dog, and Human and in 4% Solution of Flat Serum Albumin
`(RSA)
`
`Concentration
`Animal
`Percentage
`Eggs},
`Species
`Unbound”
`
`Rat
`0.1—«4000
`71.9 i 0.72 (40)
`Dog
`0.1—4oo.0
`79.9 i 0.74 (40)
`Human
`0.1—400.0
`79.4 i 0.93 (40)
`Flat
`5.0
`15.0 x 0.29 (3)
`Flat
`5.0
`17.6 i 0.75 (3)
`Rat
`5.0—40.0
`35.1 i 0.68 (9)
`100.0
`39.7 i 0.37 (3)
`200.0
`46.4 r 0.88 (4)
`400.0
`58.7 t 1.05 (3)
`5.0
`78.2 i 7.87 (3)
`
`Rat
`
`5
`
`Nalidixic acid
`
`Nortloxacin
`
`Ofloxacin
`Pipemidic acid
`
`Hat
`
`Flat
`
`RSA
`
`Flat
`
`Rat
`Flat
`
`5.0
`
`0.11—10.0
`40.0
`100.0
`200.0
`400.0
`10.0—20.0
`40.0
`100.0
`200.0
`400.0
`0.4-100.0
`
`29.4 i 1.62 (3)
`
`15.9 i 0.38 (2)
`22.0 i 0.15 (5)
`28.8 t 0.17 (5)
`33.1 i 0.60 (5)
`49.9 i 0.08 (5)
`8.4 i 0.11
`(6)
`9.7 i 0.06 (3)
`10.4 t 0.65 (3)
`16.7 i 0.39 (3)
`29.4 i 0.88 (3)
`85.3 t 0.69 (35)
`
`04-4000
`0.4—100.0
`
`77.2 i 0.46 (40)
`77.7 t 1.20 (35)
`
`M a
`
`Determined by an ultrafiltration method at 37°C. ”Determined at
`drug concentrations of0.1 (or 0.6), 0.4, 1,4 (or 5), 10, 20 (only for nalidixic
`acid in RSA solution), 40, 100, and 400 pg/ml, when the concentration
`range was indicated; the number of determinations at each concentration
`was 3—5. cMean i SEM; the numbers in parentheses are the total
`number of determinations.
`
`unbound was lower in albumin than in serum. Based on the
`results of binding of NA with RSA, it is suggested that
`quinolones bind mainly with albumin among the serum
`proteins and that they share the same binding sites on the
`protein molecule. However, the binding with albumin tended
`to be stronger than that with serum, presumably due to some
`influences by unknown endogenous substances in serum.
`The percentages ofunbound LFLX determined in the binding
`experiments with sera ofdogs and humans are given in Table II.
`The percentages of unbound LFLX were 75.0 i 0.32, 75.9 1‘
`0.96, 76.1 i 2.08, 83.6 i 0.48, 80.1 i 1.43, 81.6 i 1.98, 83.4 i
`1.72, and 83.2 i 2.14% in dog serum, and 77.3 i 0.42, 74.1 i
`0.23, 75.6 i 0.30, 77.8 i 1.82, 81.7 i 2.34, 80.2 i 0.59, 78.9 i
`0.87, and 89.7 i 4.17% in human serum at total LFLX concen-
`trations of0.1, 0.4, 1, 4, 10, 40, 100, and 400 ug/mL, respectively.
`The extent of binding of LFLX was very similar in rats, dogs,
`and humans: thus, there was no species difl‘erence of serum
`protein binding of LFLX. There is a significant difference in the
`renal clearance (CLR) among rats, dogs, and humans (rats, 9.36;
`dogs, 2.19; humans, 2.97 mL/min/kg).12s13 Considering that
`these values are much smaller than the renal plasma flow rate
`(rats, 25; dogs, 15; humans, 10 mL/min/kg),14 the renal excretion
`of LFLX is not limited by plasma flow. Therefore, CLR can be
`expressed by the following equation:
`
`CLR : fp(1 — R)(GFR + CLint,s)
`
`(1)
`
`where R, GFR, and CLW,s represent the fraction of reabsorp-
`tion, glomerular filtration rate, and secretory intrinsic clear~
`
`506/ Journal of Pharmaceutical Sciences
`Vol. 78, No. 6, June 1989
`
`Drug
`Lomefloxacin
`
`1
`2
`3
`
`4
`
`

`

`
`
`12
`
`. Nakashima, M.; Uematsu, T.; Takiguchi, Y.; Mizuno, A.; Kana—
`maru, M.; Tsuji, A.; Kubo, S.; Nagata, 0.; Okezaki, E; Takahara,
`Y. 26th Interscience Conference on Antimicrobial Agents Che-
`motherapy; New Orleans, LA, 1986; Abstract 430.
`. Nagatsu, Y.; Endo, K.; Irikura, T. Chemotherapy (Tokyo) 1981,
`29(S-4), 105—118.
`. Okazaki, 0.; Kurata, T.; Hashimoto, K.; Sudo, K.; Tsumura, M.;
`Tachizawa, H. Chemotherapy (Tokyo) 1984, 32(8-1), 1185—1202.
`. Nagata, 0.; Yamada, T.; Yamaguchi, T.; Okezaki, E.; Terasaki,
`T.; Tsuji, A. Chemotherapy (Tokyo) 1988, 36(8-2), 151—173.
`. Okezaki, E.; Terasaki, T.; Nakamura, M.; Nagata, 0.; Kato, H.;
`Tsuji, A. Drug Metab Dispos, in press.
`. Shaw, L. M.; Fields, L.; Mayock, R. Clin. Pharmacol. Ther. 1982,
`32, 490—496.
`. Okezaki, E.; Ohmichi, K.; Koike, S.; Takahashi, Y.; Makino, E.;
`
`400
`
`ml)
`
`___l
`
`400
`
`ll )
`
`Jinding
`anel A
`and 3
`h point
`
`mashi,
`
`arence
`to the
`.ficant
`
`numin
`ant of
`to the
`acule,
`
`izu, S.
`
`r Che-
`
`ki, H. ;
`other-
`
`13.
`
`14.
`
`E‘ggasaki, T.; Tsuji, A. Chemotherapy (Tokyo) 1988, 36(S-2), 132—
`Nakashima, M.; Uematsu, T.; Takiguchi, Y.; Mizuno, A.; Kana-
`maru, M.; Kubo, S.; Takahara, Y.; Okezaki, E.; Nagata, O.
`Chemotherapy (Tokyo) 1988, 36(S -2), 201—239.
`Bischofl‘, K. B.; Dedrick, R. L.; Zaharko, D. S.; Longstreth, J. A.
`J. Pharm. Sci. 1971, 60, 1128—1133.
`
`Acknowledgments
`The authors greath/Iacknowledge the technical assistance of Mr.
`Yoshihiro Kume and
`iss Miho Aoki. Thanks are also due to Daiichi
`Pharmaceutical Co. for the ifts of NA, OFLX, and [14C]OFLX. This
`work was supported in part y a Grant-in-Aid for Scientific Research
`from the Ministry of Education, Science and Culture, Japan.
`
`Journal of Pharmaceutical Sciences/ 507
`Vol. 78, No. 6, June 1989
`
`