`
`Pharmaceutical
`Sciences
`
`VOLUME 93, NUMBER 4
`APRIL 2004
`
`RESEARCH ARTICLES
`
`Biopharmaceutics of (i-Cyclodextrin Derivative-Based Formulations of Acitretin
`in Sprague-Dawley Rats
`Xin Liu, Hai-Shu Lin, Sui Yung Chan, and Paul C. Ho* .................................... (cid:9)
`Published online 16 January 2004
`
`High-Throughput Determination of the Free Fraction of Drugs Strongly Bound
`to Plasma Proteins
`Joachim Schuhmacher,* Christian Kohlsdorfer, Klaus Buhner, Tim Brandenburger, and Renate Kruk .. (cid:9)
`Published online 23 January 2004
`
`Effects of Long-Term Oral Administration of Polymeric Microcapsules Containing
`Tyrosinase on Maintaining Decreased Systemic Tyrosine Levels in Rats
`Binglan Yu and Thomas Ming Swi Chang* ............................................. (cid:9)
`Published online 20 January 2004
`
`Rapid and Accurate Prediction of Degradant Formation Rates in Pharmaceutical
`Formulations Using High-Performance Liquid Chromatography-Mass Spectrometry
`Richard T. Darrington and Jim Jiao* (cid:9) ................................................. (cid:9)
`Published online 20 January 2004
`
`Evaluation of the Protein Binding Ratio of Drugs by A Micro-Scale
`Ultracentrifugation Method
`Daisuke Nakai,* Kazuyo Kumamoto, Chisa Sakikawa, Toshiyuki I<osaka, and Taro Tokui .......... (cid:9)
`Published online 23 January 2004
`
`A Molecular Dynamics Simulation of Reactant Mobility in an Amorphous
`Formulation of a Peptide in Poly(vinylpyrrolidone)
`Tian-Xiang Xiang and Bradley D. Anderson* ........................................... (cid:9)
`Published online 28 January 2004
`
`Modulation of Intestinal P-Glycoprotein Function by Cremophor EL and Other
`Surfactants by an In Vitro Diffusion Chamber Method Using the Isolated
`Rat Intestinal Membranes
`Yasushi Shono, Hisayo Nishihara, Yasuyuki Matsuda, Shiori Furukawa, Naoki Okada, Takuya Fujita,
`and Akira Yamamoto* (cid:9) ........................................................... (cid:9)
`Published online 28 January 2004
`
`805
`
`816
`
`831
`
`838
`
`847
`
`855
`
`877
`
`(continued)
`
`Journal of Pharmaceutical Sciences
`VOL. 93, NO. 4, APRIL 2004
`
`3WILEY (cid:9)
`
`This journal is online
`
`Inter fencee
`
`Volume 93, Number 4 was mailed the week of March 22, 2004.
`
`www.interscience.wiley.com
`
`Page A
`
`ADAMIS EXHIBIT 1005
`
`
`
`Long-Term Stability Study of L-Adrenaline Injections:
`Kinetics of Sulfonation and Racemization Pathways
`of Drug Degradation
`
`DAVID STEPENSKY, MICHAEL CHORNY, ZIAD DABOUR, ILANA SCHUMACHER
`
`Research & Quality Control Laboratory, The Medical Corps, Mil. P.O. Box 02149, Israel Defense Forces,
`Israel Defense Forces, Israel
`
`Received 8 June 2003; revised 29 September 2003; accepted 21 October 2003
`
`Published online 30 January 2004 in Wiley InterScience (www.interscience.wiley.com ). DOI 10.10021 jps.20010
`
`ABSTRACT: Injectable formulations of L- adrenaline are commonly used in emergency
`medicine. Despite numerous studies, the comparative contribution and kinetics of the
`L- adrenaline inactivation pathways during storage have not been conclusively evaluated.
`We examined the kinetics of L- adrenaline degradation in a prospective study and
`determined the extent of drug inactivation by different pathways during and beyond the
`stipulated product shelf-life in 42 batches of adrenaline ampules stored under controlled
`conditions. The content of L- adrenaline and degradation products was determined with a
`chiral high-performance liquid chromatography (HPLC) assay, and the degradation
`products were identified by mass spectrometric detection as D- adrenaline and L- and
`D- adrenaline sulfonate. The kinetics of the content change with storage was analyzed
`simultaneously for L- adrenaline and the degradation products using kinetic modeling.
`The lower acceptable level of adrenaline content in the formulation stated by US
`Pharmacopoeia (90% as a sum of L- and n-isomers) was attained after 2.0 years of storage,
`at which time the content of the therapeutically active L- isomer amounted to as low as
`85%. The modeling revealed significant differences in the degradation kinetics in the
`formulations produced before and after 1997, whose cause remained unidentified in
`this study. © 2004 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci
`93:969-980, 2004
`Keywords: adrenaline; stability; chirality; HPLC (high-performance liquid
`chromatography)
`
`INTRODUCTION
`
`Adrenaline is a catecholamine compound that is
`commonly applied by intravenous injection in
`emergency medicine due to its effects on the car-
`diovascular system. In accordance with the Uni-
`ted States Pharmacopeia (USP), the injections
`contain an aqueous solution of L- adrenaline (as a
`bitartrate salt) that is several-fold more potent
`than its optical isomer. 1
`
`Correspondence to: Ilana Schumacher (Phone: +972-3-
`7374142; Fax: +972-3-7376867; E-mail: schumil@bezegint.net)
`Journal of Pharmaceutical Sciences, Vol. 93, 969-980 (2004)
`© 2004 Wiley-Liss, Inc. and the American Pharmacists Association
`
`Adrenaline in solution is subject to degradation;
`therefore, numerous studies addressed the effect
`of formulation variables on the drug inactivation
`kinetics, and attempts have been made to improve
`the formulation stability. 2-9 The results of these
`studies indicate that L- adrenaline in solution is
`inactivated by racemization and oxidation or to
`reaction with auxiliary formulation components
`(e.g., sodium metabisulfite) employed as an anti-
`oxidant (Fig. 1). The products of these reactions
`(including D- adrenaline and adrenaline sulfonate)
`possess little or no pharmacological activity com-
`pared with the parent compound. 6,10 The re-
`versibility of reactions involved in L- adrenaline
`degradation should be taken into account for
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004 (cid:9)
`
`969
`
`ADAMIS EXHIBIT 1005
`
`Page 969
`
`
`
`970 STEPENSKY ET AL.
`
`OH (cid:9)
`
`H O
`
`HO (cid:9)
`
`N H ~C H racemization H O (cid:9)
`3
`
`! (cid:9)
`
`* (cid:9)
`OH (cid:9)
`
`NH
`2
`
`CH 3
`
`/ (cid:9)
`
`L-adrenaline (cid:9)
`
`HO (cid:9)
`
`/ D-adrenaline
`
`sU
`
`o~a~io~
`
`oxidation (cid:9)
`
`HO (cid:9)
`
`HO (cid:9)
`
`OH (cid:9)
`
`N
`
`leucoadrenochrome ft CH3
`
`0 (cid:9)
`
`OH (cid:9)
`
`S0 3-
`HO (cid:9) \ (cid:9)
`
`NH2
`
`CH 3
`
`HO
`
`L-or D-adrenaline sulfonate
`
`HO \ (cid:9)
`
`/0
`
`0 / N (cid:9)
`
`adrenochrome C H 3 (cid:9)
`
`4 (cid:9)
`
`HO
`
`adrenolotin
`
`N
`
`CH3
`
`Figure 1. Degradation reactions of L-adrenaline.
`
`long-term stability studies. For instance, D-adre-
`naline, which is formed from L-adrenaline by the
`racemization process, degrades by racemization
`(to produce L-adrenaline isomer), bisulfite addi-
`tion, and oxidation reactions.
`Despite the obvious significance of L-adrenaline
`optical isomerization in the overall drug inactiva-
`tion process, the USP assay for adrenaline injec-
`tions does not provide quantification of the optical
`isomers in the formulation." Moreover, most of
`the adrenaline injections stability data available
`in the literature were obtained using compara-
`tively nonspecific colorimetric, fluorimetric, or
`bioassay techniques that do not allow for accurate
`determination of the pharmacologically active
`drug isomer. The conclusive evaluation of the com-
`parative contribution and kinetics of the drug
`degradation pathways requires well-character-
`ized analytical methods that provide enantiomeric
`separation and reliable quantification. Examples
`of such methods based on chiral liquid chromato-
`graphy have recently been published. 8° 12,13
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`The objective of this research was to examine in
`a prospective study the kinetics of the different
`pathways of L-adrenaline degradation in commer-
`cially available preparations acquired by Israel
`Defense Forces (IDF), and determine the extent of
`the drug inactivation therein during and beyond
`the stipulated product shelf-life. The analysis
`was conducted using a chiral high-performance
`liquid chromatographic (HPLC) method with
`ultraviolet—visible (UV—vis) detection, and the
`degradation products were identified by mass
`spectrometry (MS-MS).
`
`EXPERIMENTAL
`
`Samples of Adrenaline Injections
`
`Adrenaline injections (1 mg of adrenaline base per
`milliliter) were produced by manufacturer A
`(Teva Ltd, Israel; 28 batches, manufactured
`1985-1998) and manufacturer B (Biogal Ltd.,
`Hungary; 14 batches, manufactured 1998-2002)
`
`ADAMIS EXHIBIT 1005
`
`Page 970
`
`(cid:9)
`(cid:9)
`
`
`LONG-TERM STABILITY OF L-ADRENALINE INJECTIONS
`
`971
`
`using a similar preparation process, except for a
`higher minimal initial content of adrenaline start-
`ing from 1997. The minimal initial drug amount
`in the formulation was changed from 90 to 103%
`of the declared content, corresponding to 0.90 and
`1.03 mg/mL adrenaline base, respectively, be-
`cause of concerns regarding the limited stability
`of the preparations. The additional components of
`the formulation were 1 mg/mL of sodium metabi-
`sulfite, 8 mg/mL of sodium chloride, and water for
`injection.
`All the studied samples were received shortly
`(1-2 months) after the production of the corre-
`sponding batch. Following receipt, all batches
`were stored under controlled conditions recom-
`mended by the manufacturers. On the date of
`analysis, the storage period was 5.8-17.3 years
`for the batches produced by manufacturer A
`and 0.2-5.4 years for the batches produced by
`manufacturer B.
`
`Chemicals and Reagents
`
`L-Adrenaline bitartrate and potassium chloride
`were from Sigma (St. Louis, MO). Glacial acetic
`acid and HPLC-grade acetonitrile was from J.T.
`Baker (Deventer, Holland). Water was purified
`with a tandem RiOs (reverse osmosis)/Milli-Q
`Gradient A-10 system (Millipore, Molsheim,
`France). All other chemicals used in this study
`were of analytical or HPLC grade.
`
`Chiral HPLC Assay
`
`The chiral assay was a modification of the method
`provided by Showa Denko K.K., Tokyo, Japan.
`The mobile phase used for the chiral HPLC assay
`was 0.2 M potassium chloride in water: 0.2 M
`potassium chloride and 0.4% (v/v) acetic acid in
`water: acetonitrile (96:1:3, v/v). The mobile phase
`was filtered under vacuum through 0.45-µm nylon
`filters (Millipore, Bedford, MA).
`The chromatographic system consisted of an
`HPLC model HP 1100 (Hewlett Packard, Palo
`Alto, CA) interfaced to an HP ChemStation, and a
`chiral Shodex ODS 5 .tm column, 150 x 4.6 mm
`(Showa Denko K.K., Tokyo, Japan). The volume of
`injection was 50 µL, the column temperature was
`10°C, the flow rate was 0.7 mL/min, and the run
`time was 20 min. The mobile phase was deaerated
`by on-line degasser, and detection was performed
`by UV—vis photodiode-array detector at 280 nm.
`A stock solution of L-adrenaline bitartrate, ob-
`tained by dissolving 21.84 mg of the compound in
`10.0 mL of water, was further diluted in the mobile
`
`phase to prepare calibration standards in the 10-
`120% nominal drug concentration range.
`Samples for analysis were prepared by diluting
`1.0 mL of adrenaline injection solution with the
`stipulated content 1.8 mg/mL of adrenaline bitar-
`trate (equivalent to 1 mg/mL of adrenaline base)
`with mobile phase to 20 mL.
`
`HPLC-MS-MS Assay
`
`A non-chiral HPLC method applying volatile
`mobile phase was developed for identification of
`the degradation products in adrenaline injections.
`The mobile phase was ammonium acetate buffer
`(5 mM, pH 7.0)/acetonitrile/formic acid (89.03:
`10:0.07, v/v/v) pumped at a flow rate 1.0 mL/min.
`The HPLC-MS-MS system consisted of Thermo
`Separation Products HPLC system (Egelsbach,
`Germany), Millipore Solvent delivery system
`(Millipore Corp., Milford, MA), MS-MS detector
`(Micromass Ltd., UK), UV detector (HPLC detec-
`tor 432, Kontron Instruments, Switzerland), and 5
`µm Luna Phenyl-Hexyl HPLC column (250 x 4.6
`mm; Phenomenex, Inc., Torrance, CA).
`Samples of aged adrenaline batches were
`prepared for analysis by diluting 1.0 mL of the
`adrenaline injection solution with mobile phase to
`20 mL, and 20 µL of the obtained solution was
`injected into HPLC-MS-MS system.
`After the HPLC column, the eluent was split to
`obtain the flow rate of —0.25 mL/min and was
`modified by continuous injection of 0.65% formic
`acid in acetonitrile at a rate of 25 µL/min. The UV
`detector was set to 280 nm. HPLC-MS-MS was
`performed using electrospray ionization (EI) in the
`positive ion mode, with nitrogen as the nebulizer
`and drying gas. The mass range was 50-500 amu,
`and the dwell time was 0.1 s. The molecular ion
`masses of the degradation product and adrenaline
`were identified, and daughter scans of m /z = 248
`(degradation product) and m /z = 184 (adrenaline)
`were measured with the collision energies of 11
`and 18 eV.
`
`pH Measurements
`
`The pH values of the studied samples were
`determined using Metrohm Titroprocessor (model
`796, Herisau, Switzerland) with a combined.,glass
`electrode.
`
`Determination of Aluminum Concentrations
`
`Aluminum concentrations in the samples of
`adrenaline batches were determined by furnace
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`ADAMIS EXHIBIT 1005
`
`Page 971
`
`
`
`972
`
`STEPENSKY ET AL.
`
`atomic absorption analysis using Analyst 300 ap-
`paratus (Perkin Elmer, Norwalk, CT) following
`dilution 1:100 with 0.2% nitric acid. The analysis
`was performed versus standard solutions at
`309:3 nm applying the recommended pretreat-
`ment and atomization conditions. 14
`
`Analysis of Degradation Kinetics of Adrenaline
`
`Exponential Regression
`
`Data analysis by exponential regression and
`calculation of 95% confidence intervals was
`performed according to the method described by
`Zar. 15
`
`square function. 16 The variance was described b y
`the linear model: Var R = (a + b • R)2 , where a and
`b are the variance parameters. Goodness of fit for
`the individual model was assessed from the graphs
`of the predicted and observed data, the coefficients
`of variation of the resulting parameters, and
`the values of the Akaike and Schwartz criteria. 17
`The modeling was performed separately for the
`batches produced before and after 1997 (see
`Results), and for each period, the four sets of data
`(content of L- adrenaline, D- adrenaline, L- adrena-
`line sulfonate, and D-adrenaline sulfonate in the
`injections) were fitted simultaneously.
`
`Modeling
`
`RESULTS
`
`The goal of modeling was identification of the
`most parsimonious model that could appropri-
`ately describe the experimental outcomes. The
`structure of the proposed models was based on
`available data concerning adrenaline degradation
`pathways in injections (see Discussion), 6 and
`models with different kinetic order of underlying
`chemical reactions have been studied.
`Analysis of the content versus storage period
`data was performed with ADAPT II Pharmacoki-
`netic/Pharmacodynamic Systems Analysis Soft-
`ware (Biomedical Simulations Resource, Los
`Angeles, CA) applying the generalized least-
`
`The Chiral HPLC Method and Its Validation
`
`The chiral assay applied in this study enabled
`quantification of the optical isomers of ad-
`renaline and its degradant in injectable prepara-
`tions. The chromatograms of aged formulations
`obtained by this method (see Fig. 2) show four
`peaks corresponding to L- adrenaline (t =15.0 min),
`D- adrenaline (t = 16.3 min), and two degradation
`products (t = 3.4 and 3.6 min, respectively).
`The number of theoretical plates obtained for
`the L- adrenaline, D-adrenaline, and L- and D-
`degradation products were 6966, 4695, 3510, and
`
`Cr"
`
`i
`
`50
`
`IE
`
`Time, min
`
`Figure 2. Chromatogram of aged adrenaline injection.
`
`14 (cid:9)
`
`26 (cid:9)
`
`18
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`ADAMIS EXHIBIT 1005
`
`Page 972
`
`
`
`LONG-TERM STABILITY OF L-ADRENALINE INJECTIONS
`
`973
`
`2988, respectively. The corresponding peak sym-
`metry factors were 0.743, 1.01, 0.911, and 0.687,
`respectively.
`The assay exhibited high linearity, precision,
`and accuracy. Linearity of the chiral assay was
`determined in the concentration range 9.1-
`109.2 µg L- adrenaline bitartrate per milliliter,
`corresponding to 10-120% of the nominal concen-
`tration of the diluted samples. The detector
`response expressed as the peak area was linear
`with concentration. The correlation coefficient (r 2),
`intercept, and slope of the calibration curve were
`0.9999, -4.12 mAU.s, and 32.5 mAU-s•(µg/mL) -1 ,
`respectively. The detection and quantitation limits
`were 0.12; and 0.40% of the nominal drug concen-
`tration, respectively.
`Accuracy was assessed by testing solutions of
`L- adrenaline bitartrate prepared in duplicate at
`100% of the nominal concentration. The solutions
`were processed according to the sample prepara-
`tion method prior to their analysis. The recovery
`was 100.6±0.1%.
`Selectivity was assessed by chromatography of
`standard, aged, and stressed samples of adrena-
`line injections. Aged samples revealed only four
`peaks corresponding to the optical isomers of
`adrenaline and its degradation product. These
`peaks were resolved to the baseline from addi-
`tional compounds appearing in samples subjected
`to forced degradation, and the selectivity of the
`separation was confirmed by assessment of cor-
`responding peak purity factors. The shapes of
`the UV-vis absorption spectra for the isomers of
`adrenaline and degradation product were similar,
`indicating possible similarity in the chemical
`structure.
`
`Identification of the Degradation Product
`
`The degradation products were identified by
`HPLC-MS-MS, applying a nonchiral separation
`method. The degradation product eluted as a
`single peak (t = 2.9 min), and its mass spectrum
`(Fig. 3, panel A) revealed molecular ions
`m/z = 248, 265, and 270, which are attributed to
`adrenaline sulfonate (M + H +, M + NH4 , and
`M + Na', respectively) that exist in the formula-
`tions as two optical isomers. Adrenaline eluted as
`a single peak (t = 4.8 min) with its mass spectrum
`showing ml z = 184 and 166, which are attributed
`to the molecular ions M' + H+, and M' + H+-H2O,
`respectively (data not shown). The daughter
`spectra of the degradation product at m/z 248
`were obtained with different collision energies
`
`(see Fig. 3, panels B and C), and showed major
`peak at m/z of 166, which was attributed to
`M + H+-H2SO4 .
`
`Degradation Kinetics of Adrenaline
`
`The chiral method was applied for estimation
`of the long-term stability of adrenaline injections.
`The amounts of the isomers of adrenaline and
`the degradation product were determined using
`L-adrenaline as a reference assuming that the
`relative absorption of L- and D- adrenaline sulfo-
`nate at 280 nm is identical to that of L- adrenaline.
`This assumption was substantiated by nearly
`identical light absorption spectra of new and aged
`batches with close initial drug contents (data not
`shown) and by mass balance calculations: the
`total observed substance in the formulations (sum
`of L- and D- adrenaline and L- and D- adrenaline
`sulfonate expressed as their stoichiometric equi-
`valents to the drug) remained constant with the
`storage period and was -100 and 105% in batches
`produced before and after 1997, respectively (see
`Fig. 4). The results of the study show that the
`content of L-adrenaline in the formulation rapidly
`decreased with storage (Fig. 4). The content of
`D- adrenaline increased during the first 8-10 years
`of storage and decreased afterwards. The content
`of L-adrenaline sulfonate was similar to that of
`D- adrenaline sulfonate in all batches and in-
`creased with storage.
`The measured pH values of the studied samples
`are presented in Figure 5. The pH values of all the
`studied samples were in the 3.25-3.70 range and
`remained unchanged with storage.
`Aluminum concentrations in the tested batches
`increased in a linear fashion from 0 to -35 µg/L
`after 17-years of storage period (see Fig. 6). The
`aluminum level increased at a similar rate in the
`batches obtained from Manufacturers A and B,
`without significant differences in the slopes and
`intercepts of the respective linear functions.
`
`Exponential Regression of Adrenaline
`Content versus Time Data
`
`The change in the contents of L-adrenaline and
`L- and D- adrenaline together in the injections was
`appropriately described by a simple exponential
`regression (see Fig. 7 and Table 1). The difference
`in the initial drug content for the batches
`produced before and after 1997 was reflected by
`differences in the intercept of the respective
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`ADAMIS EXHIBIT 1005
`
`Page 973
`
`
`
`974 STEPENSKY ET AL.
`
`M+H+ (cid:9)
`
`248.0
`
`5O3H
`HO \ (cid:9)
`
`NH,
`~CH 3
`
`HO
`
`M+NH4+
`
`M+Na+
`
`270.0
`
`G
`
`265.1
`
`1
`
`260 (cid:9)
`
`270 (cid:9)
`
`280 (cid:9)
`
`290
`
`m/z
`
`M+H+'H2SO4
`
`
`
`XZHO (cid:9)
`
`\
`
`NH2
`
`CH 3
`
`HO
`
`248.0
`
`60 (cid:9)
`
`80 100 120 140 160 180 200 220 240
`
`m/z
`
`100 (cid:9)
`
`166.2,
`
`C
`
`n
`
`57.9
`
`107.7 (cid:9)
`
`151.5
`
`P (cid:9)
`
`248.4
`
`60 (cid:9)
`
`80 100 120 140 160 180 200 220 240 260 280 300 320 340 ' 360 360 400 420 440 460 480 500
`
`/z
`
`Figure 3. ESI mass spectra of adrenaline degradation product (panel A) and its
`daughter scans at m/z = 248, with collision energies of 11 and 18 eV (panels B and C,
`respectively).
`
`exponential decay equations (e.g., the intercept
`of 109.4+4.3 versus 99.5±5.2 for L+D-ad-
`renaline; see Figure 7, panel B and Table 1).
`For each time period, narrow confidence inter-
`vals of the drug degradation temporal profiles
`were obtained, indicating low interbatch vari-
`ability. Exponential regression resulted in lower
`coefficients of variation for the initial contents
`compared to those of the degradation rate con-
`stants (3.9-5.9 versus 19-22%, respectively; see
`Table 1).
`
`Modeling
`
`The structure of the model that was selected to
`describe the degradation of adrenaline is pre-
`sented in Figure 8 (panel A), and the underlying
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`differential equations are as follows:
`
`dXl/dt = k1X2 + k5X4 + k6X3 — (h 1 + k 2 + k3)X1
`(1)
`
`dX2/dt = k1Xl + k5X3 + k6X4 — (k 1 + k2 + k3)X2
`(2)
`
`dX3/dt = k2X1 + k3X2 + k4X4 — (k4 + k5 + k6)X3
`(3)
`
`dX4/dt = k2X1 + k3X1 + k4X3 — (k4 + k5 + k6)X4
`(4)
`where X1 , X2, X3 , and X4 are content of L-
`adrenaline, D-adrenaline, L-adrenaline sulfonate,
`and D-adrenaline sulfonate in the injections,
`
`ADAMIS EXHIBIT 1005
`
`Page 974
`
`
`
`120
`
`¤ •¤
`
`100
`
`80
`
`
`
`I 60
`
`40 (cid:9)
`
`20 (cid:9)
`
`0 (cid:9)
`
`0 (cid:9)
`
`LL
`
`2 (cid:9)
`
`LONG-TERM STABILITY OF L-ADRENALINE INJECTIONS 975
`
`•
`•p j ~ • ® ~ • (cid:9)
`
`•••• •• •% •• • ••
`
`•• •
`
`¤ •..
`
`or-bP
`
`9 a 9 99 9
`
`8080 ' a
`
`• ¤ M
`
`
`
`• ••
`
`it
`
`4 (cid:9)
`
`6 (cid:9)
`
`8 (cid:9)
`
`10 (cid:9)
`
`12 (cid:9)
`
`14 (cid:9)
`
`16
`
`18
`
`Time, years
`
`• L-adrenaline (cid:9)
`• Total substance (cid:9)
`o L-adrenaline sulfonate o D-adrenaline sulfonate
`
`D-adrenaline
`
`Figure 4. The content of L- and n-adrenaline and L- and n-adrenaline sulfonate
`(expressed as their stoichiometric equivalents to adrenaline) in injections as a function of
`storage period. Total substance was calculated as the sum of L- and n-adrenaline and L-
`and D-adrenaline sulfonate contents in each batch. The dashed line indicates the time
`point of the change in the minimal initial drug amount (see Experimental).
`
`respectively (% of the stipulated content). The
`rate constants are k l for adrenaline racemization,
`k2 for adrenaline sulfonation without change of
`conformation, k 3 for adrenaline sulfonation with
`change of conformation, k 4 for adrenaline sulfo-
`nate racemization, k5 for conversion of adrenaline
`sulfonate to adrenaline with change of con-
`
`3.75
`
`3.70
`
`3.65
`
`3.60
`
`3.55 .
`
`4Y (cid:9)
`
`~ (cid:9)
`
`• (cid:9)
`
`•
`
`3.45 (cid:9)
`
`3.40
`
`3.35 (cid:9)
`
`3.30 (cid:9)
`
`3.25
`
`•
`
`• •
`
`• . (cid:9)
`
`•
`
`0 (cid:9)
`
`2 (cid:9)
`
`4 (cid:9)
`
`6 (cid:9)
`
`8 (cid:9)
`
`10 (cid:9)
`
`12 (cid:9)
`
`14 (cid:9)
`
`16 (cid:9)
`
`18
`
`Time, years
`
`formation, and k 6 for conversion of adrenaline
`sulfonate to adrenaline without change of con-
`formation (see Fig. 8). All the degradation and
`racemization reactions were assumed to occur
`according to first-order kinetic processes. The
`
`y= 1.869x + 1.972
`R2 -0789
`
`
`
`40
`
`35
`
`30
`
`25
`
`t 20
`S
`
`15
`
`10
`
`5
`
`0
`
`¤ '
`0 (cid:9)
`
`2 (cid:9)
`
`4 (cid:9)
`
`6 (cid:9)
`
`8 (cid:9)
`
`10 (cid:9)
`
`12 (cid:9)
`
`14 (cid:9)
`
`16 (cid:9)
`
`18
`
`Time, years
`
`Figure 5. The pH values of the injections as a
`function of storage period.
`
`Figure 6. The concentration of aluminum in injec-
`tions as a function of storage period.
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`ADAMIS EXHIBIT 1005
`
`Page 975
`
`(cid:9)
`(cid:9)
`(cid:9)
`(cid:9)
`
`
`976
`
`STEPENSKY ET AL.
`
`120
`
`±00
`
`80
`
`so
`
`40
`
`20
`
`ca
`
`O
`
`04
`0
`
`120
`
`o (cid:9)
`
`100
`
`80
`
`60
`
`40 (cid:9)
`
`20 (cid:9)
`
`0
`
`0 (cid:9)
`
`L
`
`o.099x
`
`y = 109.4e'
`R 2 = 0.962
`
`0
`
`5 (cid:9)
`
`m
`o ,
`
`y = 99.5e °'loex
`R 2 = 0.959
`
`0
`
`10 (cid:9)
`Time, years
`
`15
`
`5 (cid:9)
`
`10 (cid:9)
`
`15
`
`Time, years
`
`Figure 7. The content of L-adrenaline (panel A) and sum of L- and n-adrenaline
`(panel B) in injections as a function of storage time. The circles are the observed data,
`the thick lines are the mean estimated content according to the exponential regression,
`and the dotted lines are the 95% confidence intervals.
`
`constants were assumed to be equal for L- and
`D-isomers.
`Attempts to model the results as a single data
`set were unsuccessful and led to a significant
`discrepancy with the actual findings. Therefore,
`the data sets of the batches produced before and
`after 1997 were modeled separately. According to
`the results of modeling, the rates of the reactions
`with k 4, k 5 , and k 6 rate constants (see Fig. 8,
`panel A) were negligible compared with the other
`reaction rates. This result allowed application of
`a reduced model (see Fig. 8, panel B) that was
`capable of adequately describing the content
`change with storage period for all analyzed sub-
`stances (see Fig. 9). The estimated kinetic param-
`eters for the two sets of adrenaline samples are
`presented in Table 2. The values of the Akaike and
`
`Schwartz criteria for the batches produced before
`1997 were 445 and 455, respectively, and for the
`batches produced after 1997 were 314 and 323,
`respectively. Low values of the Akaike and
`Schwartz criteria indicate close fits of the experi-
`mental data by the applied model. 17
`For short storage times, the kinetic parameters
`provided by the exponential regression analysis
`agreed well with the more inclusive kinetic
`models developed in this study, but revealed consi-
`derable discrepancies at the longer storage periods.
`For instance, the initial amount of L- adrenaline
`calculated from the exponential decay model in
`the batches produced after 1997 (110.1±4.3%,
`see Table 1) was in accordance with the corres-
`ponding values estimated from the reduced model
`(107.4 ± 0.6%, see Table 2). In contrast, the values
`
`Table 1. Kinetic Parameters of Exponential Regression: Initial Content and the Degradation Rate Constant (k) of
`L-Adrenaline and the Sum of L- and n-Adrenaline in Injections Produced Before and After 1997
`
`Isomer (cid:9)
`
`L-Adrenaline (cid:9)
`
`L- and D-Adrenaline (cid:9)
`
`Required Initial
`Drug Content
`Parameter
`
`Initial content, %
`k, year -1
`Initial content, %
`k, year -1
`
`90 -115% (Manufacturing
`Dates: 1985-1997)
`
`103-115% (Manufacturing
`Dates: 1997-2002)
`
`Estimate (cid:9)
`
`88.8
`0.135
`99.5
`0.105
`
`SD
`
`5.3
`0.029
`5.2
`0.023
`
`%CV
`
`Estimate (cid:9)
`
`5.9
`21
`5.2
`22
`
`110.1 (cid:9)
`0.132 (cid:9)
`109.4 (cid:9)
`0.099 (cid:9)
`
`SD
`
`4.3
`0.026
`4.3
`0.021
`
`%CV
`
`3.9
`19
`3.9
`21
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`ADAMIS EXHIBIT 1005
`
`Page 976
`
`
`
`LONG-TERM STABILITY OF L-ADRENALINE INJECTIONS
`
`977
`
`L-adrenaline
`
`I" (cid:9)
`
`L-adrenaline
`sulfonate
`
`L-adrenaline (cid:9)
`
`k,
`
`L-adrenaline
`
`1:1 (cid:9)
`
`k1
`
`k4 (cid:9)
`
`I.4
`
`k1 (cid:9)
`
`k1
`
`D-adrenaline (cid:9)
`
`k6 (cid:9)
`Figure 8. The full (panel A) and reduced (panel B) models of adrenaline degrdation
`pathways in injections. The reactions are assumed to have first-order kinetics (see eqs. 1-4).
`
`- D-adrenaline
`sulfonate
`
`D-adrenaline
`
`D-adrenaline
`sulfonate
`
`provided by the exponential and reduced models for
`the batches produced before 1997 were as different
`as 88.8 ± 5.3 versus 102.3 f 0.5%, respectively.
`
`DISCUSSION
`
`A chiral stability-indicating HPLC method was
`applied to study the kinetics of L- adrenaline de-
`gradation under controlled storage conditions
`and to determine the extent of drug degradation
`by the different pathways during and beyond the
`stipulated product shelf-life. Compared with
`the previous studies that did not detect optical
`isomers of adrenaline and utilized complicated
`sample preparation techniques, 2-9 ' 12 application
`of the chiral method provided a simple and re-
`liable procedure for selective determination of
`L- adrenaline and the degradation products in the
`tested samples.
`Rapid oxidation of L- adrenaline in aqueous
`solutions to strongly colored, pharmacologically
`inactive adrenochrome and adrenolotin 6 neces-
`sitated use of auxiliary compounds in the inject-
`able formulations. Several studies addressed
`
`the potential drug-stabilizing effects of sodium
`metabisulfite, EDTA, ascorbic acid, boric acid,
`acetylcysteine, and other substances, 7 ' 9 as well as
`optimized conditions of the preparation and
`packaging of the adrenaline solutions. Combina-
`tions of several approaches (i.e., use of sodium
`metabisulfite as an antioxidant, removal of oxygen
`from the ampules by packing under nitrogen, and
`keeping the pH in the 3.0-3.8 range 2,5,6,8) effec-
`tively prevented the oxidative drug inactivation.
`Use of metabisulfite, although leading to forma-
`tion of pharmacologically inactive adrenaline
`sulfonate, remains a common method for stabiliza-
`tion of adrenaline solutions because the rate of
`bisulfite addition is normally low compared with
`the rate of the drug oxidation. 18
`Based on mass balance considerations, we
`conclude that in the metabisulfite-stabilized com-
`mercial preparations examined in this study, the
`only pathways of L- adrenaline degradation were
`racemization and sulfonation (see Fig. 1) and that
`there was no significant degradation to additional
`products that were not detected by the chiral
`HPLC assay method used herein. Insignificance of
`additional degradation pathways (e.g., oxidation)
`
`Table 2. Kinetic Parameters of Racemization and Sulfonation of Adrenaline Estimated using the Reduced Model
`(see Fig. 8, panel B) in Injections Produced Before and After 1997
`
`Required Initial Drug
`Content Par ameter
`ki (year (cid:9)
`k2 (year (cid:9)
`1)
`k3 (year')
`Initial content of L-adrenaline (%)
`
`1)
`
`90-115% (Manufacturing
`Dates: 1985-1997)
`
`103-115% (Manufacturing
`Dates: 1997-2002)
`
`Estimate
`
`SD
`
`0.0575
`0.0527
`0.0544
`102.3
`
`0.0017
`0.0005
`0.0005
`0.5
`
`%CV
`
`3.00
`0.918
`0.915
`0.442
`
`Estimate (cid:9)
`
`SD
`
`0.0334 (cid:9)
`0.0458 (cid:9)
`0.0469 (cid:9)
`107.4 (cid:9)
`
`0.0008
`0.0008
`0.0008
`0.6
`
`%CV
`
`2.36
`1.68
`1.67
`0.605
`
`JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 93, NO. 4, APRIL 2004
`
`ADAMIS EXHIBIT 1005
`
`Page 977
`
`
`
`978 STEPENSKY ET AL.
`
`120
`
`•
`
`80
`
`U
`
`~
`
`+~ (cid:9)
`
`40
`
`0
`
`0 (cid:9)
`
`15
`
`10
`
`5
`
`0
`
`0 (cid:9)
`
`c7
`Q (cid:9)
`
`O
`
`1000 (cid:9)
`
`100
`
`10
`
`A (cid:9)
`
`, ~ (cid:9)
`
`•-
`
`50
`
`40
`
`30
`
`tic +•
`
`C
`
`- (cid:9)
`
`0 (cid:9)
`
`5
`
`10 (cid:9)
`
`15 (cid:9)
`
`20
`
`5 (cid:9)
`
`10
`
`15
`
`Time, years
`
`•.
`
`•..__
`
`0
`
`50
`
`40
`
`•_• (cid:9)
`
`30
`
`20
`
`10
`
`0
`
`0 (cid:9)
`
`5 (cid:9)
`
`10
`
`15
`
`Time, years
`
`10 (cid:9)
`
`15
`
`Time, years
`
`5 (cid:9)
`
`10 (cid:9)
`
`15
`
`Time, years
`
`Figure 9. The content of L-adrenaline (panel A), n-adrenaline (panel B), L-adrenaline
`sulfonate (panel C), and n- adrenaline sulfonate (panel D) in injections as a function of
`storage period. The circles are the observed data, and the lines are the best fits according
`to the reduced model described in Figure 8. The modeling was performed separately for
`two data sets corresponding to batches produced before and after 1997 (see text for
`details). The insert is the semilogarithmic presentation of panel A.
`
`was supported by absence of additional peaks on
`the chromatograms on one hand and the purity of
`the observed substances confirmed by MS exam-
`ination on the other hand. The drug exhibited
`rapid degradation by reactions with first-order
`kinetics in adrenaline when stored under recom-
`mended conditions. In the batches produced after
`the application of the new requirement for the
`minimal drug content, the lowest acceptable limit
`(90% as a sum