Pharmaceutical Development and Technology, 7(1), 1–32 (2002)
`
`REVIEW ARTICLE
`
`Stabilization of Pharmaceuticals to
`Oxidative Degradation
`
`Kenneth C. Waterman,* Roger C. Adami,
`Karen M. Alsante, Jinyang Hong, Margaret S. Landis,
`Franco Lombardo, and Christopher J. Roberts
`
`Pfizer Global Research and Development, Eastern Point Road,
`Groton, CT 06340
`
`ABSTRACT
`
`A guide for stabilization of pharmaceuticals to oxidation is presented. Literature is
`presented with an attempt to be a ready source for data and recommendations for
`formulators. Liquid and solid dosage forms are discussed with options including
`formulation changes, additives, and packaging documented. In particular, selection
`of and methods for use of antioxidants are discussed including recommended levels.
`
`INTRODUCTION
`
`Scope
`
`This review article sets the stage for a pharmaceutical
`formulator to deal with the problem of drug-product
`chemical instability. In particular, this article focuses on
`one of the more common modes of degradation in drug
`products, namely oxidation. Methods are suggested for
`establishing that oxidation is indeed the problem, and
`what the particular oxidative pathway is for degradation.
`Although each new drug presents unique challenges,
`guidance is provided for resolving this problem based on
`the best information currently available.
`This review is organized to provide information on
`recognizing and predicting drug oxidation in dosage
`
`forms. Relevant oxidation mechanisms and various
`remedies are discussed for the different dosage forms
`including traditional
`liquid and solid dosage forms
`containing small molecules, as well as formulations
`containing protein and DNA-based pharmaceuticals.
`This review also provides a decision tree for addressing
`oxidative degradation, along with a detailed table of
`antioxidant additives and their commercial precedence.
`Other approaches discussed include packaging, counter-
`ions and pH, and mitigation of impurities.
`
`Formal Oxidation (Recognition of Oxidation)
`
`Oxidation is defined broadly as the loss of electrons
`from a molecule (increase in oxidation number). For
`organic molecules, this can be restated as an increase in
`
`*Corresponding author. Fax: (860) 441-3972; E-mail: ken_waterman@groton.pfizer.com
`
`Copyright q 2002 by Marcel Dekker, Inc.
`
`www.dekker.com
`
`1
`
`Eton Ex. 1027
`1 of 32
`
`

`

`rm §1
`
`H+H H
`
`1
`
`2
`
`3
`
`4
`
`alkene
`I
`
`alkyne
`2
`
`Oxidation
`Increase in oxygen content
`Decrease in hydrogen content
`
`l
`
`RCH,OH
`
`[OJ
`
`[HJ
`
`RCH,
`
`t
`
`Reduction
`Decrease in oxygen content
`Increase in hydrogen content
`
`-
`
`1--C'-'-
`, ,,
`--y
`I alkane
`I O
`- i-CH20R
`
`ether (alcohol for R=H)
`
`1
`
`/C=CH2 - -C=CH
`"
`-i-~0
`- -i-(; - i-}-o,
`
`ketone
`(aldehyde for R=H)
`2
`
`ester
`(carboxylic acid
`for R=H)
`3
`
`R
`
`O
`carbonate ester
`(carbonic acid for R=H)
`4
`
`~
`
`1
`i ·c:
`~
`li
`M
`
`I @
`J
`MARCEL DEKKER, INC. n
`3
`
`270 Madison Avenue, New York, New York 10016
`
`•
`
`ORDER REPRINTS
`
`2
`
`Waterman et al.
`
`oxygen or decrease in hydrogen content (1). Alterna-
`tively, oxidation can be defined as a reaction that
`increases the content of more electronegative atoms in a
`molecule (2). With organic systems, these electronega-
`tive heteroatoms are generally oxygen or halogens.
`
`structures shown below:
`
`Listed below are the general structures and names of
`some common pharmaceutically relevant species in the
`order of increasing oxidation states with the oxidation
`numbers listed below each species:
`
`When a compound is oxidized, another compound
`must be reduced. Hydration and dehydration are not
`oxidation/reduction reactions, though they add oxygen,
`since the reaction is essentially an internal oxidation and
`reduction: one carbon atom is oxidized while another is
`reduced. The net change to the oxidation state of the
`molecule is therefore zero. In the example below,
`ethylene is hydrated to ethanol. While the carbon on the
`left gains a hydrogen atom and is therefore effectively
`reduced, the carbon on the right gains an oxygen atom
`and is effectively oxidized. The net change to the
`molecule is zero.
`
`H2C ¼ CH2
`
`H2Oÿ! CH3CH2OH
`
`If there is any doubt as to the oxidation status of a
`molecule, the following procedure helps to identify the
`oxidation state of a given compound (3):
`
`1.
`
`Imagine a water molecule being added onto all
`unsaturated bonds in that molecule. If there is a
`ring in the molecule, open the ring with a water
`molecule.
`2. Count the number of heteroatoms in the water
`addition product. This is the oxidation number of
`that molecule.
`3. By comparing the oxidation numbers for reactants
`and products, one can determine if the reaction
`represents an oxidation.
`
`The italicized numbers in the following scheme
`represent the corresponding oxidation numbers for the
`
`is
`it
`To study an oxidation reaction mechanism,
`important to understand the electron transfer between the
`species involved by tracking the fate of electrons (using
`Lewis structures and arrows). For single-electron
`transfers, single-headed arrows are used, while for two-
`electron processes, double-headed arrows are used. As an
`example of tracking an electron transfer, disproportiona-
`tion of two radicals is shown with arrows indicating the
`“flow” of electrons:
`
`Preliminary Screening for Oxidative Instability
`
`Solid Dosage Forms
`
`There is currently no best practice purposeful
`degradation protocol to study oxidative degradation in
`solid-state drug substances or drug products. As a result,
`solid-state reactivity of the drug substances towards
`oxidation in drug products is explored most readily using
`excipient compatibility screening under thermal/humid-
`ity challenge conditions. Although there has been some
`discussion on using ternary mixtures of drug and
`excipients (with statistical design) (4), currently, binary
`mixtures of drug compound and tableting excipients
`(binders, fillers, disintegrants, etc.) are used commonly in
`preliminary screening experiments. Samples are pre-
`pared using standard mortar and pestle geometric
`dilution techniques (i.e., over a wide range of drug-to-
`
`Eton Ex. 1027
`2 of 32
`
`

`

`rm
`~
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`ORDER REPRINTS
`
`Oxidation Stabilization
`
`3
`
`excipient ratios). The mechanical forces involved in using
`the mortar and pestle are hoped to mimic some of the
`forces involved in milling and tableting, which often lead
`to amorphous or disordered drug regions (see “Oxidation
`in Solid Dosage Forms”). These blends are stored under
`different temperature and humidity conditions (indicated
`typically 58C,
`by the relative humidity, RH),
`room
`temperature/ambient humidity, 408C/closed bottle,
`408C/75% RH open bottle, 508C/closed bottle and
`508C/20% RH open bottle. These studies provide an
`indication of the oxidative instability of a drug in a binary
`mixture where the pure drug substance may be quite
`stable. Such studies help to identify particular excipients
`that may need to be avoided. Unfortunately, these blend
`stability studies do not, in general, predict rates for the
`system after granulation, milling, and tableting, and can
`therefore only be used qualitatively.
`If oxidative instability is suspected, studies can be
`performed to determine if molecular oxygen is involved in
`the oxidation process. Stability challenges that involve
`filling the headspace of vials with nitrogen (negative
`control) and pure O2 (positive control) are often useful to
`determine if molecular oxygen is involved in the
`degradation. If the headspace environments have little or
`no effect on the oxidative reactivity of the drug substance
`in these stability screenings, molecular oxygen may be
`involved in the reaction but not
`in the rate-limiting
`oxidation step, or the level of oxygen may still be
`sufficiently high that the reaction readily occurs. More
`thorough removal of oxygen in the headspace can be
`accomplished by a freeze – pump – thaw cycle. Alterna-
`tively, oxidation may be related to highly reactive
`impurities (peroxides, superoxides, hypochlorites, formic
`acid) present in the excipients as manufacturing-related
`impurities (see “Detecting and Controlling Impurities”).
`Such drug oxidative instability will often show itself in the
`form of greater decomposition rates for more dilute drug
`mixtures. Pre-treatment of the excipients with heat and
`radical scavengers such as nitric oxide or benzoquinone
`may also be helpful in implicating these impurities as the
`source of oxidative instability. While these techniques
`may be helpful in determining the causes and mechanisms
`of a drug oxidation, they have not been adapted to use in
`the actual solid dosage form stabilization.
`
`Liquid Dosage Forms
`
`Liquid dosage form oxidative stability screening
`generally involves examining the drug stability under a
`
`number of conditions. These conditions will vary
`depending on the type of dosage form (oral, parenteral,
`etc.) and limitations due to solubility and decomposition
`pathways competitive with oxidation. Among the factors
`to evaluate for a new drug candidate are the following:
`
`1. Acidity. The pH can impact the oxidative stability
`of ionizable drugs (see “Acidity and pH Effects”).
`2. Concentration. Excipient
`impurities will have
`more relative impact on dilute solutions than on
`concentrated solutions.
`3. Temperature. Although Arrhenius behavior
`is
`seldom observed, an indication of oxidative
`stability is found by examining the reactivity over
`the range 5 – 708C.
`4. Oxygen in the headspace. Both oxygen enriched
`and inert atmosphere samples can provide an
`indication of the tendency for a drug to oxidize in a
`particular formulation.
`5. Photo-oxidation. The stability of the drug in the
`presence of light and oxygen can be important. In
`addition, the use of added sensitizers (such as rose
`bengal) in examining the photostability of the drug
`can help determine whether
`the mechanism
`involves singlet oxygen.
`6. Metals. Addition of metal ions to solutions can
`indicate whether there is a tendency for the drug to
`be catalytically oxidized (see “Catalysis”). Typi-
`cally, FeCl3 or CuCl2 are added at levels less than
`100 ppm.
`7. Packaging. For parenteral dosage forms, a range of
`stoppers should be examined. For oral dosage
`forms, both plastic and glass bottles should be
`evaluated (see “Packaging/Liquid Dosage Forms”).
`
`General Issues
`
`It is difficult to use the Arrhenius equation to describe
`oxidative instability, or accelerated screening methods in
`general to predict room temperature shelf-life. This can be
`due to the following factors:
`
`1.
`
`Instability may be related to the amount of a
`peroxide impurity present in the particular lot of
`the excipient
`(see “Detecting and Controlling
`Impurities”). This could be correlated with the
`manufacturing processes involved, age of
`the
`excipient, and the conditions (temperature, humid-
`ity, sunlight) under which the material was stored.
`Oxidation rates can be very rapid at early time
`
`Eton Ex. 1027
`3 of 32
`
`

`

`roil
`~
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`ORDER REPRINTS
`
`4
`
`Waterman et al.
`
`2.
`
`points when peroxide impurities are plentiful and
`plateau or fall off after the impurity is consumed
`by the degradation reaction.
`If the generation of radicals is rate limiting, the
`kinetics can show autocatalysis; i.e., the rate of
`drug degradation increases as the radical concen-
`tration increases.
`3. There can be more than one mechanism involved
`in the degradation with the differences in
`activation parameters similar enough that tem-
`perature changes can essentially lead to different
`mechanisms dominating.
`In the solid state, temperature-sensitive properties
`of the drug product such as percent amorphous
`content, degree of hydration of the components,
`and molecular mobility may affect the oxidative
`degradation.
`5. The permeability of oxygen through packaging is
`temperature dependent.
`6. The solubility of oxygen in excipients (solvents)
`is inversely temperature dependent.
`
`4.
`
`Detection and identification of oxidative degradants is
`aided by the use of mass spectrometry. Table 1 gives
`some characteristic mass peaks that may be associated
`with a drug substance that has undergone some type of
`oxidative reaction. Often,
`tandem liquid chromato-
`graphy-mass spectrometry/mass (LC-MS/MS) techniques
`can be used in identifying specific sites of oxidation
`within complex molecules.
`
`Predicting Oxidation
`
`Purposeful Degradation (5,6)
`
`Oxidative studies executed to force drug substance
`degradation are useful
`to predict primary oxidative
`
`degradants in drug products. Published stability guide-
`lines (7) suggest
`the use of a concentrated oxygen
`atmosphere to generate oxidative degradation products
`for chromatographic identification. Since degradants can
`arise from reaction of the drug product with molecular
`oxygen (see “Autoxidation—Chain Processes” and
`“Electron Transfer”) or with oxidizing agents present
`in the formulation (usually peroxides, see “Peroxides and
`Other Oxidizing Agents”), it is important to conduct
`purposeful degradation studies with oxygen as well as
`hydrogen peroxide. Hydrogen peroxide is often non-
`predictive of molecular oxygen reactions (8) because it
`does not involve the radical chain process common with
`the oxygen-based reactions (see “Autoxidation—Chain
`Processes”). Hydrogen peroxide stress testing is useful in
`drug-product studies where hydrogen peroxide itself is
`an expected impurity in an excipient (see “Formation and
`Presence of Oxidants in Excipients” and “Detecting and
`Controlling Impurities”).
`For an oxygen-atmosphere purposeful degradation
`study, a solvent must be chosen that solubilizes the
`drug sufficiently (1 – 10 mg/mL) and ideally mimics
`the proposed formulation. Although ideally one would
`use protic
`solvents
`to mimic
`common protic
`excipients,
`this is complicated by the tendency for
`alcohols to slow oxidation reactions by competing
`with the drug for initiator radicals (9). For this reason,
`the polar aprotic solvent acetonitrile is often used
`in place of alcohols in model studies. A co-solvent
`may be necessary to achieve a sufficiently high
`concentration.
`Radical
`initiators can be effective in accelerating
`autoxidation (see “Initiation”), thereby allowing for easier
`characterization (10 – 12). Although electron-transfer (see
`“Peroxides and Other Oxidizing Agents”) rather than free-
`
`Possible Products of Drug Oxidation Based on Mass Spectral Data
`
`Table 1
`
`Mass Spectral Analyses
`
`Possible Products
`
`Parent 2 2
`Parent þ 14
`Parent þ 15
`Parent þ 16
`Parent þ 32
`
`Oxidation of – CH(OH) – to – CO – , primary or secondary amine to imine
`Oxidation of – CH to – C(O) –
`Oxidation of secondary amine to N-oxide
`Hydroxylation or conversion of – C – H to – C – OH; epoxidation of a double
`bond; sulfide to sulfoxide conversion; tertiary amine to N-oxide
`Hydro- or endoperoxide formation; sulfide to sulfone conversion
`
`Eton Ex. 1027
`4 of 32
`
`

`

`roil
`~
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`ORDER REPRINTS
`
`Oxidation Stabilization
`
`5
`
`radical propagation may dominate the mechanism in the
`dosage form, in most cases the products (though not their
`relative distributions) are the same in either case. Because
`the addition of initiators allows for generation of 10 – 20%
`degradation product within 10 days (using 1 – 10 mol%
`radical initiator in the presence of pressurized oxygen), it is
`generally advantageous to use this approach along with
`appropriate controls.
`To perform an autoxidative degradation study, the
`drug substance is dissolved in an appropriate solvent and
`transferred to a reaction vessel pressurized at 50 – 300 psi
`O2 to increase the oxygen concentration in solution, and
`heated to form radicals from the initiator (see Table 2 for
`a list of typical initiators).
`In carrying out a purposeful oxidative degradation
`study, it is critical to run the appropriate controls in order
`to get a better mechanistic understanding of whether the
`degradation results from a thermal,
`free-radical, or
`nonfree-radical process. Controls for these experiments
`include the following:
`
`1.
`2.
`
`3.
`
`4.
`
`drug with oxygen without initiator;
`drug with initiator purged of oxygen (with
`nitrogen or argon);
`drug without initiator purged of oxygen (at the
`reaction temperature); and
`initiator at appropriate level without drug
`substance to determine if any observed high
`pressure liquid chromatography (HPLC) peaks
`result from oxidation products that are not drug
`substance related.
`
`Redox Potential
`
`As discussed in “Formal Oxidation (Recognition of
`Oxidation)”, oxidation involves the (formal) loss of
`
`electrons. A compound’s oxidation potential (or redox
`potential) gives its thermodynamic tendency to lose an
`electron. A classical way to determine the redox potential
`and provide some kinetics for the decomposition of the
`oxidized species, involves the use of cyclic voltammetry
`(CV). Sweep rates of several thousand volts per second
`are possible with adequate instrumentation, which offers
`the opportunity of clocking the lifetime of radicals
`generated by oxidation, even for very fast processes (13).
`Because of the preparation and analysis time involved,
`CV techniques are more appropriate for detailed
`mechanistic investigations rather than for fast prelimi-
`nary screening of drug candidates.
`A comparison based on the use of known standards
`(antioxidants, drugs, and general organic compounds)
`can be used to assess the relative ease with which a
`compound undergoes electron transfer. This allows a
`prediction of stability based on a compound’s redox
`potential and the relative stability of standards with
`similar potentials. A compound, which by comparison to
`known oxidatively labile compounds yields a low
`oxidative potential (more easily oxidized), is likely to
`be prone to oxidative degradation. In Table 3 are
`tabulated some redox potentials of reference compounds
`and rough potentials of some oxidatively labile groups
`(the lower the potential, the more readily the species will
`be oxidized). Molecules that are generally stable to
`electron-transfer oxidation could still be unstable to
`oxidation by hydrogen-atom abstraction (see “Propa-
`gation”). These data allow for general rules for predicting
`electron-transfer based oxidative stability:
`
`1.
`
`2.
`
`oxidation potential $ 1300 mV: stable to electron-
`transfer oxidation;
`oxidation potential between 850 and 1300 mV:
`depends on specific drug and formulation; and
`
`Table 2
`
`Free-Radical Initiators Useful in Purposeful Oxidative Degradation Studies (from Waco Pure Chemical Industries)
`
`Initiator
`
`Chemical Abstract Service
`(CAS) #
`
`Temperature for
`10 hr T1/2 (8C)
`
`Solubility
`(mg/mL)
`
`0
`0
`0
`0
`0
`0
`
`-Azobis(N,N0
`-dimethyleneisobutyramidine)dihydrochloride
`-Azobis(4-cyanopentanoic acid)
`-Azobis(2-amidinopropane)dihydrochloride
`-Azobis(2-methyl-N-hydroxymethyl) propionamide))
`-Azobisisobutyronitrile (AIBN)
`-Azobis(2,4-dimethylvaleronitrile)
`
`2,2
`4,4
`2,2
`2,2
`2,2
`2,2
`
`27776-21-2
`61630-29-3
`2997-92-4
`61551-69-7
`927-83-3
`52406-55-0
`
`44 (H2O)
`69 (H2O)
`56 (H2O)
`86 (H2O)
`65 (toluene)
`51 (toluene)
`
`35.2 (H2O)
`1 (H2O)
`23.2 (H2O)
`2.4 (H2O)
`7.5 (MeOH)
`22 (MeOH)
`
`Eton Ex. 1027
`5 of 32
`
`

`

`roil
`~
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`ORDER REPRINTS
`
`6
`
`Waterman et al.
`
`Table 3
`
`Selected Potentials for Antioxidants and Drugs
`
`Compound
`
`Oxidizable Functional Group
`
`Oxidative Peak Potential
`
`Observed Stability Towards Oxidation
`
`Dextromethorphan
`Amitriptyline
`Fluoxetine
`Flurazepam
`Captopril
`Chlorpromazine
`Morphine
`L-Ascorbic acid
`Vitamin E
`Phenylbutazone
`Tetracyline(s)
`
`Tertiary amine
`Tertiary amine
`Secondary amine, benzylic ether
`Tertiary amine, imine
`Thiol
`Thioether
`Allylic alcohol, phenol
`Allylic alcohol
`Phenol
`3,5-Dioxopirazolidine
`Phenol, enols, tertiary amine
`
`1400 (14)
`1300 (15)
`1300 (15)
`900 (15)
`, 900 (19)
`840 (15)
`820 (21)
`700 (21)
`700 (21)
`660 (22)
`550 (23)
`
`Stable (9)
`Stable (16)
`Stable (17)
`Stable (18)
`Unstable (20)
`Unstable (20)
`Unstable (20)
`Unstable (20)
`Unstable (20)
`Unstable (9)
`Unstable (20,24)
`
`þ
`The potentials were obtained under different conditions, and are intended only as illustrative examples. All potentials are in mV vs. Ag/Ag
`electrode.
`
`reference
`
`3.
`
`oxidation potential # 850 mV: easily oxidized.
`
`Identifying Oxidative Reactivity Based on Structure
`
`Autoxidation
`
`Assuming autoxidation of a drug molecule occurs by
`an initial abstraction of a labile hydrogen atom followed
`by reaction with molecular oxygen (see “Autoxidation—
`Chain Processes”), oxidative degradation should be
`linked to the lability of hydrogen atoms within the
`molecular framework. The lability of hydrogen atoms
`within a molecular structure is described by the
`corresponding bond-dissociation energies;
`i.e.,
`the
`lower the bond energy, the more likely a hydrogen
`atom will be abstracted by a radical. Table 4 shows the
`representative bond-dissociation energies of some
`common functional moieties.
`A qualitative way to estimate hydrogen atom lability
`is to evaluate the stability of the radical that is formed.
`In general, the more stable the resulting radical, the
`easier for the hydrogen atom to be abstracted from the
`corresponding position on the molecule. The rules of
`radical stability are similar to the rules that govern
`reactive center stability (i.e., cations and anions). In
`general, for sp3 hybridized centers,
`tertiary centered
`radicals are more stable than secondary centered
`radicals which are more stable than primary radicals
`so that the corresponding hydrogen atom abstraction
`will occur most readily from tertiary hydrogens, and
`
`least readily from primary hydrogens. This stability
`order is supported by the bond-dissociation energies in
`Table 4, where the C – H bond energy for ethane is less
`than the C – H bond energy for methane. With sp2 and
`
`Table 4
`
`Representative Hydrogen Bond-Dissociation
`Energies for Some Common Structural
`Moieties (25,26)
`
`Species
`
`CH3 – H
`CH3CH2 – H
`vCH – H
`CH2
`CH2vCHCH2 – H
`PhCH2 – H
`H2N – H
`CH3NH – H
`CH3O – H
`PhNH – H
`HO – H
`HOCH2 – H
`HOCH(CH3) – H
`CNCH2 – H
`HO – OH
`CH3O – OCH3
`HOO – H
`
`Bond-Dissociation
`Energy (kcal/mol)
`
`104
`98
`103
`85
`85
`103
`92
`102
`80
`119
`93
`90
`86
`51
`36
`90
`
`Eton Ex. 1027
`6 of 32
`
`

`

`rm]
`~
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`
`
`HO lw /
`
`nitrogen can undergo
`electron transfer oxidation
`I
`
`adjacent to
`heteroatom
`
`-
`+ H CH3
`H zN~OC
`HO ~ /;
`::::.....
`N
`I
`HHH HHH
`\
`\
`' b 1-
`phenolic position can
`enzy 1c
`J
`d
`benzylic and tertiary and
`un ergo e ectron
`adjacent to
`adjacent to
`transfer
`oxidation and_ hydrogen heteroatom
`heteroatom
`atom abstractJon
`
`ORDER REPRINTS
`
`Oxidation Stabilization
`
`7
`
`is
`radicals
`stabilization of
`sp hybridized centers,
`generally poor due to the orthogonal arrangement of
`the p-orbitals with the unpaired electron. Referring to
`Table 4,
`the hydrogen bond energy for ethylene is
`similar to that of methane.
`Particularly stable radicals are formed from hydrogen
`atom abstraction of allylic or benzylic positions. These
`methylene/methine positions adjacent
`to aromatic or
`vinylic systems are very susceptible to hydrogen-atom
`abstraction because the resulting radical centers are
`stabilized through delocalization (see Table 4). Simi-
`larly, hydrogen atom abstraction from carbons adjacent
`to heteroatoms (nitrogen or oxygen) is facile due to
`electronic stabilization of the product.
`
`Other Oxidation Mechanisms
`
`Oxidation can also occur via an electron-transfer
`reaction (see “Electron Transfer”) to form reactive
`radical anions or cations. Nitrogen (amines), sulfur
`heteroatoms (sulfides, disulfides, and sulfoxides), and
`oxygen-based anions (phenol anions) are common sites
`for electron-transfer induced oxidation, producing final
`products such as N-oxides, sulfoxides, sulfones, and
`ketones. These sites can also act as nucleophiles to react
`with peroxide impurities in the formulation (see
`“Peroxide Reactions with Drugs”) giving similar
`products.
`
`Example
`
`As an example of using these rules to predict reactive
`centers, Labetalol is shown below with the potentially
`labile hydrogen atoms and other oxidation sites
`indicated:
`
`Computational Modeling
`
`Computational prediction of degradation can be
`extremely helpful in the understanding of degradation
`
`mechanisms and the characterization of degradation
`products found by purposeful degradation studies (see
`“Purposeful Degradation”). Several commercial compu-
`ter programs designed to predict reaction products are
`available including elaboration of reactions for the
`organic syntheses (EROS)
`(27), workbench for
`the
`organization of data for chemical applications (WODCA)
`(28), and logic and heuristics applied to synthetic analysis
`(LHASA) (29). Since these programs are parameterized
`around the solution chemistry reactions, they may be less
`predictive for reactivity of drug molecules in the solid
`state. The EPIWIN (estimation program interface for
`Windowse) (30,31) is designed to predict the reactivity
`of organic molecules towards hydroxyl radicals (HOz).
`This approach can be useful in estimating the reactivity of
`a given molecule towards hydrogen abstraction; how-
`ever, its use in predicting actual degradation products and
`reaction rates has not been reported. One program found
`to be particularly useful is computer assisted mechanistic
`evaluation of organic reactions (CAMEO)
`(32). The
`CAMEO program computationally predicts the products of
`organic reactions given the starting materials, reagents,
`and conditions including the oxidative conditions.
`Solvent variation (protic/aprotic) and temperature
`changes can be considered, as well as more than one
`equivalent of reagents. The CAMEO degradation predic-
`tion results can be used as an initial guide to possible
`decay products. The CAMEO predicted decay products
`may not in reality be observed during the stability studies,
`and some degradants can be formed,
`that are not
`predicted in the calculations. In general, however, the
`CAMEO algorithms have been biased to predict more
`products than are likely to be observed. It is also likely
`that certain products predicted can undergo further
`decomposition. For example, primary and secondary
`hydroperoxides predicted in many oxidation processes
`typically undergo further reactions in actual degradation
`studies to produce ketones and alcohols. Although not
`reported in the literature, CAMEO allows for a standard
`oxidative set of reagents and conditions to be created and
`used routinely to predict oxidation products. This
`approach would allow for consistency for all compounds
`being evaluated. The reagents and conditions used do not
`necessarily need to mimic the real stability conditions,
`but rather could be predictive of those conditions. In spite
`of CAMEO’s limitations (only available for Macintoshes,
`limited to 64 total atoms), future adaptations promise to
`make this and other computational methods more
`routinely useful for pharmaceutical scientists.
`
`Eton Ex. 1027
`7 of 32
`
`

`

`roil
`~
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`ORDER REPRINTS
`
`8
`
`Waterman et al.
`
`MECHANISMS OF OXIDATION
`
`Autoxidation—Chain Processes
`
`The free-radical process of autoxidation consists of a
`chain sequence involving three distinct
`types of
`reactions: initiation, propagation, and termination.
`
`Initiation
`
`Initiation produces a free radical to begin a chain
`reaction (11,33,34). Initiation can occur due to homolytic
`cleavage of a weak bond such as with a polymeric
`excipient, by electron-transfer processes (see “Electron
`Transfer”), by light-induced reactions, or by metal
`catalysis (see “Catalysis”).
`In the solid state, special issues exist with regard to
`initiation. In particular, homolytic bond cleavage to
`give a pair of radicals (caged radicals) will generally
`not lead to free radicals since cage escape is strongly
`dependent on the diffusional mobility within a solid
`matrix. Because of this, initiation is slow to nonexistent
`in the solid state, though any radicals generated tend to
`survive for long periods since termination is slow. At
`this point,
`it
`remains unknown how much drug
`oxidation in the solid state is a product of chain
`processes vs. reaction of drug with oxidant impurities
`(see “Peroxides and Other Oxidizing Agents”) and
`mobile active oxygen species (hydroxyl, peroxyl, and
`superoxide) generated by metal catalysis.
`
`Propagation (11,34)
`
`Once a free radical is produced, its fate can include
`propagation by hydrogen-atom abstraction from the drug
`or excipient (solvent), propagation by a radical addition
`reaction to the drug or excipient, propagation by reaction
`with molecular oxygen to form a peroxyl
`radical,
`rearrangement, cyclization, or termination (see “Termin-
`ation”). The peroxyl radical itself can propagate through
`the same types of reactions.
`Propagation:
`Inz þ RH ! InH þ Rz
`Rz þ O2 ! ROOz
`Rz þ RH ! HR – Rz
`ROOz þ RH ! ROOH þ Rz
`
`Though hydrogen-atom abstraction from the drug is
`formally an oxidation process, it is not an autoxidation
`process since there is no oxygen in the reaction path.
`When oxygen is absent, the drug radical will reform the
`drug by abstraction of a hydrogen atom from other
`molecules in the system. When oxygen is involved,
`however,
`the hydroperoxide product rarely loses an
`oxygen molecule, such that irreversible drug decompo-
`sition occurs. In solution, reaction of a radical with
`molecular oxygen typically occurs with no activation
`barrier;
`i.e.,
`the reaction occurs at
`the diffusion-
`controlled rate of approximately 109 M 21sec21 depend-
`ing on the solvent viscosity (35,36).
`In solution,
`under normal atmospheric conditions, the rate-limiting
`step is hydrogen-atom abstraction,
`therefore the
`resulting rate expression is determined by the rate of
`radical
`formation (initiation), propagation (typically
`104 – 1027 M 21 sec21) (37,38) and termination, but is
`not dependent on the concentration of oxygen in solution.
`This is shown in the following rate expression for chain
`oxidation (37,38):
`kiÿ!Inz þ Inz
`In – In
`Inz þ R – Hfastÿ!Rz þ InH
`Rz þ Rz fastÿ!R – R typically 109 – 1010 M 21 sec21
`fastÿ!R – OOz
`Rz þ O2
`typically 108 – 109 M 21 sec21
`
`R – OOz þ H – R kpÿ!R – OOH þ Rz
`
`typically 1 – 102 M 21 sec21
`
`R – OOz þ R – OOz ktÿ!R – OOH þ R
`ðRef:ð38ÞÞ
`
`typically 106 – 109 M 21 sec21
`
` 
`1=2½RHŠ
`
`d½RHŠ
`dt
`
`2
`
`¼ kp
`
`ki
`kt
`
`One of the by-products of this chain process is a
`hydroperoxide of either the drug or an excipient, which
`itself can act as an oxidant for the drug (see “Peroxides
`and Other Oxidizing Agents”). With oxygen present,
`the radical propagation (with oxygen consumption) can
`lead to large turnovers where every step creates a
`degradation product of the drug. Eventually, the build-
`
`Eton Ex. 1027
`8 of 32
`
`

`

`rm]
`~
`
`OH I
`
`H2C:?"' -
`
`#CH
`
`+
`
`0
`
`II
`
`/CH
`H3C
`
`R + R----- R-R
`Termination: R + ROD· ------ R-OOR
`/
`ROD· + ROD·-
`ROO-OOR
`R + R _ . , . disproportionatio~
`
`l + o, + Ho-b--
`
`H
`
`/
`
`\
`'--.
`(R= pri mary or secondary)
`ROOR + 0 2
`(R= teni ary)
`
`-
`
`H" ::=/H
`,..,..-N-.-"-: +HOO'
`R
`Ar
`
`j
`I (Q)
`i ·c t
`MARCEL DEKKER, INC. ~
`270 Madison Avenue, New York, New York I 0016 ~
`
`ORDER REPRINTS
`
`Oxidation Stabilization
`
`9
`
`up of hydroperoxides can lead to changes in the
`reaction mechanism, where the hydroperoxide com-
`petes with the drug in the hydrogen-atom abstraction
`step.
`In the solid state, propagation is hindered by low
`mobility. Reactions therefore tend to occur over
`extremely short distances with fewer chain propagation
`steps. Alternatively, solid-state oxidation can be
`propagated through diffusable species such as peroxide
`and superoxide. Although the products from these
`processes are often the same as for the chain process
`described above for solutions, the mechanism and rates
`in the solid state can be significantly different (see
`“Oxidation in Solid Dosage Forms”). Another impli-
`cation of having fewer propagation steps in the solid state
`compared to the solution state is that antioxidants
`designed to intercept the chain (see “Chain Termin-
`ators”) may be considerably less effective in the solid
`state. The low mobility of the solid state requires that an
`antioxidant be dispersed into each small domain to
`quench the reaction effectively.
`
`Termination (11,34)
`
`Termination of propagating radicals occurs on
`combination of two radicals to form nonradical products.
`This can manifest itself in radical combination reactions
`where two radicals form a new bond joining them
`together or by disproportionation where one radical is
`reduced while the other is oxidized. This latter process
`typically involves