Radiolabelling diverse positron emission tomography (PET) tracers using a single digital microfluidic reactor chip

Article abstract of DOI:10.1039/c3lc51195b

Radiotracer synthesis is an ideal application for microfluidics because only nanogram quantities are needed for positron emission tomography (PET) imaging. Thousands of radiotracers have been developed in research settings but only a few are readily avail

Full text of DOI:10.1039/c3lc51195b

Lab on a Chip
PAPER
Radiolabelling diverse positron emission
tomography (PET) tracers using a single digital
Cite this: Lab Chip, 2014, 14, 902
microfluidic reactor chip†
Supin Chen, a Muhammad Rashed Javed,bc Hee-Kwon Kim,‡bc Jack Lei,bc
*
Mark Lazari,abc Gaurav J. Shah,bcd R. Michael van Dam,abc Pei-Yuin Kengbc
and Chang-Jin “CJ” Kimae
Radiotracer synthesis is an ideal application for microfluidics because only nanogram quantities are needed
for positron emission tomography (PET) imaging. Thousands of radiotracers have been developed in
research settings but only a few are readily available, severely limiting the biological problems that can be
studied in vivo via PET. We report the development of an electrowetting-on-dielectric (EWOD) digital
microfluidic chip that can synthesize a variety of 18F-labeled tracers targeting a range of biological processes
by confirming complete syntheses of four radiotracers: a sugar, a DNA nucleoside, a protein labelling
compound, and a neurotransmitter. The chip employs concentric multifunctional electrodes that are used
for heating, temperature sensing, and EWOD actuation. All of the key synthesis steps for each of the four
18F-labeled tracers are demonstrated and characterized with the chip: concentration of fluoride ion, solvent
exchange, and chemical reactions. The obtained fluorination efficiencies of 90–95% are comparable to, or
greater than, those achieved by conventional approaches.
Received 22nd October 2013,
Accepted 5th December 2013
DOI: 10.1039/c3lc51195b
www.rsc.org/loc
powerful medical tool used for diagnosing disease (like
Introduction
cancer and Alzheimer's disease), monitoring treatment, and
Positron emission tomography (PET) is a non-invasive imaging
technique for quantification of biomolecular processes through
detection of gamma rays arising from the annihilation of a
developing pharmaceuticals.2
Fluorine-18 is an ideal radioisotope for PET imaging
because it emits a low energy positron, has a shorter positron
range compared to other positron-emitters, decays to an
innocuous 18O, and has a moderate half-life that is long
enough for chemical synthesis, transport to imaging clinic,
and imaging molecules in vivo.3 The fluorine atom can also
be labelled onto many biomolecules because it is bioisostere
to oxygen and can often substitute for a hydroxyl group or
replace a hydrogen atom bound to a carbon atom.4 More
than 1400 fluorine-18 radiotracers have been developed in
research settings for imaging biological processes,5 but only
a handful are commercially available. Production of radio-
tracers is limited by large capital investment in equipment
and infrastructure, high operating costs, and short tracer
lifetime.6 Recently, there has been a movement towards tech-
nologies for decentralizing PET radiotracer production, in
which imaging centers synthesize their own radiotracers
onsite after receiving radioisotopes from radiopharmacies.
For that reason, commercial automated synthesizers (e.g., Siemens
Explora® One, Eckert & Ziegler Modular-Lab Standard, and
GE FASTlab) were developed for flexible synthesis of various
PET radiotracers using one machine.7 However, these systems
may require manual reconfiguration to synthesize different
radiotracers8 and must be operated within expensive and
positron (emanated from a radiotracer) and an electron.1
A
radiotracer is a biochemical compound tagged with a short-
lived radionuclide, and the type of radiotracer determines the
specific biological function that can be quantified via PET:
sugars for monitoring metabolism, nucleic acids for examining
cell replication, and neurotransmitters for studying brain activity.
The ability to capture these molecular functions make PET a
a Bioengineering Department, University of California, Los Angeles, CA 90095,
USA. E-mail: supin.chen@engineering.ucla.edu; Fax: +1(310)825 0267;
Tel: +1(310)825 3977
b Molecular and Medical Pharmacology, University of California, Los Angeles,
CA 90095, USA
c Crump Institute for Molecular Imaging, University of California, Los Angeles,
CA 90095, USA
d Sofie Biosciences, Culver City, CA 90230, USA
e Mechanical and Aerospace Engineering, University of California, Los Angeles,
CA 90095, USA
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c3lc51195b
‡ Current affiliation: Department of Nuclear Medicine, Molecular Imaging &
Therapeutic Medicine Research Center, Biomedical Research Institute,
Chonbuk National University Medical School and Hospital, Jeonju, Jeonbuk
561-712, South Korea.
902 | Lab Chip, 2014, 14, 902–910
This journal is © The Royal Society of Chemistry 2014

Lab on a Chip
Paper
bulky hot cells (special chemical fume hoods that are surrounded
by lead to provide radiation shielding).
be absorbed in the chip substrate rather than solvent,
which would significantly reduce the formation of radicals and
radiolysis, in comparison with a macroscale geometry.24
Miniaturization by microfluidic systems is of direct interest
for radiotracer development because of the substantial savings
from both using less lead shielding and reducing use of
expensive reagents. Reaction efficiency can also be improved
with reduced volumes because of increased concentration of
the radioisotope, expedited heat transfer, and quicker mass
transport.3 This is important for radiosynthesis reactions that
sometimes involve low-yields and short half-life isotopes.4
While the small amount of product limits the utility of
microfluidic chemical synthesis in some cases, it is not an
issue for PET imaging, which requires product in only nano-
gram quantities. Several microfluidic systems have been
developed for radiotracer synthesis, including flow-through
systems and elastomer batch-flow reactors.9,10 A commercial
microfluidic reactor, the Advion NanoTek®, has also been
used for radiotracer synthesis. It incorporates distribution
valves, syringe pumps, pressure sensors, and heated vials to
store and deliver reagents, concentrate radioisotopes, and
control reactions.11 The NanoTek® can transfer liquid volumes
as small as 100 μL through a 16 μL capillary reactor and has
been used to synthesize [18F]fluoro-2-deoxy-D-glucose ([18F]FDG)
in 3 minutes with 62% yield.12
A new and promising microfluidic approach of digital
microfluidics13–15 does not require mechanical valves, pumps,
or channels, but instead uses electric potentials to manipulate
liquids through the mechanism of electrowetting-on-dielectric
(EWOD).16 Past reports have demonstrated a range of droplet
functions (i.e., generating, moving, splitting, and merging),17
precise control of droplet volumes,18 on chip liquid composi-
tion measurement,19 and suitability for multiple reagent
chemical synthesis.20 Our group has recently reported the use
of EWOD for the synthesis of the most commonly used PET
radiotracer, [18F]FDG.21
The EWOD-driven digital microfluidics has key advantages for
radiotracer synthesis. Because droplet movement is electronically
controlled without valves, pumps or tubes, liquid pathways can
be defined in software, enabling diverse chemical synthesis pro-
cesses to be carried out with just one type of EWOD chip. Since
the sidewalls of channels are not necessary, an EWOD chip can
be open to air, as first demonstrated by Lee et al.13 The open-to-
air configuration enables rapid drying, evaporation, and solvent
exchange; these are critical steps for no-carrier-added fluorination
reactions, which are water-sensitive but begin with [18F]fluoride
ion obtained in [18O]-enriched water from a cyclotron.4 The low
volumes (2–12 μL) used on the EWOD chip are conducive to sim-
plifying the purification process and increasing specific activity
(radioactivity per mass of tracer) due to the minute amount of
reagents used in a single batch of synthesis.11,22,23
To avoid different chip designs for different tracers, in this
report we explore and confirm a single EWOD chip design that
could produce a variety of radiotracers, completing our prelimi-
nary efforts,22,25 along with significantly improved loading
practices, shorter processing time, increased [18F]fluoride con-
centration, and radiochemical yields. Four exemplary radio-
tracers currently used in clinical and preclinical research are
demonstrated: (1) [18F]FDG (a sugar analogue), (2) 3′-deoxy-
3′[18F]fluorothymidine ([18F]FLT, a DNA nucleoside analogue),
(3) N-succinimidyl 4-[18F]fluorobenzoate ([18F]SFB, a prosthetic
group for protein labelling), and (4) [18F]fallypride (a neuro-
transmitter analogue). The EWOD platform, as an affordable
and flexible synthesizer, has the potential to empower final
users to produce tracers of their choice locally on demand and
eliminate bottlenecks due to centralized production.
Experimental
Device description and fabrication
The EWOD chip has a configuration of two parallel plates
with a gap space, in which droplets are sandwiched to a disk
shape. The electrical ground plate is on top and uses a trans-
parent conductive layer of indium tin oxide (ITO) to maintain
a reference electrical connection for EWOD actuation. The
actuation plate is on the bottom and has defined electrodes
(also transparent ITO) for six droplet pathways that meet at
a circular heating site (Fig. 1). The heating site consists
of 4 concentric multifunctional electrodes,26 each a resistive
element that can be used either for EWOD actuation of drop-
lets (when voltage but no current is applied) or for feedback-
controlled heating (when both voltage and current are
applied). The multi-element heating site can center the drop-
let on the heater and maintain temperature more uniformly
than a single-element heater26,27 as the droplet shrinks
during evaporations.
Both EWOD actuation plates and electrical ground plates were
diced from 700 μm thick glass wafers coated with 140 nm ITO
(Semiconductor Solutions). Chrome (20 nm) and gold (200 nm)
were evaporated onto the wafers. The gold, chrome, and ITO
layers were patterned to form EWOD electrodes, heaters, connec-
tion lines, and contact pads by photolithography and wet etching.
The silicon nitride dielectric was deposited by plasma-enhanced
chemical vapor deposition at a thickness of 2 μm on the actua-
tion plate and a thinner 100 nm on the electrical ground plate.
Teflon® (250 nm) was spin-coated and annealed at 340 °C
under vacuum to make the surfaces hydrophobic. A 140 μm
gap between the assembled actuation plate and electrical
ground plate was maintained by two layers of double-sided
tape (3M Inc.).
Another potential advantage yet to be confirmed experimen-
tally, is reduced radiolysis, which is damage to the reagents
due to formation of radicals by energy that is mainly deposited
from positron emission. Because the EWOD droplets are
squeezed within a gap that is smaller than the positron range,
it is expected that a significant portion of positrons will
The hydrophobic coating consisted of Cytop® (glass
transition temperature Tg = 108 °C) in early devices,21,26 but
was switched to Teflon® AF 2400 (Tg = 240 °C) in the current
devices for its higher thermal stability. Teflon® AF 2400 was
This journal is © The Royal Society of Chemistry 2014
Lab Chip, 2014, 14, 902–910 | 903

Paper
Lab on a Chip
Fig. 2 Electronic control scheme to operate the EWOD chip for tracer
synthesis. A computer controls a switchboard through a digital input/
output device. The switches connect the electrodes on the EWOD
chip to an amplified voltage for EWOD actuation. When necessary,
some switches connect the multifunctional electrodes to a heater
driver for temperature sensing and resistive heating via feedback
control from the computer through a second digital input/output
device.
program using a digital I/O device (NI USB-6509, National
Instruments).
Fig.
1 EWOD chip designed and fabricated for tracer synthesis.
A multichannel heater driver was designed and built in
house and interfaced with a digital I/O device (NI USB-6343,
National Instruments) to maintain feedback-controlled tem-
peratures over the four individual resistive heaters on each
actuation plate. The multifunctional electrodes were individu-
ally connected to a switch to alternate between EWOD actua-
tion or temperature measurement and heating.
(A) Schematic of the chip, showing the actuation glass plate patterned
with regular EWOD electrodes (blue), multifunctional electrodes
named heating site (orange and red), and electric contact pads along
the two edges, as well as the electrical ground glass plate assembled
on top. (B) A picture of the fabricated chip, showing the EWOD and
multifunctional electrodes (slightly darker) and gold connection lines
and contact pads (yellow).
Reagents and materials
considered even more stable than Teflon® AF 1600 (Tg = 160 °C)
used in the work preliminary to this report.22,25 The highest
temperature used in our radiolabelling is 120 °C. After the
hydrophobic topcoat was switched from Cytop® to Teflon® AF
2400, less residual radioactivity remained “stuck” to the chip
surface after synthesis. Compared to the initially loaded activity,
the extracted product radioactivity has increased from 45%
(in the preliminary work although not specifically reported in
ref. 22) to 77% (decay-corrected) for [18F]FLT synthesis.
One concern with Teflon® is that it can be a source for
fluorine-19 contamination into the reagents and result in a
lower specific activity of the final product.28 However, because
[18F]fallypride synthesized by EWOD has some of the highest
reported specific activity (19 Ci μmol−1), fluorine-19 contamina-
tion from Teflon® appears not likely to be a major issue.23
No-carrier-added [18F]fluoride ion was obtained by irradiation
of 97% [18O]-enriched water with an 11 MeV proton beam
using a cyclotron (RDS-112, Siemens, Knoxville, TN) at the
UCLA Ahmanson Biomedical Cyclotron Facility.
Potassium carbonate (K2CO3, 99%), 2,3-dimethyl-2-butanol
(thexyl alcohol), anhydrous acetonitrile (MeCN, 99.8%),
anhydrous dimethyl sulfoxide (DMSO, 99.9%), anhydrous
methanol (MeOH, 99.8%), and N,N,N′,N′-tetramethyl-O-
(N-succinimidyl)uronium hexafluorophosphate (HSTU, 98%)
were purchased from Sigma Aldrich. Hydrochloric acid (HCl,
1 N) and sodium hydroxide (NaOH, 1 N) were purchased
from Fisher Scientific. Mannose triflate (FDG precursor),
4,7,13,16,21,24-hexaoxa-1,10-diazobicyclo (8.8.8)-hexacosane
(K2.2.2, 98%), tetrabutylammonium bicarbonate solution (75 mM)
(TBAHCO3),
3-N-Boc-5′-O-dimethoxytrityl-3′-O-nosyl-thymidine
(FLT precursor), 4-(tert-butoxycarbonyl)-trimethylbenzeneammonium
triflate (SFB precursor), and tosyl-fallypride (fallypride precursor)
were purchased from ABX Advanced Biochemical Compounds
(Radeberg, Germany). All chemicals were used as received. A
vortex mixer (Vortex Genie 2, Scientific Industries) was used for
off-chip preparation of reagents.
Device operation
EWOD actuation voltage for droplet movement was generated
from a 10 kHz signal (33220A waveform generator, Agilent
Technologies) and amplified to 100 Vrms (Model 601C, Trek)
(Fig. 2). The voltage was applied selectively to desired electrodes
for droplet movement by solid-state relays (AQW610EH
PhotoMOS relay, Panasonic) that were controlled by a LabVIEW
Neutral alumina (50–300 μm particle size), C-18 resin
(55–105 μm particle size), and hydrophilic–lipophilic-balanced
904 | Lab Chip, 2014, 14, 902–910
This journal is © The Royal Society of Chemistry 2014

Lab on a Chip
Paper
resin were purchased from Waters (Milford, MA). Ion retarda-
tion resin (AG11 A8, 180–425 μm particle size) and cation
exchange resin (AG-50W-X4, 63–150 μm particle size) were
purchased from BioRad Laboratories (Hercules, CA).
Solvents and solutions commonly used for radiotracer
synthesis were tested for EWOD compatibility. Repeatable move-
ment, splitting, and merging of droplets was accomplished with
MeCN, DMSO, MeOH, TBAHCO3, TSTU, K2CO3 in water (1 M),
HCl (1 N), and NaOH (1 N), with voltages below 80 Vrms and
frequencies ranging 0–1 kHz.
to a maximum volume of 16 μL fluoride solution that could
be reduced by evaporation.26 More radioactivity can be used
by repeated loading and heating steps, but this lengthens the
synthesis time. Further development allowed us to concen-
trate larger volumes by not squeezing the initial droplet
between the plates (i.e., into a disk shape) but instead loading
and heating a much larger droplet on the open surface of the
bottom plate (i.e., in a spherical shape) near the edge of the
top electrical ground plate, as shown in the top figure of
Fig. 3a. Here, a 200 μL droplet of fluoride with base and phase
transfer catalyst could be heated outside of the gap and then
pulled between the gap by EWOD after its volume reduced to
5 μL (Fig. 3). Within the gap, the droplet could shrink more or
be completely dried with further heating. The total heating
time was 12 minutes, but the 10 minutes outside of cover plate
could potentially be reduced with the addition of a heater out-
side of the cover plate, specifically for concentration. Calibrated
ion chamber measurements before and after concentration
showed that negligible radioactivity was lost during the whole
process. This method of removing bulk water for radiotracer
synthesis is different from the conventional method, in which
[18F]fluoride is typically trapped on a cartridge of anion
exchange resin and then eluted off in a release solution. The
concentration cartridge can both reduce evaporation time
(if the release solution contains less water) and also remove
impurities that are released by the cyclotron target. However,
the ability to concentrate fluoride ion directly on the EWOD
chip without using anion exchange resin provides an alterna-
tive means of preparing radiotracers with high specific activ-
ity. A recent report by Lu et al.11 showed high fluorine-19
carrier contamination when the fluorine-18 solution was first
concentrated with anion exchange resin.
Analytical methods
Radioactivity was measured from samples using a calibrated
ion chamber (Capintec CRC-15R). Radio thin layer chromatog-
raphy (radio-TLC) was analysed with a scanner equipped with
scintillation probe for gamma rays (MiniGITA star, Raytest).
High performance liquid chromatography (HPLC) was carried
out on a reversed-phase C-18 column (250 × 4.6 mm, 5 μm,
Phenomenex Luna) equipped with a variable wavelength UV
detector and a radiation detector (Eckert & Ziegler).
Fluoride concentration
[18F]fluoride is delivered in [18O]-enriched water from the
cyclotron and is mixed with base and phase transfer catalyst.
In earlier work21 including those preliminary to this report,25,26
the mixing was done off chip with a vortex mixer prior to
loading onto the EWOD chip. For [18F]FDG and [18F]SFB
radiolabeling in this work, mixing was instead performed on
chip by EWOD actuation after loading solutions of base and
phase transfer catalyst separate from [18F]fluoride onto the
chip. After being loaded onto the electrical ground plate edge
by using a micropipette, the 2–10 μL droplets were pulled in
between the plates and moved to the circular multifunctional
electrode site by EWOD actuation. The loading process was
repeated until the desired volume was reached, upon which
the mixed droplet was heated at 105 °C to dry.
Solvent exchange and azeotropic drying
Because of its open-to-air configuration, the EWOD chip facil-
itates evaporative removal of solvent, unlike many other types
of microfluidic devices.30 Solvent exchange is performed on
the EWOD chip by heating until the solvent has been evaporated
and its vapours removed with a nitrogen stream, leaving
The amount of radioactivity loaded was initially restricted
by the droplet volume that could be squeezed within the
140 μm gap over the 12 mm diameter heater site, limiting us
Fig. 3 Images of Cerenkov radiation emitted during fluorine-18 positron decay during fluoride concentration and cross-section schematics of the
chip in corresponding stages. Cerenkov analysis of on-chip synthesis has been demonstrated by Dooraghi et al.29 In the Cerenkov images, red
represents a higher concentration of radioactivity. Orange lines and circles were added to depict the cover plate edge and the heating site, respectively.
(A) 200 μL droplet of [18F]fluoride with base and phase transfer catalyst loaded to the cover plate edge (depicted with an orange line). (B) Droplet at
cover plate edge after its volume was reduced by heating the nearby multifunctional electrodes to 150 °C. (C) Droplet at center of heater after EWOD
actuation. (D) Residue at the heater center after solvent evaporation.
This journal is © The Royal Society of Chemistry 2014
Lab Chip, 2014, 14, 902–910 | 905

Paper
Lab on a Chip
behind a solute residue at the heater region. A droplet of the
new solvent is loaded at the cover plate edge, and then EWOD
actuation is used to pull the droplet in between the plates,
move it to the residue site, and dissolve the residue.
drying step was used, reducing the overall synthesis time and
therefore radioactive decay.
Fluorination reaction
For azeotropic drying, MeCN is added at the electrical
ground plate edge, moved to the reactor site by electrowetting
actuation, mixed, and heated at 105 °C for 1 min. Due to the
properties of the water-MeCN azeotrope, this is an effective
method for removing residual water. Typically, fluorine-18
radiochemistry processes using the macroscale synthesizer
include 2–3 successive cycles of azeotropic evaporation with
MeCN to obtain the “naked” [18F]fluoride ion for the subse-
quent fluorination reaction.4 Multiple azeotropic evapora-
tions were also used in the preliminary devices21,26 as well as
our early work.22,25 However, in the current microfluidic
droplet synthesis, no significant improvements in reaction
yields were found when increasing the number of azeotropic
drying steps on the EWOD chip. Thus, only one azeotropic
Fluorination was realized on the EWOD chip by first loading
a precursor solution (Fig. 4). After electrowetting pulled the
precursor droplet from the electrical ground plate edge and
moved it to the dried fluoride, the heating site was heated to
105–120 °C for 2–6 minutes depending on the type of radio-
tracer. For [18F]FDG, the 10 minute fluorination time used
before21 was reduced to 5 minutes in this report without a
noticeable change in reaction yield.
Conventionally, nucleophilic radiofluorination reactions
use MeCN as a solvent because its low boiling point (82 °C)
facilitates rapid removal by evaporation after the synthesis.
Although many nucleophilic fluorination reactions have been
reported to be more efficient in DMSO due to its high polariz-
ability and ability to perform reactions at higher temperatures,31
Fig. 4 Synthesis schemes for the four radiotracers produced on the EWOD chip. (A) Solvent exchange of fluoride with phase transfer catalyst.
(B) Two-step synthesis of [18F]FDG. (C) Two-step synthesis of [18F]FLT. (D) Three-step synthesis of [18F]SFB. (E) One-step synthesis of [18F]fallypride.
906 | Lab Chip, 2014, 14, 902–910
This journal is © The Royal Society of Chemistry 2014

Lab on a Chip
Paper
DMSO is not typically used in macroscale synthesis because its
low volatility (189 °C boiling point) makes its evaporative
removal difficult in a short time. However, microliter volumes
of DMSO could be effectively evaporated at 120 °C in the
EWOD chip after fluorination. The remaining DMSO did not
affect following reactions in the synthesis, and the residual
amount after synthesis (870 ppm) was below the acceptable
limits for human use (5000 ppm).21
DMSO, MeOH, H2O). An automated approach was also tested,
in which the crude product droplet were moved away from
the reactor after synthesis reactions to an outlet droplet path-
way by EWOD actuation, and then pulled off chip into a vial
by vacuum. An additional 3–6 μL solvent was required to
move the product droplets away from the heating site for
[18F]FDG, [18F]FLT, and [18F]fallypride, but no additional
solvent was required for [18F]SFB.
Radioactivities of the [18F]fluoride complex solution before
and after loading, the total EWOD chip after the synthesis,
the extracted product, actuation plate, electrical ground plate,
and pipette tips were measured with a calibrated ion chamber.
The extraction efficiency was determined by dividing the radio-
activity of the crude product droplet after extraction by the post-
synthesis radioactivity of the total EWOD chip before extraction.
Other reactions
Ideally, fluorine incorporation is the last synthesis step for
radiotracer preparation in order to minimize the overall
synthesis time and thus minimize the amount of radioactivity
decay and reduce radiation exposure to the operator. A syn-
thesis process with fewer steps is also inherently more reliable
and simpler to automate. However, due to the high basicity of
the “naked” fluoride ion, the majority of complex molecules
require additional synthesis steps after fluorination, such as:
removal of protecting groups (by acidic or alkaline hydrolysis),
saponification, esterification, and coupling reactions.
To accomplish multistep reactions in the EWOD chip, the
reagents were loaded from the electrical ground plate edge
and mixed with the previous reaction's products by electro-
wetting. The multifunctional electrodes were then heated to
the set reaction temperature over a predetermined reaction
time. All of the synthesis reactions were performed at the
same heating site without any intermediate purification steps.
Although efforts were made to maximize individual iso-
lated pathways for loading reagents, the synthesis results
were not affected in tests where the same route was used for
all reagents. This was the case, even if a reagent was water
sensitive and followed the same path as an aqueous reagent.
Contamination from adsorption was not an issue, because all
of the radiotracer processes demonstrated involved a one-pot
synthesis and only one reactor site was used on chip. Also,
there was very limited radioactive residue on the chip during
droplet routing based on previous work using Cerenkov imaging
for analysis.29 However, residue became significant on the
reactor site after heating.
Radio-TLC was used to analyse the reaction efficiencies
after crude reaction mixtures were spotted onto silica gel
plates and developed in appropriate mobile phases. Multiple
peaks were visible in each TLC chromatogram, one corre-
sponding to the desired reaction product and the other(s) to
unreacted starting material and/or byproducts. The reaction
efficiency was determined from the area under the product
peak divided by the total area under all peaks. For radiotracer
synthesis processes requiring multiple reactions, the com-
bined reaction efficiency was the multiplied product of all
the individual reaction efficiencies. Radio-TLC was also used
to measure radiochemical purity of the purified product.
Purification and analysis
The crude reaction mixtures were purified off chip to isolate
the desired radiotracer from unreacted reagents, intermedi-
ates, and by-products, which are not safe to be administered
into animals and patients, and can complicate interpreta-
tion of images if the by-products are radioactive. Two differ-
ent purification methods were used depending on the type
of radiotracer.
[18F]FDG and [18F]FLT were purified via a solid phase
extraction (SPE) method using miniaturized cartridges pre-
pared in-house. Each cartridge contained 4 types of SPE
sorbents. For [18F]FDG purification, neutral alumina, ion
retardation, cation exchange, and C-18 sorbents removed
impurities as the crude radiotracer mixture was pushed into
the cartridge and the purified radiotracer was outputted. For
[18F]FLT purification, neutral alumina, ion retardation, and
cation exchange sorbents were also used, but C-18 was
replaced with hydrophilic–lipophilic-balanced resins to retain
[18F]FLT as impurities from the crude radiotracer were out-
putted from the cartridge into waste. The desired [18F]FLT
radiotracer was then eluted as a purified product in ethanol.
HPLC was used to purify [18F]SFB and [18F]fallypride, as
the final radiotracer and other impurities were difficult to be
purified via solid phase extraction. Due to the small quanti-
ties of reagents used, the mass of impurities is much lower
from the EWOD chip than conventional systems, and thus an
analytical-scale HPLC column was sufficient for preparative
use. A separate analytical HPLC system was used to test the
chemical and radiochemical purity for all of the synthesized
radiotracers after purification.
While we used fluorination efficiency and hydrolysis effi-
ciency to characterize radiotracer synthesis for the prelimi-
nary results,22,25 in this report we demonstrated the entire
PET probe production process starting from synthesis, and
including purification and reformulation. Here we report the
final decay corrected radiochemical yield obtained after the
purification and reformulation in saline. The final radio-
chemical yield was determined by comparing the decay-
corrected radioactivity of the purified product with the initial
fluoride radioactivity loaded onto the chip for synthesis.
Extraction
The final reaction mixture from the EWOD chip was manu-
ally extracted by pipette after removing the electrical ground
plate and washing both plates with solvent (i.e., MeCN,
This journal is © The Royal Society of Chemistry 2014
Lab Chip, 2014, 14, 902–910 | 907

Paper
Lab on a Chip
908 | Lab Chip, 2014, 14, 902–910
This journal is © The Royal Society of Chemistry 2014

Lab on a Chip
Paper
to be reduced with integrated reagent delivery and further
automation in the electrowetting and heating control.32
Results and discussion
Radiotracer synthesis yield
Upon setting the main reaction parameters of reagent, con-
centration, temperature, and time, the reaction efficiencies
for each radiotracer synthesis process were reliable and gen-
erally high (Table 1). The four example tracers demonstrated
here include both aliphatic nucleophilic fluorination reac-
tions as well as aromatic nucleophilic fluorination, which
generally require harsher reaction conditions. Compared with
preliminary results,22,25 the fluorination efficiency has
improved from 80% to 90% or greater for all four tracers.
Purified radiochemical yield for all four tracers is presented
for the first time on the right-most column of Table 1.
As summarized in the 6th column of Table 1, the
combined reaction efficiency (decay corrected) was 93% 3%
(n = 2) for [18F]FDG, 95% 3% (n = 6) for FLT, 85% 5%
(n = 3) for [18F]SFB, and 90% 9% (n = 6) for [18F]fallypride.
These reaction efficiencies were comparable to or exceeded
macroscale methods.
Extraction efficiency varied with each radiotracer's synthesis:
63% 8% (n = 2) for [18F]FDG, 81% 5% (n = 6) for [18F]FLT,
80% 17% (n = 3) for [18F]SFB, and 94% 3% (n = 6) for
[18F]fallypride. Although the extraction efficiency was comparable
with other microfluidic synthesizers, it was poor compared to
macroscale synthesizers where extraction efficiency is typically
greater than 95%. During the extraction step, 0–3% of the
radioactivity was lost to the pipette tip and the remaining
radioactivity was left on the actuation and cover plates.
Before-purification yields were determined by multiplying
combined reaction efficiency by measured extraction radioac-
tivity and dividing it by the initially loaded radioactivity.
Compared with the before-purification yields presented in
ref. 25 (50% for [18F]FLT and 72% for [18F]fallypride), the
current before-purification yields (73% for [18F]FLT and 84%
for [18F]fallypride) have improved by 45% for [18F]FLT and
16% for [18F]fallypride.
Conclusions
We have developed an EWOD radiosynthesizer in an effort
to create an affordable and flexible platform to enable researchers
and clinicians to produce their own tracer of interest on demand.
To demonstrate the success of the effort, we have successfully
performed several classes of radiochemical reactions (including
aliphatic nucleophilic substitution, aromatic nucleophilic substi-
tution, hydrolytic deprotection, saponification, and esterifica-
tion) using one EWOD chip design. Our goal is a universal chip
that can produce the radiotracer of choice as long as reagents
are provided. A mass-produced universal chip may be dispos-
able and fresh for each radiosynthesis, allowing imaging
centers to produce multiple, different radiotracers back-to-back
by simply swapping the chips and reagents.
To date, [18F]FDG and [18F]NaF are the only two radiotracers
that are routinely available from commercial radiopharmacies.
Other PET tracers are highly expensive or only available from
specialized radiochemistry facilities. Several research groups
have advanced microfluidic radiosynthesizer technology in
recent years, and the first microfluidic device for clinical
production of human doses of 18F-labeled PET tracers was
presented in early 2013.30
The EWOD radiosythesizer is distinctive as the only batch
reactor using droplets in air or an inert gas, which provides
unique advantages in its flexibility for radiosynthesis (because
fluid paths are software-programmable) and greatly facilitates
evaporation steps. The EWOD radiosynthesizer also has the
potential to produce radiotracers with unprecedented high spe-
cific radioactivity. We will investigate various parameters in the
microdroplet radiochemistry on EWOD chip to further under-
stand the factors that influence specific activity and develop a
generic EWOD radiosynthesizer to synthesize radiotracers with
ultrahigh specific activity.
The EWOD chip can use different solvents and shorter reaction
times than conventional synthesizers, but because it is open to
air, care must be taken to not lose radioactivity as volatile side
products and intermediates. Also because the reactor was not
sealed, the droplet size reduced during heating, even if the tem-
perature was below the boiling point. Attempts were made to keep
a stable concentration by replenishing the volume with smaller
solvent droplets, but we found that the reaction yields were higher
with shorter heating times when the droplet size was reduced,
which resulted in an overall increase in the concentration of
the reagents.21
Acknowledgements
This work was supported in part by the UCLA Foundation from
a donation made by Ralph & Marjorie Crump for the UCLA
Crump Institute for Molecular Imaging and the Department of
Energy [DE-SC0005056]. We thank Prof. David Stout for use of
radiolabeling facilities, Dr. Wyatt Nelson for developing
the multi-functional EWOD electrodes, Alex Dooraghi and
Prof. Arion Chatziioannou for developing the Cerenkov setup,
and Prof. Saman Sadeghi and Bob Silverman for developing
the multi-channel temperature controller and software.
Although reaction times are typically shorter when performed
with microfluidic volumes,4 the overall radiotracer synthesis time
(from initial loading to extraction) on the EWOD device was
longer than desired and ranged from 31
1 min for
Notes and references
[18F]fallypride (n = 6) to 58 8 min for [18F]SFB (n = 3). During
the developmental period, the synthesis times were longer
due to tedious process of manual reagent loading and manual
activation of electrodes. The overall synthesis time is anticipated
1 M. E. Phelps, Proc. Natl. Acad. Sci. U. S. A., 2000, 97, 9226–9233.
2 K. Serdons, A. Verbruggen and G. M. Bormans, Methods,
2009, 48, 104–11 .
This journal is © The Royal Society of Chemistry 2014
Lab Chip, 2014, 14, 902–910 | 909

Paper
Lab on a Chip
3 A. M. Elizarov, Lab Chip, 2009, 9, 1326–1333.
4 L. S. Cai, S. Y. Lu and V. W. Pike, Eur. J. Org. Chem., 2008, 17,
2853–2873.
20 D. Witters, N. Vergauwe, R. Ameloot, S. Vermeir, D. De Vos,
R. Puers, B. Sels and J. Lammertyn, Adv. Mater., 2012, 24,
1316–1320.
5 R. Iwata, Radiosynthesis database of PET probes, National
Institute of Radiological Sciences, Japan, 2013. Retrieved
from http://www.nirs.go.jp/research/division/mic/db2/.
6 C. H. Hsieh, Positron Emission Tomography – Current Clinical
and Research Aspects, InTech, New York, United States, 2012,
pp. 153–182.
21 P. Y. Keng, S. Chen, H. J. Ding, S. Sadeghi, G. J. Shah,
A. Dooraghi, M. E. Phelps, N. Satyamurthy, A. F. Chatziioannou,
C.-J. Kim and R. M. van Dam, Proc. Natl. Acad. Sci. U. S. A.,
2012, 109, 690–695.
22 S. Chen, R. Javed, J. Lei, H.-K. Kim, G. Flores, R. M. van Dam,
P. Y. Keng and C.-J. Kim, Tech. Dig. Solid-State Sensor and
Actuator Workshop, Hilton Head Island, SC, United States,
189–192, 2012.
7 E. Webster, M. S. Haka and P. Gutierrez, et al. J. Labelled
Compd. Radiopharm., 2011, 54, S438.
8 R. Krasikova, in PET Chemistry – The Driving Force in Molecular
Imaging, Springer, Berlin, 2007, ch. 11, pp. 289–316.
9 G. Pascali, P. Watts and P. A. Salvadori, Nucl. Med. Biol.,
2013, 40, 776–787.
10 C. Rensch, A. Jackson, S. Lindner, R. Salvamoser, V. Samper,
S. Riese, P. Bartenstein, C. Wangler and B. Wangler, Molecules,
2013, 18, 7930–7956.
11 S. Y. Lu, A. M. Giamis and V. W. Pike, Curr. Radiopharm.,
2009, 2, 49–55.
12 M. Yu, J. Matteo, D. Townsend and R. Nutt, J. Nucl. Med.,
2006, 47, 159P.
13 J. Lee, H. Moon, J. Fowler, T. Schoelhammer and C.-J. Kim,
Sens. Actuators, A, 2002, 95, 259–268.
14 K. Choi, A. H. C. Ng, R. Fobel and A. R. Wheeler, Annu. Rev.
Anal. Chem., 2012, 5, 413–440.
15 M. J. Jebrail, M. S. Bartsch and K. D. Patel, Lab Chip, 2012, 12,
2452–2463.
16 W. Nelson and C.-J. Kim, J. Adhes. Sci. Technol., 2012, 26,
1747–1771.
23 M. R. Javed, S. Chen, J. Lei, J. Collins, M. Sergeev, H.-K. Kim,
C.-J. Kim, R. M. van Dam and P. Y. Keng, Chem. Commun.,
2014, DOI: 10.1039/C3CC47616B.
24 C. Rensch, B. Waengler, A. Yaroshenko, V. Samper,
M. Baller, N. Heumesser, J. Ulin, S. Riese and G. Resichl,
Appl. Radiat. Isot., 2012, 70, 1691–1697.
25 H. K. Kim, S. Chen, R. Javed, J. Lei, C.-J. Kim, P. Y. Keng and
R. M. van Dam, Proc. Int. Conf. Miniaturized Systems for
Chemistry and Life Sciences, Okinawa, Japan, 617–619, (2012).
26 S. Chen, P. Y. Keng, G. J. Shah, R. M. van Dam and
C.-J. Kim, Proc. Int. Conf. MEMS, Cancun, Mexico, 980–983,
(2011).
27 W. C. Nelson, I. Peng, G.-A. Lee, J. A. Loo, R. L. Garrell and
C.-J. Kim, Anal. Chem., 2010, 82, 9932–9937.
28 M. S. Berridge, S. M. Apana and J. M. Hersh, J. Labelled
Compd. Radiopharm., 2009, 52, 543–548.
29 A. A. Dooraghi, P. Y. Keng, S. Chen, M. R. Javed, C.-J. Kim,
A. F. Chatziioannou and R. M. van Dam, Analyst, 2013, 138,
5654–5664.
17 S. K. Cho, H. J. Moon and C.-J. Kim, J. Microelectromech.
Syst., 2003, 12, 70–80.
18 J. Gong and C.-J. Kim, Lab Chip, 2008, 8, 898–906.
19 S. Sadeghi, H. J. Ding, G. J. Shah, S. Chen, P. Y. Keng,
C.-J. Kim and R. M. van Dam, Anal. Chem., 2012, 84,
1915–1923.
30 A. Lebedev, R. Miraghai, K. Kotta, C. E. Ball, J. Zhang,
M. S. Buchsbaum, H. C. Kolb and A. Elizarov, Lab Chip,
2013, 13, 136–145.
31 A. J. Parker, Chem. Rev., 1969, 69, 1–32.
32 G. J. Shah, H. J. Ding, S. Sadeghi, S. Chen, C.-J. Kim and
R. M. van Dam, Lab Chip, 2013, 13, 2785–2795.
910 | Lab Chip, 2014, 14, 902–910
This journal is © The Royal Society of Chemistry 2014

Copyright of Lab on a Chip - Miniaturisation for Chemistry & Biology is the property of
Royal Society of Chemistry and its content may not be copied or emailed to multiple sites or
posted to a listserv without the copyright holder's express written permission. However, users
may print, download, or email articles for individual use.