Rapid carbon-11 radiolabelling for PET using microfluidics

Article abstract of DOI:10.1002/chem.201002644

Labels all around: A gas-liquid microfluidic reactor is used for the first time to efficiently perform [¹¹C]carbon monoxide carbonylation reactions for the preparation of ¹¹C-radiolabelled amide molecules within a short timeframe (<15 min; see scheme).

Full text of DOI:10.1002/chem.201002644

DOI: 10.1002/chem.201002644
Rapid Carbon-11 Radiolabelling for PET Using Microfluidics
Philip W. Miller,*[a, b] Hꢀlꢁne Audrain,[b] Dirk Bender,[b] Andrew J. deMello,[a]
Antony D. Gee,[a, c] Nicholas J. Long,[a] and Ramon Vilar[a]
Positron emission tomography (PET) is an imaging tech-
nique routinely used for screening, diagnosing and staging
chronic conditions such as cancer and neurodegenerative
diseases.[1] In addition to clinical applications, PET is also
widely used to gain a fundamental understanding of the un-
derlying biology of these diseases and to discover new treat-
ments.[2] All PET scans require a positron emitting radioiso-
tope to enter the body, usually in the form of an injected ra-
diopharmaceutical. The synthesis, preparation and purifica-
tion of radiopharmaceuticals for PET imaging are not easy
processes. The short half-lives of the common cyclotron-gen-
the challenges of increasing the speed of labelling reactions,
reducing their scale and improving the overall efficiency of
radiolabelling reaction processes.
Our group[6] and others[7] have recently reported the use
of microfluidic reactors for rapid and high-yielding carbony-
lation reactions. These Pd-catalysed carbonylation reactions
present an efficient synthetic route to a range of carbonyl-
containing organic compounds with biological relevance.[8]
For this reason Pd-mediated 11C carbonylation reactions
(Scheme 1) have attracted considerable interest for the
erated PET radioisotopes (11C: t1/2 =20.4 min, 18F: t1/2
=
110 min, 13N: t1/2 =9.96 min, 15O: t1/2 =2.04 min), coupled
with extremely low radioisotopic concentrations (pm–nm)
represent the main challenges in the synthesis of positron
emitting labelled compounds.[3] Microfluidics has recently
emerged as an important technology for the rapid synthesis
of short-lived radiopharmaceuticals for PET.[4] The advan-
tages of using microfluidic reactors for organic chemistry
are well documented and include the benefits associated
with miniaturisation: smaller reaction volumes (nL–mL) and
lower reagent quantities (nmol–mmol), controlled and predi-
cable mixing, efficient heat transfer and enhanced process-
ing capabilities.[5] Microfluidic reactors for PET radiosynthe-
sis have generated considerable interest primarily because
miniaturised reaction systems have the potential to address
Scheme 1. The three-component Pd-mediated aminocarbonylation reac-
tion to form an amide. The asterisk indicates the 11C-labelling position.
preparation of 11C-carbonyl-labelled compounds for PET.[9]
Radiochemical synthesis with 11CO is challenging and pres-
ents two key difficulties: firstly, carbon monoxide is sparing-
ly soluble in common organic solvents, and secondly, 11CO,
produced through the reduction of 11CO2, is delivered at ex-
tremely low concentrations (<nm) in an inert carrier gas.
11CO is always accompanied by a much larger quantity of
stable [12C/13C]CO, typically in >1000-fold excess. Several
technologies have emerged in recent years to improve the
processing and the reactivity of 11CO; these include high-
pressure micro-autoclave systems,[10] supported catalysts[11]
and inorganic 11CO trapping reagents.[12] Here, we report the
first example of a 11CO-labelling procedure using a micro-
fluidic reactor that exploits enhanced gas–liquid contact.
[a] Dr. P. W. Miller, Prof. A. J. deMello, Prof. A. D. Gee, Prof. N. J. Long,
Dr. R. Vilar
Department of Chemistry, Imperial College London
South Kensington, London, SW7 2AZ (UK)
Fax : (+44)20-7594-5804
E-mail: philip.miller@imperial.ac.uk
[b] Dr. P. W. Miller, Dr. H. Audrain, Dr. D. Bender
PET-Centre, Aarhus University Hospital
We designed
a glass-fabricated microfluidic reactor
Norrebrogade 44, 8000 Aarhus C (Denmark)
(Figure 1) with two inlet channels for the gaseous and liquid
reagents, a mixing-tee motif to permit gas and liquid con-
tact, a long residence channel to enhance reactivity and an
outlet port for product collection. The etched microchannels
have a semicircular cross-sectional profile and are 220 mm
[c] Prof. A. D. Gee
Division of Imaging Sciences, Kingꢀs College London
St Thomasꢀ Hospital, London SE1 7EH (UK)
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201002644.
460
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 460 – 463

COMMUNICATION
sieve trap into a stream of helium gas at room temperature.
A sample vial attached to the device exit port was used to
collect the labelled product after flushing the chip device
with toluene. A second sample vial containing a solution of
copper tris(3,5-dimethylpyrazolyl)-borate ([Cu
the Supporting Information) was used to trap and measure
any unreacted 11CO. This [Cu
(Tp*)] trapping solution has
ACHTUNGTRENUN(NG Tp*)], see
AHCTUNGTRENNUNG
recently been developed within our group[12a] and has
proven to be a highly efficient reagent for trapping 11CO.
Here, it is used as a convenient way of trapping unreacted
11CO in order to calculate the radioactivity trapping efficien-
cies (RTE) of these reactions.
Figure 1. Microfluidic device used for 11CO carbonylation reactions. The
device is filled with a blue dye to emphasise the channel structure.
Bottom left hand quadrant shows a close-up of the inlet channels. See
the Supporting Information for details of the microfluidic reactor design
and substrate solution preparation.
An annular type flow regime was imposed on our reaction
system as it provides a large gas–liquid contact area[13] and
importantly can be generated much faster and more reliably
than segmented flow within the short timeframes required
for 11C-labelling reactions. The entire labelling process was
typically complete within 15 min from the end of radionu-
clide production. The 11CO trapping, pre-concentration and
release process takes approximately 5–6 min whereas the
microfluidic chip reaction and subsequent chip flushing step
take a further 7–8 min to complete.
wide and 100 mm deep. The residence channel is 5 m in
length and occupies most of the deviceꢀs footprint area (90ꢂ
15 mm). In a previous study we found that a long residence
channel was necessary to give reasonably high chemical
yields for Pd-catalysed carbonylations under annular type
flow regimes.[6b] A schematic of our microfluidic radiolabel-
ling reaction setup is shown in Figure 2 and consists of a
A series of [carbonyl-11C]amides and a lactone were syn-
thesised from aryl iodide substrates (Table 1) through the
Pd-mediated [11C]carbonylation reaction by using our setup.
The catalyst used in all labelling reactions was [PdCl2-
AHCTUNGTREG(NNUN xantphos)] (xantphos=4,5-bis(diphenylphosphino)-9,9-di-
methylxanthene), which we previously found gave excep-
tionally high yields within short reaction times for carbony-
lation reactions.[6a] The model Pd-mediated 11C carbonyla-
tion reaction of iodobenzene and benzylamine was used as a
benchmark to test the efficiency of 11CO labelling reactions.
Under our microfluidic reaction conditions the 11C-carbony-
lative coupling reaction of iodobenzene with benzylamine
(Table 1, entry 1) gave encouragingly high RTEs averaging
88%. Analysis of this radiolabelled product by HPLC
showed the exclusive formation of the [carbonyl-11C]N-ben-
zylbenzamide with exceptionally high radiochemical purities
(RCP > 99%). The intramolecular Pd-mediated 11CO cou-
pling reaction of 2-iodobenzyl alcohol (Table 1, entry 2) to
form the lactone, [carbonyl-11C]phthalide, also gave excel-
lent RTE (87%) and RCP values (>99%). A range of
other aryl iodide substrates with the activated electron with-
drawing groups p-nitrile, o-pyridyl and p-trifluoromethyl, in
addition to the deactivated p-anisole group, was investigated
(Table 1). Good to excellent RTEs were obtained in all
cases, however, RCPs were decreased owing to the forma-
tion of unknown radioactive byproducts.
Figure 2. Schematic of the microfluidic 11CO carbonylation setup.
series of valves to control and direct gas flow, a stainless
steel loop packed with molecular sieves, liquid reagent injec-
tor port, microfluidic device, heating plate, solvent pump,
collection vial and 11CO trapping vial. In a typical 11CO-la-
belling experiment the coupling reagents (aryl halide, Pd
catalyst and amine) were premixed and loaded into a 50 mL
loop on a Rheodyne valve connected to an external syringe
pump charged with toluene. The 11CO was trapped and con-
centrated into a smaller volume at À1968C by using a mo-
lecular sieve stainless steel loop. This trapping stage is nec-
essary to increase the speed of the overall labelling process
by reducing the total volume of gas that has to be processed
through the chip device. The coupling reagents were infused
into the microfluidic device while, simultaneously, the
trapped 11CO was controllably released from the molecular
The microfluidic reactions gave exceptionally good radio-
chemical yields considering the short residence times of
both the gas (2 s) and liquid (2 min) reagents.1 The high effi-
ciency of radiolabelling is attributed to the improved gas/
liquid contact and heat transfer within the microchannels of
1
The liquid residence times were determined experimentally to be
ꢀ2 min whereas the gas residence time was calculated to be 2.1 s at a
gas flow rate of 2.5 cm3 minÀ1 (device volume/flow rate=0.087/2.5ꢂ
60=2.1 s).
Chem. Eur. J. 2011, 17, 460 – 463
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
461

P. W. Miller et al.
Table 1. Results from Pd-mediated microfluidic [11C]carbonylation reac-
tions.
and V2 (Figure 2) were switched to direct helium flow through the micro-
fluidic device at a pressure of 3 bar and flow rate of 2.5 cm3 minÀ1. The
trapped 11CO was controllably released by allowing the molecular sieve
loop to slowly warm to room temperature by removing the liquid nitro-
gen Dewar over a period of two minutes by using a remote-controlled
linear actuated platform that was built in-house. The radiodetector indi-
cated the controlled loss of activity from the trapping loop. At the same
time as 11CO release from the trap, the premixed carbonylation reagents
were infused into the microfluidic device at a flow rate of 5 mLminÀ1
from a pre-loaded 50 mL loop on a Rheodyne injector 7725i valve (V3)
by using a Harvard pump 11 syringe pump charged with anhydrous tolu-
ene. The microfluidic device was pre-heated to 1508C for the synthesis of
[carbonyl-11C]amides and 1108C for [carbonyl-11C]phthalide. A collection
vial was connected to the chip outlet vial PTFE tubing. A second vial
Entry
Substrate
Product
RTE [%][a] RCY [%][b]
1
88 (n=4)
87 (n=2)
88
87
2
3
4
5
78 (n=2)
65 (n=2)
81 (n=2)
76
44
56
containing the 11CO trapping complex [Cu
ACTHNUTRGNE(UNG Tp*)] in anhydrous toluene
solution (1 mL) was connected to the product collection vial to receive
the vented gases and to trap any unreacted 11CO. Following a 10 min re-
action period, the microfluidic device was rapidly flushed with 100 mL of
toluene. After this time the collection and trapping vials were removed
from the hot cell and their radioactivity measured in a dose calibrator. In
a typical labelling reaction 400–700 MBq of radioactivity was recorded in
the collection vial after 20 min from EOB (typical bombardment time
was 10 min at 40 mA). The identities of the 11C-labelled compounds were
confirmed by using HPLC by co-injection of authentic reference samples.
The contents of the collection vial were diluted with acetonitrile (250 mL)
and a 20 mL aliquot removed for UV/Vis and radio-HPLC analysis (sol-
vent: 60:40 acetonitrile/sodium phosphate buffer solution (70 mm), flow:
6
90 (n=2)
58
[a] Radioactivity trapping efficiency, based on the radioactivity trapped
in the collection vial expressed as a fraction of total of radioactivity (i.e.,
combination of collection and trapping vials). [b] Decay corrected RCY
(radiochemical yield), based on the total radioactivity delivered to vial
corrected for radiochemical purity; n=number of runs.
1.5 mLminÀ1
, column: phenomenex sphereclone ODS, 250ꢂ4.6 mm).
The radio-HPLC system consisted of a Dionex Summit system (P680
pump and UVD 170U detector) connected in series with a NaI radiode-
tector of in-house design. Dionex Chromeleon 6.8 software was used for
data acquisition and analysis.
the device, which we believe enhances the 11CO insertion
step even at these low isotopic dilutions. One factor that
limits the overall speed of this labelling procedure is the
rate of 11CO gas processing through the device, currently
this process takes 7–8 min. Significant reductions in time
may be achieved by either increasing the gas flow rate or by
adding extra devices to run in parallel, thus, there is the po-
tential to reduce reaction times from minutes to seconds.
In summary, a microfluidic reactor has been used to rap-
idly and efficiently radiolabel simple amide molecules
through the highly versatile Pd-mediated 11CO carbonyla-
tion reaction. RCYs are on a par with, or exceed that, of
currently used methods.[9,11,12] This method is technically
straightforward and further demonstrates the utility and po-
tential of 11CO-labelling reactions. We are presently apply-
ing this method to more challenging coupling reactions for
the ultra rapid synthesis of potential 11C PET tracer mole-
cules.
Acknowledgements
P.W.M. is grateful to the EPSRC for a Life Sciences Interface Fellowship
(EP/E039278/1) and to J. K. Graverholt, T. Knak and J. Welander-
Madsen for technical support.
Keywords: carbon · microfluidic reactors · positron emission
tomography · radiochemistry · radiopharmaceuticals
[1] a) E. M. Rohren, T. G. Turkington, R. E. Coleman, Radiology 2004,
231, 305–332; b) W. A. Weber, J. Clin. Oncol. 2006, 24, 3282–3292;
c) K. Herholz, W. D. Heiss, Mol. Imaging. Biol. 2004, 6, 239–269.
[2] a) M. E. Phelps, Proc. Natl. Acad. Sci. USA 2000, 97, 9226–9233;
b) J. S. Fowler, N. D. Volkow, G. J. Wang, Y. S. Ding, S. L. Dewey, J.
Nucl. Med. 1999, 40, 1154–1163.
[3] a) J. S. Fowler, A. P. Wolf, Acc. Chem. Res. 1997, 30, 181–188;
b) P. W. Miller, N. J. Long, R. Vilar, A. D. Gee, Angew. Chem. 2008,
120, 9136–9172; Angew. Chem. Int. Ed. 2008, 47, 8998–9033.
[4] a) P. W. Miller, J. Chem. Technol. Biotechnol. 2009, 84, 309–315;
b) P. W. Miller, A. J. deMello, A. D. Gee, Curr. Radiopharm. 2010, 3,
254–262; c) C. C. Lee, G. D. Sui, A. Elizarov, C. Y. J. Shu, Y. S. Shin,
A. N. Dooley, J. Huang, A. Daridon, P. Wyatt, D. Stout, H. C. Kolb,
O. N. Witte, N. Satyamurthy, J. R. Heath, M. E. Phelps, S. R. Quake,
H. R. Tseng, Science 2005, 310, 1793–1796.
[5] a) A. J. deMello, Nature 2006, 442, 394–402; b) K. Jꢃhnisch, V.
Hessel, H. Lowe, M. Baerns, Angew. Chem. 2004, 116, 410–451;
Angew. Chem. Int. Ed. 2004, 43, 406–446.
[6] a) P. W. Miller, L. E. Jennings, A. J. deMello, A. D. Gee, N. J. Long,
R. Vilar, Adv. Synth. Catal. 2009, 351, 3260–3268; b) P. W. Miller,
N. J. Long, A. J. deMello, R. Vilar, J. Passchier, A. Gee, Chem.
Commun. 2006, 546–548.
Experimental Section
General microfluidic radiolabelling procedure: [11C]carbon dioxide was
produced by using a GE PETtrace cyclotron by 16 MeV proton bom-
bardment of a target containing nitrogen and 1% oxygen. 11CO was pro-
duced by using the GE process cabinet reduction module by passing
11CO2 over a molybdenum wire packed into a quartz tube at 8508C. In a
typical procedure, 11CO/N2 gas stream was passed through the molecular
sieve trap at À1968C at a flow rate of 50 cm3 minÀ1. A radioactivity detec-
tor was placed adjacent to the molecular sieve loop to monitor the activi-
ty trapping. Following the peak in activity on the 11CO trap, valves V1
462
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 460 – 463

Carbon-11 Radiolabelling
COMMUNICATION
[7] a) E. R. Murphy, J. R. Martinelli, N. Zaborenko, S. L. Buchwald,
K. F. Jensen, Angew. Chem. 2007, 119, 1764–1767; Angew. Chem.
Int. Ed. 2007, 46, 1734–1737; b) M. T. Rahman, T. Fukuyama, N.
Kamata, M. Sato, I. Ryu, Chem. Commun. 2006, 2236–2238.
[8] a) C. F. J. Barnard, Organometallics 2008, 27, 5402–5422; b) A.
Brennfꢄhrer, H. Neumann, M. Beller, Angew. Chem. 2009, 121,
4176–4196; Angew. Chem. Int. Ed. 2009, 48, 4114–4133; c) C. Tor-
borg, M. Beller, Adv. Synth. Catal. 2009, 351, 3027–3043.
845–848; c) J. Eriksson, O. Aberg, B. Langstrom, Eur. J. Org. Chem.
2007, 455–461.
[11] P. W. Miller, N. J. Long, A. J. deMello, R. Vilar, H. Audrain, D.
Bender, J. Passchier, A. Gee, Angew. Chem. 2007, 119, 2933–2936;
Angew. Chem. Int. Ed. 2007, 46, 2875–2878.
[12] a) S. Kealey, P. W. Miller, N. J. Long, C. Plisson, L. Martarello, A. D.
Gee, Chem. Commun. 2009, 3696–3698; b) H. Audrain, L. Martarel-
lo, A. Gee, D. Bender, Chem. Commun. 2004, 558–559.
[13] A. Gunther, K. F. Jensen, Lab Chip 2006, 6, 1487–1503.
[9] B. Langstom, O. Itsenko, O. Rahman, J. Labelled Compd. Radio-
pharm. 2007, 50, 794–810.
[10] a) T. Kihlberg, B. Langstrom, T. Ferm, J. Eriksson, US2008095693,
2005; b) E. D. Hostetler, H. D. Burns, Nucl. Med. Biol. 2002, 29,
Received: September 14, 2010
Published online: December 3, 2010
Chem. Eur. J. 2011, 17, 460 – 463
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
463