ARTICLE IN PRESS
`
`Applied Radiation and Isotopes 67 (2009) 88–94
`
`Contents lists available at ScienceDirect
`
`Applied Radiation and Isotopes
`
`journal homepage: www.elsevier.com/locate/apradiso
`
`Studies into radiolytic decomposition of fluorine-18 labeled
`radiopharmaceuticals for positron emission tomography
`
`Peter J.H. Scott a, Brian G. Hockley a, Hank F. Kung b, Rajesh Manchanda c, Wei Zhang c,
`Michael R. Kilbourn a,
`
`a Department of Radiology, University of Michigan Medical School, Ann Arbor, MI, USA
`b Department of Radiology, University of Pennsylvania, Philadelphia, PA, USA
`c Avid Radiopharmaceuticals, Inc., Philadelphia, PA, USA
`
`a r t i c l e i n f o
`
`a b s t r a c t
`
`Radiolytic decomposition of high specific concentration radiopharmaceuticals is an undesired side-
`effect that can hamper development of novel PET tracers. This was particularly evident in a series of
`carbon-11 and fluorine-18 labeled mono- and dimethyl-substituted aryl amines, where rapid
`decomposition was observed in isolation and formulation steps. We tested a number of additives that
`inhibit radiolysis and can be safely added to the synthesis procedures (purification and isolation) and
`reformulation steps to provide suitable clinical formulations. Ethanol and sodium ascorbate are
`established anti-oxidant stabilizers that completely inhibit radiolytic decomposition and are amenable
`to human use. Herein, we also demonstrate for the first time that nitrones are non-toxic radical
`scavengers that are capable of inhibiting radiolysis.
`
`& 2008 Elsevier Ltd. All rights reserved.
`
`Article history:
`Received 4 December 2007
`Received in revised form
`21 August 2008
`Accepted 26 August 2008
`
`Keywords:
`Tomography
`Emission computed
`Radiolysis
`Stability
`Anti-oxidants
`
`1.
`
`Introduction
`
`Positron emission tomography (PET) imaging is a rapidly
`growing field of research in which molecules labeled with short-
`lived radionuclides such as carbon-11 or fluorine-18 are utilized to
`non-invasively examine biochemistry in living human subjects.
`As part of a broad program to deliver not only established PET
`radiopharmaceuticals for clinical care (2-deoxy-2-[18F]fluoro-
`D-glucose, [18F]FDG) and for clinical research (carbon-11 and
`fluorine-18 labeled radiotracers for brain, heart, pancreas and
`tumor imaging) we have a research program dedicated to the
`development and clinical
`implementation of new PET radio-
`pharmaceuticals. In particular, we and others have significant
`interest in the development of new fluorine-18 labeled PET
`imaging agents for neurodegenerative diseases such as Alzhei-
`mer’s disease (AD) and Parkinson’s disease (PD). The application
`of fluorine-18, with a convenient 110 min half-life, will facilitate
`the more widespread distribution and use of such radiopharma-
`ceuticals in clinical populations. During the development of
`radiochemical syntheses of such new fluorine-18 labeled radio-
`pharmaceuticals, including the validation of methods to prepare
`suitably high amounts for potential distribution, we observed a
`
`
`
`Corresponding author.
`E-mail address: mkilbour@umich.edu (M.R. Kilbourn).
`
`0969-8043/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
`doi:10.1016/j.apradiso.2008.08.015
`
`distressing occurrence of decomposition of the final formulated
`products. As we suspected this might be due to radiolysis, and
`that it was particularly acute for the types of structures (N-
`methylanilines) common in our recently developed radiotracers,
`we undertook and report here a limited study of the factors
`involved in this decomposition process, examining what chemical
`structures seem particularly susceptible and what steps can be
`taken to effectively prevent decomposition. Radiolysis in radio-
`pharmaceutical preparations is certainly not an unknown phe-
`nomenon, having been implicated originally in the decomposition
`of carbon-14 and hydrogen-3 labeled species (Bayly and Evans,
`1966) and more recently of carbon-11 labeled species (Suzuki
`et al., 1990; Bogni et al., 2003; Fukumura et al., 2003, 2004a, b).
`These studies suggest that the degree of radiolysis of labeled
`compounds depends on the level of radioactivity, the level of
`specific activity, the structure of the radiopharmaceutical and the
`position of the radiolabel. These hypotheses are all supported by
`the results of the current study.
`In contrast to the above studies, radiolytic decomposition has not
`been extensively studied for fluorine-18 labeled compounds,
`although many of the mechanisms are expected to be similar to
`those discussed by Fukumura et al., (2004a, b). One example, a recent
`study of the stability of [18F]fluorodeoxyglucose (FDG), (Fawdry,
`2007) reported a slow but steady decomposition of the radio-
`pharmaceutical to release free [18F]fluoride ion. Similarly, MacGregor
`[18F]-N-methylspiroperidol,
`reported on the decomposition of
`
`Evergreen Ex. 1015
`1 of 7
`
`

`

`ARTICLE IN PRESS
`
`P.J.H. Scott et al. / Applied Radiation and Isotopes 67 (2009) 88–94
`
`89
`
`in which it was shown that rate of radiolysis was proportional to
`specific activity (MacGregor et al., 1987). With the more
`widespread availability of medical cyclotrons capable of produ-
`cing high levels of [18F]fluoride ion coupled with the development
`and implementation of automated synthesis modules, high level
`preparations of fluorine-18 compounds are becoming increasingly
`common. Therefore, the problem of radiolytic decomposition of
`fluorine-18 labeled radiopharmaceuticals needs to be recognized
`and addressed as a part of the overall radiopharmaceutical
`development program.
`
`2. Materials and methods
`
`2.1. General considerations
`
`Fluorine-18 ([18F]fluoride in H2[18O]O) and carbon-11 ([11C]CO2
`in nitrogen gas) radionuclides were generated using a General
`Electric Medical Systems (GEMS) PETtrace cyclotron. Fluorine-18
`labeled radiopharmaceuticals were prepared in a GEMS TRACER-
`lab FX F-N synthesis module and carbon-11 tracers were
`synthesized using a GEMS TRACERlab FX C-Pro synthesis module
`or Bioscan Loop synthesis system. N,N-Dimethylaniline, N-methy-
`laniline, 4-aminophenol and N-tert-butyl-a-phenylnitrone (PBN)
`were purchased from Sigma-Aldrich and used as received. Sodium
`ascorbate (SA) (Cenolates Ascorbic Acid (as SA) Injection, USP,
`500 mg/mL) was purchased from Hospira Inc. and used as
`received. Ethanol (sterile dehydrated alcohol injection, USP) was
`supplied by American Regent Inc. HPLC analysis of radiochemical
`purity was conducted using a Shimadzu LC-2010AHT Liquid
`Chromatograph fitted with UV and Bioscan g-detectors. HPLC
`peaks corresponding to products were identified by co-injection
`with unlabeled reference standards. Gas chromatography (GC)
`analysis used to determine residual levels of organic solvents was
`performed on a Shimadzu GC-2010 Gas Chromatograph.
`
`2.3. Typical carbon-11 radiolabeling procedure
`
`[11C]CO2 (100 GBq) in a nitrogen stream was delivered to the
`TRACERlab FX C-Pro and trapped on molecular sieves. Heating
`(350 1C) under an atmosphere of hydrogen over a nickel catalyst
`reduced the [11C]CO2 to [11C]CH4 and subsequent reaction with
`iodine vapor at 720 1C provided [11C]CH3I. [11C]CH3I was then
`converted to [11C]CH3OTf by passing over silver triflate. Labeling
`reactions were carried out by bubbling methyl triflate through a
`solution of precursor either in the TRACERlab FX C-Pro reactor or
`in a Bioscan Autoloop methylation system.
`In both cases,
`purification by semi-preparative HPLC or Sep-Pak provided
`11C-labeled radiopharmaceuticals as isotonic solutions suitable
`for injection that were released for QC analysis.
`
`2.4. Modified radiopharmaceutical preparation procedures
`incorporating anti-oxidant stabilizers
`
`to radiolytic decomposition
`Radiopharmaceuticals subject
`were prepared following modified versions of
`the general
`procedures described above to allow for incorporation of anti-
`oxidant stabilizers into the manufacturing processes. HPLC
`solvent systems were prepared containing 0.5% w/v (5 g/L) sodium
`ascorbate (SA) (or ascorbic acid depending upon the pH of the
`buffer in question). HPLC fractions were simultaneously collected
`and diluted into aqueous SA (0.5% w/v, 0.25 g in 50 mL sterile
`water). The resulting solution was passed through a C18-light
`Sep-Pak that trapped the product and residual solvents were
`washed away with additional aqueous SA (50 mg in 10 mL sterile
`water). Products were eluted with ethanol (USP, 0.5 mL) and
`diluted with SA (USP, 500 mg/mL, 0.1 mL) in 0.9% saline (USP,
`9.4 mL). This provided formulations (5% v/v ethanol in saline
`containing 0.5% w/v SA) suitable for
`injection that were
`transferred into sterile 10 mL dose vials through Millex 0.22 mm
`sterile filters and released for QC analysis.
`
`2.2. Typical fluorine-18 radiolabeling procedure
`
`2.5. Quality control
`
`[18F]Fluoride in [18O]H2O (37–74 GBq) was delivered to the
`TRACERlab FX F-N synthesis module and collected on a QMA-light
`Sep-Pak. The [18F]fluoride was then eluted from the QMA cartridge
`using aqueous potassium carbonate (3.5 mg in 0.5 mL) and
`transferred into the reactor vessel. Kryptofix-2.2.2 in acetonitrile
`(15 mg in 1 mL) was added and the water–acetonitrile azeotrope
`was evaporated (60 1C for 7 min under vacuum with argon
`stream followed by 120 1C for 5 min under vacuum). Precursors
`(1–1.5 mg) in anhydrous dimethylsulfoxide (DMSO, 1 mL) were
`added and the reaction was heated to 120 1C for 10 min. Following
`labeling, the reactor was cooled to 50 1C and the reaction
`mixture was diluted with HPLC solvent (3 mL). The crude mixture
`was passed through an alumina-light Sep-Pak and purified by
`semi-preparative HPLC. Those compounds purified using HPLC
`solvent systems suitable for injection (e.g., ethanol and water)
`were diluted with saline and transferred into a sterile 10 mL dose
`vial through a Millex 0.22 mm sterile filter.
`For those compounds purified using solvent systems unsui-
`table for
`injection (e.g., acetonitrile–water), HPLC fractions
`containing desired products were collected, diluted with sterile
`water (50 mL), and the resulting solution was passed through a
`C18-light Sep-Pak. Radiopharmaceuticals remained bound to the
`Sep-Pak whilst residual HPLC solvent was washed away with
`further sterile water (10 mL). Products were then eluted with USP
`ethanol (0.5 mL) and diluted with 0.9% sterile saline (9.5 mL). The
`resulting isotonic solution was passed into a sterile 10 mL dose
`vial through a Millex 0.22 mm sterile filter.
`
`All radiopharmaceuticals prepared for human use undergo
`extensive quality control according to the USP guidelines before
`they are released for use. Formulations are analyzed for chemical
`and radiochemical purity (HPLC); specific activity (HPLC);
`pyrogenicity (Charles Rivers); residual Kryptofix-2.2.2 (TLC);
`residual organic solvents (GC); formulation pH; radionuclidic
`half-life; and sterile filter integrity (bubble-point test). Every
`batch must pass all of these tests before release to the clinic is
`approved.
`
`2.6. Determination of rates of radiolytic decomposition
`
`The radiolytic decomposition of fluorine-18 and carbon-11
`labeled products was monitored as a function of radiochemical
`purity through repeated injections of formulated products on the
`analytical HPLC. Decomposition of unlabeled standard com-
`pounds placed either next to or into a vial containing high
`[18F]fluoride ion was monitored by HPLC
`concentrations of
`coupled with UV detection of the mass peak.
`
`3. Results and discussion
`
`3.1. Potential PET ligands for imaging AD pathophysiology
`
`We have been involved in two research projects investigating
`new tracers to image AD pathophysiology. The disease is
`Evergreen Ex. 1015
`2 of 7
`
`

`

`ARTICLE IN PRESS
`
`90
`
`P.J.H. Scott et al. / Applied Radiation and Isotopes 67 (2009) 88–94
`
`characterized by the deposition of b-amyloid in plaques and
`formation of neurofibrillary tangles resulting from the abnormal
`aggregation of tau protein (La Ferla and Green, 2007; Pallas and
`Camins, 2006). The ability to determine levels of plaques and
`tangles early enough to allow effective diagnosis and treatment
`represents a major goal in AD therapy and radionuclide imaging
`techniques may play a key role in meeting this need (Nordberg,
`2004; Cohen, 2007; Cai et al., 2007).
`2-(4-([11C]methylamino)phenyl)benzo[d]thiazol-
`Currently,
`([11C]PIB, 3, Fig. 1)
`is the most commonly employed
`6-ol
`radioligand for imaging amyloid pathology (Mathis et al., 2007).
`However, due to the short 20 min half-life of carbon-11, there is
`significant interest in developing fluorine-18 labeled amyloid
`imaging agents: the 110 min half-life of fluorine-18 is advanta-
`geous because it allows for multiple patients to be scanned
`from a single synthetic preparation and provides for possible
`distribution of radiopharmaceuticals to PET imaging centers that
`do not possess cyclotrons or
`the synthetic radiochemistry
`facilities. Numerous structures of potential fluorine-containing
`amyloid binding agents have now been reported, among them
`2-(1-(6-((2-[18F]fluoroethyl)(methyl)amino)naphthalene-2-yl)
`ethylidene)malononitrile ([18F]FDDNP, 5) (Liu et al., 2007; Small
`et al., 2006), stilbenes related to SB-13 ((E)-4-(4-(dimethylamino)
`styryl)phenol 2) (Ono et al., 2003), and styrylpyridines (e.g., (E)-4-
`(2-(6-(2-(2-(2-[18F]fluoroethoxy)ethoxy)ethoxy)pyridine-3-yl)vinyl)
`-N,N-dimethylbenzenamine (Zhang et al., 2007), ([18F]AV-19, 1,
`Fig. 1). Our interests were in the scale-up of the preparation of this
`last compound, [18F]AV-19, to allow initial clinical trials.
`The second project targets neurofibrillary tangles, resulting
`from abnormal aggregation of tau protein, using substituted
`(e.g., N-[11C]methyl-4-(quinolin-2-yl)benzenamine
`quinolines
`(MQB, 4a) and N,N-[11C]dimethyl-4-(quinolin-2-yl)benzenamine
`
`Fig. 1. Structures of representative N-methyl substituted anilines used as PET
`radiopharmaceuticals.
`
`(DMQB, 4b), Fig. 1) (Okamura et al., 2005; Kudo et al., 2004).
`A common structural element in all of the radiopharmaceuticals
`illustrated in Fig. 1, and seen in many drug candidates, is the
`presence of N-methylamine and N,N-dimethylamine substituents
`on the aryl rings. N-Methylamine groups are hydrogen bond
`donors that are able to coordinate to active sites and also allow
`for metabolism and subsequent elimination of drugs by, for
`example, cytochrome P450 mediated oxidative N-demethylation
`(Sheweita, 2000; Uehleke, 1973; Hlavica, 2002) or amine-oxidase/
`monoamine oxidase driven N-oxidation (Tipton et al., 2004;
`Edmondson et al., 2004). However, despite the benefits of
`incorporating amines into potential tracer molecules, it quickly
`became apparent during lead development studies for both the
`amyloid and tau imaging projects that many potential tracers
`containing methylaniline components are prone to radiolytic
`decomposition.
`
`3.2. Radiolytic decomposition of amino-substituted molecules
`
`[18F]AV-19 1 has been evaluated as an imaging agent for
`quantifying levels of amyloid plaques in the brains of AD patients
`(Skovronsky et al., 2008). During initial development of a
`production method for [18F]AV-19 using a TRACERlab FX F-N,
`trial labeling studies were attempted by reacting a tosylated
`precursor in DMSO with 11 GBq of [18F]fluoride. A method was
`developed which routinely provided 1.5 GBq of [18F]AV-19 in
`485% radiochemical purity at end-of-synthesis (Table 1, Entry 1,
`n ¼ 3). However, the product as formulated in a standard 5%
`ethanol : saline formulation was unstable and found to decom-
`pose into four polar radioactive species (having lower retention
`times in the radiochemical HPLC traces) such that after 20 min
`radiochemical purity had decreased to 73%. This decomposition
`was attributed to radiolysis as the corresponding formulations of
`unlabeled reference standard are known to be chemically stable
`for extended times and at elevated temperatures. Unfortunately,
`and in agreement with previous reports (Fukumura et al.,
`2004a, b, MacGregor et al., 1987), the rate of radiolytic decom-
`position greatly increased when the synthesis was scaled up to
`provide high specific concentration formulations suitable for
`imaging multiple patients from a single dose. Full-scale produc-
`tion using 66.6 GBq of [18F]fluoride gave the expected increase in
`yield of [18F]AV-19 (7.4 GBq) but there was a dramatic decrease in
`radiochemical purity (72% at EOS). Repeated analysis of the
`product by HPLC after end-of-synthesis revealed that rapid
`radiolytic decomposition was occurring to the extent that after
`20 min, radiochemical purity had decreased to 37% and radi-
`olabeled decomposition products were the major components of
`the formulation (Table 1, Entry 2). As per previous reports into
`radiolytic decomposition of radiopharmaceuticals (Suzuki et al.,
`1990; Bogni et al., 2003; Fukumura et al., 2003, 2004a, b; Fawdry,
`2007; MacGregor et al., 1987), this decomposition was thought to
`be due to transient reactive species such as hydroxyl radicals
`generated by radiolysis of water or hydrated electrons. In further
`experiments intended to elucidate the mechanism of decomposi-
`tion, a vial containing a solution of [19F]AV-19 standard was
`placed in close proximity (1 cm away) to the TRACERlab FX F-N
`target vial containing 66.6 GBq of fluoride in an aqueous solution.
`Under these conditions [19F]AV-19 was found to be very stable and
`underwent none of the expected decomposition (Table 1, Entry 3).
`This confirmed that decomposition and/or the free-radicals
`suspected of causing radiolysis were not the result of the high
`concentration of gamma rays. Rather, for radiolysis to occur the
`radioactivity had to be in the same vial as the product and this
`was suggestive of decomposition possibly involving short-range
`positrons. Since the range of a fluorine-18 positron in water is
`Evergreen Ex. 1015
`3 of 7
`
`

`

`ARTICLE IN PRESS
`
`P.J.H. Scott et al. / Applied Radiation and Isotopes 67 (2009) 88–94
`
`91
`
`Table 1
`Time-dependent stabilities of selected PET radioligands. All radiochemical products formulated in 8 ml of 5% ethanol in isotonic saline
`
`Entry
`
`Compound
`
`Conditions
`
`1
`2
`3
`4
`5
`6
`7
`8
`9
`10
`11
`
`[18F]AV-19 (1)
`[18F]AV-19 (1)
`[19F]AV-19 (1)
`[19F]AV-19 (1)
`[11C]PIB (3)
`[12C]PIB (3)
`[12C]PIB (3)
`[11C]MQB (4a)
`[11C]DMQB (4b)
`[12C]DMQB (4b)
`[11C]DASB (6)
`
`1.48 GBq dose
`7.4 GBq dose
`1 cm from vial with 66.6 GBq of [18F]fluoride
`In vial with 66.6 GBq [18F]fluoride
`2.5 GBq dose
`1 cm from vial with 66.6 GBq of [18F]fluoride
`In vial via with 66.6 GBq [18F]fluoride
`1.73 GBq dose
`1.85 GBq dose
`In vial with 11.1 GBq [18F]fluoride
`1.85 GBq dose
`
`Puritya
`
`0 (%)
`
`85
`72
`100
`100
`99
`100
`100
`95
`58
`60
`90
`
`a Radiochemical purity is reported for labeled compounds and UV purity for unlabeled standards.
`b Purity recorded after 1 h to assess long term stability.
`
`10 min (%)
`
`20 min (%)
`
`77
`55
`100
`73
`99
`100
`100
`70
`21
`22
`
`
`73
`37
`100
`
`99
`100
`100
`
`5
`0b
`
`
`2.4 mm and they are unable to penetrate glass (Fowler and Wolf,
`1982), decomposition does not occur when radioactivity is
`isolated in the target vial and separate from the solution of
`[19F]AV-19. As expected, when the [19F]AV-19 standard was
`exposed to 66.6 GBq of [18F]fluoride in solution, rapid decomposi-
`tion occurred (Table 1, Entry 4).
`It is noteworthy that in contrast to the above findings and
`despite similarities in structure, [11C]PIB (3) was very stable in
`high-specific activity formulations prepared from 37 GBq of
`methyl triflate using a GE TRACERlab FX C-Pro and Bioscan
`Autoloop (Table 1, Entry 5). Similarly, when placed into or next to
`a vial containing 66.6 GBq of [18F]fluoride, no decomposition
`occurred (Table 1, Entries 6 and 7). Presumably, the thiazole
`linkage provided formulated [11C]PIB with enhanced stability
`when compared to the alkene linkage found in AV-19.
`Concomitantly with these studies, we have also investigated
`the possibility of radiolabeling quinolines (4) according to a
`literature procedure and using them to image tau pathology in AD.
`Labeling of these compounds was carried out with solutions of the
`corresponding aniline or N-methylaniline and carbon-11 methyl
`triflate (37 GBq)
`in methylethylketone (MEK)
`to provide
`N-methyl-(MQB, 4a) and N,N-dimethyl-labeled (DMQB, 4b)
`tracers, respectively. We were surprised during this work to
`observe similar radiolytic decomposition to that described above
`for the [18F]styrylpyridines. [11C]MQB (4a) was 95% radiochemi-
`cally pure at EOS but this dropped off to 70% after 10 min (Table 1,
`Entry 8). Similar to that found earlier with the fluorine-18 labeled
`compounds,
`the rate of radiolysis was faster for the N,N-
`dimethylamine (DMQB, 4b) than for the N-methylamine (MQB,
`4a). The dimethyl derivative [11C]DMQB (4b) was only 58% pure at
`EOS and by 20 min post-EOS the radiochemical purity had
`dropped to 5% (Table 1, Entry 9). Similar results were obtained
`when unlabeled standards of these compounds were exposed in
`solution to high levels of [18F]fluoride (Table 1, Entry 10). An
`additional drawback is that solutions of
`the corresponding
`unlabelled MQB and DMQB standards are known to be chemically
`unstable, however the process is noticeably slower (weeks to
`months) and is not responsible for the rapid decomposition of the
`labeled species.
`
`3.3. Radiolysis of simple anilines
`
`The results with the amyloid and tau radioligands raised the
`question of whether or not anilines in general are particularly
`prone to radiolytic decomposition. As one potential reaction of
`anilines is the formation of the N-oxide species, that might be
`
`expected to occur with most if not all anilines. Therefore
`three simple anilines (N,N-dimethylaniline, N-methylaniline and
`4-aminophenol) were exposed to high levels of fluorine-18 to test
`this hypothesis. There was no evidence of decomposition when
`any of the three were combined with 66.6 GBq of fluoride-18 in
`aqueous solution, using UV–HPLC analysis after 2 h of exposure. At
`present it is unclear what intrinsic feature of substituted anilines
`may make them more susceptible to radiolysis and further
`investigation is ongoing.
`
`3.4.
`
`Inhibition of radiolysis
`
`All of the potential radiopharmaceuticals described in this note
`have been successfully radiolabeled with carbon-11 or fluorine-
`18. However, to progress any of them into clinical studies required
`inhibition of radiolytic decomposition so that stable formulations
`meeting all of the high radiochemical purity and specific activity
`requirements could be prepared. Therefore we investigated a
`number of means of inhibiting radiolytic decomposition.
`
`3.4.1. Dilution of formulated products
`The radiolytic decomposition of tracers was thought to be
`caused by hydroxyl radicals resulting from the reaction of high
`energy positrons with the aqueous medium. Since the range of a
`fluorine-18 positron in water is 2.4 mm, it followed that radiolytic
`decomposition should be slowed down simply by diluting the
`radiopharmaceutical formulations. Therefore, [18F]AV-19 (7.4 GBq)
`was prepared as described above and formulated in 0.5 mL
`ethanol and 9.5 mL saline. Five mL of this formulation was then
`removed and diluted with additional saline (5 mL) and the rate of
`radiolysis of the parent dose vs. the diluted dose were compared
`(Table 2). Radiolysis occurred in both cases, but decomposition
`was noticeably slower in the diluted formulation, despite the
`lower ethanol concentration. Whilst additional dilution might
`further slow radiolysis occurring during short-term storage of a
`fluorine-18 preparation, large injection volumes would begin to be
`required to obtain levels of
`injected activity necessary for
`successful PET imaging. Furthermore, the initial 60% radiochemi-
`cal purity at
`formulation demonstrated the difficulties in
`preparation of this radiochemical without using some means to
`prevent the decomposition during the synthesis and formulation
`steps.
`
`3.4.2. Addition of anti-oxidant stabilizers
`Since dilution was capable of slowing down radiolysis but not
`completely inhibiting it, incorporation of additional stabilizers
`Evergreen Ex. 1015
`4 of 7
`
`

`

`ARTICLE IN PRESS
`
`92
`
`P.J.H. Scott et al. / Applied Radiation and Isotopes 67 (2009) 88–94
`
`Table 2
`Test of dilution as a method for inhibiting radiolysis of a fluorine-18 labeled N,N-dimethylaniline
`
`Compound
`
`Formulation
`
`Radiochemical purity
`
`1
`
`2
`
`[18F]AV-19 (1)
`
`[18F]AV-19 (1)
`
`3.7 GBq in 5% EtOH/saline
`(5 mL)
`3.7 GBq in 2.5% EtOH/saline
`(10 mL)
`
`60
`
`60
`
`48
`
`62
`
`21
`
`60
`
`5
`
`52
`
`EOS (%)
`
`10 min (%)
`
`20 min (%)
`
`3 h (%)
`
`Table 3
`Applications of antioxidants (EtOH, PBN and sodium ascorbate) to inhibit radiolytic decomposition of PET radiopharmaceuticals containing N-methylaniline structures
`
`Compound
`
`Formulation
`
`Purity
`
`EOS (%)
`
`10 min (%)
`
`20 min (%)
`
`1
`2
`3
`4
`5
`6
`
`[18F]AV-19 (1)
`[18F]AV-19 (1)
`[18F]AV-19 (1)
`[18F]AV-19 (1)
`[12C]DMQB (4b)
`[11C]DASB (6)
`
`50 : 50 EtOH saline
`100% EtOH
`5% EtOH in saline + 0.5% w/v PBN
`5% EtOH in saline + 0.5% w/v Na–ascorbate
`11.1 GBq fluoride added to 5% EtOH in saline + 0.5% w/v Na–ascorbate
`5% EtOH in saline + 0.2% w/v Na–ascorbate
`
`60
`66
`92
`96
`100
`99
`
`62
`66
`93
`96
`100
`99
`
`64
`66
`93
`96
`100
`99
`
`Long-term
`stability
`(%)
`
`66 (3 h)
`–
`93 (1 h)
`100 (6 h)
`100 (1 h)
`99 (1 h)
`
`into radiopharmaceutical formulations was considered (Table 3).
`Working on our hypothesis that radiolysis was attributable to
`reactive oxygen species, such as transient free-radicals, addition of
`anti-oxidant stabilizers was considered. Ethanol is noted for its
`anti-oxidant properties and ability to scavange hydroxyl radicals
`in aqueous solutions. Formulating [18F]AV-19 in a solution of 50%
`ethanol : 50% saline or 100% ethanol was found to completely
`inhibit
`radiolytic decomposition of
`the formulated product
`(Table 3, Entries 1 and 2), although it should be noted that the
`product was isolated in poor radiochemical purity after HPLC
`purification (a method for solving this problem is discussed later).
`These results using ethanol as inhibitor agree with Fukumura’s
`results with carbon-11 labelled species (Fukumura et al., 2003,
`2004a, b) and studies of
`the radiolytic decomposition of
`[18F]FDDNP reported in the literature (Liu et al., 2007; Small
`et al., 2006; Klok et al., 2008). [18F]FDDNP contains a mono-
`methylaniline and is stable in neat ethanol but undergoes rapid
`radiolytic decomposition when the ethanol is diluted with water.
`High concentrations of ethanol do not represent a practical
`solution to the problem of radiolysis as 5–10% is the maximum
`amount of ethanol permitted in isotonic solutions formulated for
`injection. As formulating in 5% ethanol was not sufficient to
`prevent radiolysis in formulated products and also could not be
`employed to prevent radiolytic decomposition occurring during
`synthesis, alternative anti-oxidants were considered. A range of
`compounds are known to inhibit decomposition attributable to
`free radicals including ascorbic acid (Chen et al., 2005; Elmore,
`2005; Liu et al., 2003; Werner et al., 1990), potassium iodide
`(Suzuki et al., 1990), nitrones (Reybier et al., 2006; Green et al.,
`2003) and thiourea (Halliwell and Gutteridge, 2005). Thiourea is
`highly toxic and unsuitable for clinical formulations and so we
`focused our efforts on ascorbic acid and nitrones as non-toxic
`anti-oxidants amenable for human use.
`N-tert-Butyl-a-phenylnitrone (PBN) is a nitrone widely used as
`a means of trapping and detecting free-radicals in both chemistry
`and biology (Halliwell and Gutteridge, 2005), particularly finding
`widespread application in electron spin resonance (ESR) spectro-
`scopy. The ability of PBN to trap reactive oxygen species in vivo has
`also led to extensive investigation as a neuroprotective agent
`(Kotake, 1999; Kim et al., 2007). We have also previously worked
`with PBN and radiolabeled analogs as a means of quantifying
`levels of free-radicals in vivo (Bormans and Kilbourn, 1995). We
`
`were therefore interested in employing PBN to prevent radiolytic
`decomposition by trapping free radicals potentially formed in
`high specific concentration radiopharmaceutical
`formulations.
`[18F]AV-19 (1) was prepared as outlined above but the procedure
`was modified so that the semi-preparative HPLC buffer contained
`PBN (0.1% w/v) and the final product was formulated in 5% ethanol
`in saline containing 0.5% w/v PBN. This provided a high specific
`concentration formulation of [18F]AV-19 which had a radio-
`chemical purity 493%. Incorporation of PBN inhibited radiolytic
`decomposition during synthesis and the high radiochemical
`purity of the formulation remained unchanged for 1 h after end-
`of-synthesis (Table 3, Entry 3). It thus appears that the inclusion of
`a nitrone effectively inhibits radiolytic decomposition; application
`to human studies is feasible as nitrones have been investigated in
`clinical studies as neuroprotective agents in cerebral ischemia
`(Green et al., 2003). This is a preliminary proof-of-concept study
`and further investigations to determine the minimum effective
`level of PBN required to inhibit radiolytic decomposition are
`ongoing and will be reported in due course.
`Ascorbic acid is also a well known anti-oxidant that a number
`of groups have employed to inhibit radiolytic decomposition
`(Fukumura et al., 2004a, b; Chen et al., 2005; Elmore, 2005; Liu
`et al., 2003; Werner et al., 1990). For example, Klok et al., 2008.
`used it to overcome stability problems associated with the
`preparation of [18F]FDDNP (Klok et al., 2008). One disadvantage
`is that addition of ascorbic acid is known to make final
`formulations very acidic (pH2) and outside of physiological
`pH, making them unsuitable for human use without additional
`buffering. However, we have found that SA is a suitable
`alternative. SA has the same excellent anti-oxidant properties as
`the free acid, but its addition to clinical
`formulations has
`negligible affect on pH. However, a drawback of adding SA to
`[18F]-labeled radiopharmaceutical formulations is that it compli-
`cates analysis of residual Kryptofix-2.2.2 in formulations during
`quality control. Kryptofix-2.2.2 is a crown-ether phase transfer
`reagent used to enhance nucleophilic substitution reactions using
`[18F]fluoride. Due to its toxicity, Kryptofix must be removed after
`synthesis and levels in clinical formulations must be quantified
`during quality control. Typically this is achieved using the
`classical iodoplatinate TLC spot-test (Mock et al., 1997). However,
`if formulations containing SA are analyzed using this spot-test
`then bleaching of the iodoplatinate TLC strips occurs and renders
`Evergreen Ex. 1015
`5 of 7
`
`

`

`ARTICLE IN PRESS
`
`P.J.H. Scott et al. / Applied Radiation and Isotopes 67 (2009) 88–94
`
`93
`
`the test subject to false negative results. To overcome this
`problem, we have recently developed an alternative TLC method
`that allows for the analysis of residual Kryptofix in the presence of
`SA (Scott and Kilbourn, 2007).
`The full-scale synthesis of [18F]AV-19 (1) was repeated after
`incorporation of 0.5% w/v SA into the semi-preparative HPLC
`buffer (5 g/L), reformulation flask (250 mg/50 mL water) and
`final formulation (50 mg/10 mL). The result was a high-specific
`concentration formulation of [18F]AV-19 (1) with excellent radio-
`chemical purity (495%, n ¼ 20) that was suitable for injection
`(typical formulation pH ¼ 6.5). As with PBN, SA also inhibited
`radiolysis during and after synthesis. The formulation was stable
`up to 6 h post end-of-synthesis and there was no decrease in
`radiochemical purity as analyzed by HPLC (Table 3, Entry 4).
`Similarly, carbon-11 labeled quinoline compounds (MQB, 4a and
`DMQB, 4b) were stable when formulated with SA in a similar
`fashion (Table 3, Entry 5).
`The representative examples provided here show the worst-
`case scenario of radiolytic decomposition since without addition
`of anti-oxidant stabilizers, the tracers are unusable. However,
`given that SA is very efficient at inhibiting radiolysis, we re-visited
`formulation of other aniline containing radiopharmaceuticals that
`we routinely prepare and consider to be stable. [11C]DASB (3-
`amino-4-(2-dimethylaminomethyl-phenylsulfanyl)benzonitrile 6)
`is known to undergo very slow radiolysis and, not surprisingly
`given our previous findings, contains an aniline component.
`Slow radiolysis of a primary aniline is in keeping with the
`apparent trend that the rate of radiolysis increases with substitu-
`tion at the aniline: ArNMe2 4 ArNHMe 4 ArNH2. Seemingly, the
`more electron rich the aniline nitrogen, the more susceptible the
`compound is to radiolysis. Whilst formulation without ascorbate
`does not give an unusable product (radiochemical purity ¼ 90%),
`addition of 0.4% w/v SA to the isolation and formulation steps
`increases radiochemical purity of the formulated product to 99%
`and the product is stable for at least three half-lives of the carbon-
`11 radionuclide with no evidence of any decomposition (Table 3,
`Entry 6).
`
`4. Conclusions
`
`In conclusion we have shown that a number of potential new
`PET ligands as well as some established compounds in hum