Automated solid-phase radiofluorination using polymer-supported phosphazenes

Article abstract of DOI:10.3390/molecules180910531

The polymer supported phosphazene bases PS-P₂ ^tBu and the novel PS-P₂ ^PEG allowed for efficient extraction of [¹⁸F]F^- from proton irradiated [¹⁸O]H ₂O and subsequent radiofluorination of a broad range of substrates directly on the resin. The highest radiochemical yields were obtained with aliphatic sulfonates (69%) and bromides (42%); the total radiosynthesis time was 35-45 min. The multivariate analysis showed that the radiochemical yields and purities were controlled by the resin load, reaction temperature, and column packing effects. The resins could be reused several times with the same or different substrates. The fully automated on-column radiofluorination methodology was applied to the radiosynthesis of the important PET radiotracers [¹⁸F]FLT and [¹⁸F]FDG. The latter was produced with 40% yield on a 120 GBq scale and passed GMP-regulated quality control required for commercial production of [¹⁸F]FDG. The combination of compact form factor, simplicity of [¹⁸F]F^- recovery and processing, and column reusability can make solid phase radiofluorination an attractive radiochemistry platform for the emerging dose-on-demand instruments for bedside production of PET radiotracers.

Full text of DOI:10.3390/molecules180910531

Molecules 2013, 18, 10531-10547; doi:10.3390/molecules180910531
OPEN ACCESS
molecules
ISSN 1420-3049
www.mdpi.com/journal/molecules
Article
Automated Solid-Phase Radiofluorination Using
Polymer-Supported Phosphazenes
Bente Mathiessen and Fedor Zhuravlev *
Hevesy Laboratory, Center for Nuclear Technologies, Technical University of Denmark, Risø Campus,
Frederiksborgvej 399, P.O. Box 49, 4000 Roskilde, Denmark; E-Mail: beem@dtu.dk
*
Author to whom correspondence should be addressed; E-Mail: fezh@dtu.dk;
Tel.: +45-4677-5337; Fax: +45-4677-5347.
Received: 24 July 2013; in revised form: 26 August 2013 / Accepted: 27 August 2013 /
Published: 30 August 2013
tBu
PEG
Abstract: The polymer supported phosphazene bases PS-P
2
and the novel PS-P
2
1
8
−
18
allowed for efficient extraction of [ F]F from proton irradiated [ O]H
2
O and subsequent
radiofluorination of a broad range of substrates directly on the resin. The highest
radiochemical yields were obtained with aliphatic sulfonates (69%) and bromides (42%);
the total radiosynthesis time was 35–45 min. The multivariate analysis showed that the
radiochemical yields and purities were controlled by the resin load, reaction temperature,
and column packing effects. The resins could be reused several times with the same or
different substrates. The fully automated on-column radiofluorination methodology was
1
8
18
applied to the radiosynthesis of the important PET radiotracers [ F]FLT and [ F]FDG.
The latter was produced with 40% yield on a 120 GBq scale and passed GMP-regulated
1
8
quality control required for commercial production of [ F]FDG. The combination of
1
8
−
compact form factor, simplicity of [ F]F recovery and processing, and column reusability
can make solid phase radiofluorination an attractive radiochemistry platform for the
emerging dose-on-demand instruments for bedside production of PET radiotracers.
Keywords: solid phase radiofluorination; dose-on-demand; polymer-supported phosphazene;
1
8
[
F]; multivariate analysis

Molecules 2013, 18
. Introduction
Over the years positron emission tomography (PET) has developed into one of the most successful
10532
1
−12
3
nuclear imaging modalities [1]. PET is a non-invasive, 10 M/cm -sensitive imaging method [2], able
to provide quantitative biomolecular information regarding physiological processes in real time [3].
PET is currently finding applications in cardiology, neuroscience, oncology, gene therapy [4] and drug
development [5]. Although more positron emitting radionuclides find their way into PET applications,
fluorine-18 remains the most used PET radioisotope [6] due to its optimal half-life (T1/2 = 109.8 min),
high-yielding positron decay (97%, no γ-emission) and a low and narrow energy spectrum
(Eavg = 249.3 keV, Emax = 633.5 keV), which translates into lower noise and higher image spatial
1
8
−
resolution [7]. In a typical PET radiopharmacy setting, 200 GBq of cyclotron-produced [ F]F would
be used to synthesize 15-20 doses of a given radiopharmaceutical and to ship it to a number of satellite
PET centers. Due to high cost of establishing and maintaining the production facilities as well as
difficulties associated with radiotracer development, the centralized PET radiopharmacies tend to have
a lean and expensive radiotracer “menu”. This limits the diversity of existing, and creates a barrier to
adoption of new PET radiotracers [8]. Allowing researchers and clinicians to produce a tracer of their
choice wherever and whenever there is a need for it, is a new paradigm often referred to as the
“
dose-on-demand” radiotracer production. Whether the tracer is produced from a radioisotope made at
a centralized facility, or made on-site using the emerging compact cyclotron/radiosynthesis instruments [9],
the dose-on-demand approach requires a significant change in current radiosynthesizer technology.
The preferred embodiment of the dose-on-demand concept is an affordably priced, GMP-compliant,
fully automated, compact (preferably table-top) instrument featuring: (1) single dose on demand tracer
delivery via tracer-specific kits; (2) ability to run custom radiosyntheses; (3) execution of back-to-back
production runs with minimal cassette replacement and radiation exposure to the operator;
(
4) integrated quality control (QC) module; (5) easy maintenance and operational simplicity. While
some of these challenges can be met with new engineering approaches [10], the development of
simpler and more efficient radiochemistry remains among the top priorities in radiochemical research.
1
8
−
In radiofluorination, the bottleneck is often considered to be the process of [ F]F recovery and
1
8
−
activation. In routine production, the nanomolar amounts of cyclotron-produced [ F]F are recovered
from several milliliters of its water solution by absorption onto an anion-exchange resin followed by
2 3
elution with aqueous acetonitrile/ K CO and Kryptofix. The subsequent azeotropic evaporations yield
1
8
the [ F]KF·Kryprofix complex in varying degrees of hydration and reactivity. The overall procedure
requires multiple liquid and gas transfers resulting in lengthening of synthesis, diminished yields, and
complex automation. Furthermore, the cartridge is not reusable and does not lend itself to efficient
miniaturization. Significant efforts focused on eliminating the azeotropic evaporation step, for
1
8
−
example, by using ionic liquids [11] or by eluting [ F]F with a strong organic base in the presence of
protic additives [12]. In the light of potential applicability to dose-on-demand radiosynthesis, a more
attractive approach would involve radiofluorination performed directly on the solid phase extraction
(
SPE) media. This would obviate the need for fluoride elution and azeotropic evaporation and allow
for seamless integration of potentially reusable SPE resin into an automated dose-on-demand
radiosynthesis module. The viability of radiofluorination at the point of fluoride capture was
previously demonstrated [13,14], although the non-reusable aminopyridine-based resin suffered from

Molecules 2013, 18
10533
18
−
poor stability and inconsistent [ F]F trapping and radiofluorination. Herein, we report fully
automated on-column radiofluorination of a wide variety of substrates using reusable polymer
supported phosphazenes.
2
. Results and Discussion
1
8
Et
We have recently reported that phosphazenium hydrofluoride [ F]P ·HF obtained by the reaction
2
Et
18
of P
2
(Scheme 1) with [ F]HF [15] is an efficient radiofluorination reagent (Scheme 2A) [16].
Scheme 1. Some phosphazene bases and substrates used in this study.
To access the radiolabeling ability of its polymer-supported analogue, we reacted the commercially
tBu
18
18
tBu
available polystyrene-supported PS-P2 with [ F]HF
2
trapping 90% of activity as [ F]PS-P ·HF.
(gas)
The subsequent reaction of the resin with the reference substrate Np(CH
2
)
2
OMs [16,17] (Scheme 1)
18
yielded the desired [ F]Np(CH ) F, albeit in only 5% radiochemical yield (RCY) (Scheme 2B).
2 2
Scheme 2. Radiofluorination using phosphazene bases.
RCY aside, we were encouraged by these results: based on Lemaire’s report,[12] one could expect
1
8
tBu
+
−
tBu
+
−
18
−
−
the formation of [ F]PS-P
2
+
H F from the putative PS-P
2
H OH via a [ F]F /OH anion
tBu
−
tBu
2
exchange, whereas PS-P
2
H OH would result from the reaction of PS-P
with water (Scheme 2C).
18
−
Therefore, we decided to treat the resin with aqueous [ F]F and after drying, use the resulting
1
8
−
[
F]F -loaded resin for radiofluorination directly on the solid support. Indeed, this strategy proved

Molecules 2013, 18
10534
tBu
18
−
fruitful, as PS-P2 was able to extract the [ F]F from the cyclotron-irradiated target water and
deliver it to a variety of substrates (Table 1).
tBu
Table 1. Manual on-column radiofluorination using PS-P
2
: Preliminary experiments.
1
8
−
[
F]F trapping
RCP
RCY
[
a]
Entry
Substrate
[
b]
[c]
[
%]
[%]
[%]
1
2
3
94
98
55
95
99
92
61
64
16
4
5
84
98
76
45
25
14
[
a]
tBu
New column for each experiment; Substrate: 0.2 mmol (52µmol for entry 4); PS-P
2
: 100 mg;
18
−
[
F]F : 200–800 MBq; Column drying: MeCN, 7 mL at 30 mL/h, 60 °C; Radiofluorination: toluene, 5 mL at
[b]
[c]
18
−
1
0 mL/h, 90 °C; RCP: radiochemical purity; RCY: decay-corrected radiochemical yield based on [ F]F (aq).
The preliminary experiments listed in Table 1 were performed manually, by sequential single
passes of the target water, acetonitrile and the solution of a substrate through a flexible PTFE tube
packed with the resin and immersed in an oil bath. We were pleased to see that both the aliphatic
(
entries 1–4) and the aromatic (entry 5) substrates could be radiofluorinated. On the other hand, the
fluoride trapping, RCP and RCY varied significantly across the substrates/runs. While some
variabilities were undoubtedly substrate-related, others could be due to operational variations,
warranting systematic screening and optimization studies performed with automated equipment. To
that end we prepared a column consisting of a short glass tube fitted with standard SwageLok fittings
tBu
and packed with PS-P
2
resin and glass beads to decrease back pressure. The column was connected
to our home-made automated apparatus, where the necessary chemicals, solvents and gases could be
remotely controlled and manipulated in a self-shielded and reproducible environment (Figure 1). Since
the column could be as short as 5 cm, the size of the apparatus was limited only by the peripheral
equipment, such as syringe pumps, valves and tubing.
In a typical automated experiment the target water was passed through the column, resulting in
1
8
−
absorption of [ F]F on the resin. The residual water was then removed from the column by rinsing of
the column with dry MeCN. Further drying could be assisted by passing a flow of He through the
column at 60 °C. Finally, the solution of substrate was passed through the column at a suitable flow
1
8
−
rate and temperature, effecting on-column radiofluorination. If necessary, the unreacted [ F]F could
be removed from the product using a silica Sep-Pak cartridge (Figure 1). At the outset we recognized
that the screening and optimization studies would require the exploration of the experimental domain
defined by various physical and chemical parameters. While changing one variable at a time is a
common strategy, a more efficient approach is the use of multivariate experimental design known as
the design of experiments (DoE) [18]. In DoE the experimental conditions (factors) are varied
systematically and simultaneously according to a DoE algorithm and the outcome of the experiments

Molecules 2013, 18
10535
(
responses) are measured, thereby allowing for a statistically driven analysis of data in an optimal
number of experiments. The final result is a model which describes the relationships between the
factors and responses. The trapping efficiency of the resin was analysed first by varying the amount of
resin (5–500 μmol), the volume of target water (0.5–3.5 mL), the flow rate (1.5–30 mL/min) and the
number of times the column was used. The reaction matrix suggested by the DoE (Table 2) yielded an
2
2
excellent single component partial least square (PLS) model (R = 0.91, Q = 0.89).
Figure 1. Automated radiosynthesis module for on-column radiofluorination.
1
8
−
tBu
[a]
Table 2. On-column [ F]F trapping using PS-P : trapping efficiency studies .
2
1
8
−
Number of times
Resin
[µmol]
5.8
Water
[mL]
0.5
0.5
0.5
0.5
3.5
3.5
3.5
3.5
30
Flow rate
[ F]F
[
b]
column was used
[mL/min]
1.5
trapping [%]
1
4
1
4
1
4
1
4
1
4
1
4
77
67
5.8
1.5
509
509
5.5
1.5
100
100
57
1.5
1.5
5.5
1.5
58
499
499
5.1
1.5
100
100
56
1.5
30
5.1
30
30
52
501
501
30
30
100
99
30
30
[
a]
[b]
Full factorial design at two levels reduced to 12 experiments and analyzed by PLS; after each trapping
the column was washed with dry MeCN (50 mL, flow rate 1.5 mL/min, rt) and dried with a flow of He
flow rate: 100 mL/min, 20 min).
(

Molecules 2013, 18
10536
The variables important for the projection (VIP) plot indicated that the only factor controlling the
trapping efficiency was the amount of resin (Figure 2). The column could be reused at least 3 times
1
8
−
without any loss in [ F]F trapping efficiency and without resin degradation.
1
8
−
tBu
Figure 2. On-column [ F]F trapping using PS-P : variables important for the
2
projection (VIP) plot.
We found that the minimal amount of resin necessary to obtain 99% trapping efficiency was
6
1 μmol (Figure 3).
1
8
−
tBu
Figure 3. On-column [ F]F trapping using PS-P
2
: Correlation between resin amount
1
8
and [ F]fluoride trapping efficiency.
Although it is well known that hydration reduces fluoride nucleophilicity, some radiofluorination
reactions show certain tolerance for water [11,12,16]. Since residual water entrapped in the resin could
potentially influence both RCY and the radiochemical purity (RCP) we have attempted to optimize our
drying protocol, varying the nature, the flow rate and the volume of the drying solvent, as well as the
drying temperature (Table 3).

Molecules 2013, 18
Table 3. Effect of column drying on radiofluorination of Np(CH ) OMs with PS-P
2
10537
tBu
2
2
.
[
a]
[b]
[c]
Entry
Drying solvent Vol [mL] Flow rate [mL/min] Drying T [°C] RCP [%]
RCY [%]
1
2
3
4
5
6
7
8
9
MeCN
Acetone
MeCN
Acetone
MeCN
Acetone
MeCN
Acetone
MeCN
MeCN
MeCN
1
1.5
25
50
50
25
50
25
25
50
38
38
38
88
77
77
75
74
90
85
88
85
83
91
32
26
18
24
20
23
32
34
25
30
31
1
1.5
10
10
1
1.5
1.5
20
1
20
10
10
5.5
5.5
5.5
20
20
10.75
10.75
10.75
1
1
0
1
[
a]
tBu
2
18
−
New column for each experiment; PS-P
: 180 µmol; [ F]F : 100-1000 MBq; Radiofluorination:
[
b]
Np(CH ) OMs: 52 µmol; toluene, 3 mL, flow rate 0.55 mL/min, 90 °C;
RCP: radiochemical purity;
2
2
[
c]
18
−
RCY: decay-corrected radiochemical yield based on [ F]F (aq)
.
Despite good reproducibility (Table 3, entries 9–11), the analysis of the data showed little
variability in RCY and did not result in a statistically valid model. No leaching of activity from the
column was observed upon drying. For practical reasons, the drying protocol was adjusted to the use of
MeCN (3–5 mL, 1.5 mL/min, 25 °C) for most of the following experiments.
The drying conditions established above were used to perform limited optimization studies where
we looked at the influence of the amount of resin and the reaction temperature on RCP and RCY
(Table 4). For consistency with our previous studies [16] the substrate loading was chosen at 52 μmol
as the preliminary experiments showed that the high (150 μmol), medium (77.5 μmol) and low (5 μmol)
substrate loading gave similar RCY.
Column packing effects, affecting the flow and likely to be responsible for some of the inter-
column variability we have seen, were left out of optimization, as we did not find a reliable way to
tBu
quantify them. It was also important to find out whether PS-P
2
resin could be reused not only at the
trapping stage (Table 2), but also after radiofluorination. To that end, the trapping and radiofluorination
sequence was repeated several times on the same column (columns A, C, D, and E, Table 4).
2
2
The optimization studies yielded a 2-component PLS model (R = 0.99. Q = 0.98) which indicated
a positive correlation for both the amount of resin and the reaction temperature. Higher amount of resin
and higher temperature (T) led to higher RCP and RCY (Figure 4, left). Interestingly, the cross term
resin* T was also important and positively correlated. The high covariance between the X-values (T, resin)
and the Y values (RCY, RCP) can be clearly seen in the bi-plot (Figure 4, right) as the factors (X) and
responses (Y) are grouped together. The RCP and RCY for columns A and B, and to a lesser extent C,
are dominated by the resin and resin* T terms. The T variable overlaps with RCP and RCY, suggesting
that the three are highly correlated. The cross term can be interpreted as interaction: the effect of T on
RCY and RCP depends on the amount of resin.

Molecules 2013, 18
Table 4. On-column radiofluorination of Np(CH ) OMs with PS-P : Optimization.
10538
tBu
2
2
2
[
a]
[b]
tBu
[c]
[d]
Entry
Column/run
PS-P
2
, [μmol] T [°C] Trapping [%] RCP [%]
RCY [%]
1
2
3
4
5
6
7
8
9
A/1
A/2
B
180
180
180
140
140
100
100
100
60
90
100
100
100
100
100
98
79
79
84
74
94
80
89
93
82
88
38
39
50
43
69
46
62
64
49
63
90
120
120
120
120
120
120
120
120
C/1
C/2
D/1
D/2
D/3
E/1
E/2
100
100
98
1
0
60
100
[
a]
18
− 18
Trapping: [ F]F /[ O]H
2
O, 200–800 MBq, 3mL, 1.5mL/min; Drying: MeCN, 3mL, 1.5mL/min;
[
b]
Radiofluorination: Np(CH ) OMs: 52 μmol; toluene, 3 mL, 0.55 mL/min, 90 °C;
Letter-coded columns
2
2
tBu
A–E differed in PS-P2 loading and reaction temperature, the Arabic numerals refer to the number of
[
c]
consecutive trapping and radiofluorination runs performed on that particular column; RCP: radiochemical
[
d]
18
−
purity; RCY: decay-corrected radiochemical yield based on [ F]F (aq)
.
Importantly, Table 4 showed a consistent increase in the RCY and RCP when the columns were
reused (columns A, C, D and E). The highest RCY and RCP (69% and 94%) were obtained on the
18
second use of column C. Radio-TLC analysis showed that during radiosynthesis the only F-containing
1
8
18
materials found in the reaction mixture eluted from the column were [ F]Np(CH
2
)
2
F and [ F]fluoride
(
[
Figure 5). The formation of the latter was consistent with the substrate-dependent elimination of
1
8
18
−
F]HF triggered by the basicity of [ F] F , as we have demonstrated for homogeneous fluorination
with phosphazene hydrofluorides [16].
tBu
Figure 4. On-column radiofluorination of Np(CH
2
)
2
OMs using PS-P
2
: variables
important for the projection (VIP) plot (left) and bi-plot of loadings and scores (t), (right).

Molecules 2013, 18
Figure 5. On-column radiofluorination of Np(CH ) OMs with PS-P : Radio-TLC.
10539
tBu
2
2
2
The remaining activity was retained on the column and could be removed neither with the fresh
portion of substrate nor with a variety of solvents, such as toluene, DMF or water. Increasing
2 2
substrate’s on-column residence time by recirculating the solution of Np(CH ) OMs did not lead to
any appreciable increase in RCY. The activity retained on the column after radiofluorination could
indicate a physical entrapment of small fluoride anion within the polymer matrix, or a chemical
deactivation due to reaction with the resin or impurities inside the resin.
Next, we turned to investigation of polymer matrix effects. The mobility of polymer chains and
hence the accessibility of functional groups depends on polymer swelling. The highly cross-linked
tBu
PS-P
2
used in this study showed negligible swelling in MeCN, toluene and DMF. Therefore,
resin, which is
based on PEG-grafted polystyrene matrix offers a high and largely solvent-independent swelling. From
swellable polymer matrix was sought after. The commercially available TentaGel-NH
2
the design point of view the TentaGel-NH
2
resin was also attractive as its alkyl-amino side chains
PEG
could be easily functionalized with phosphazene residues. Consequently, the desired PS-P
2
was
synthesized by applying the Schwesinger’s protocol [19] to the Tentagel –NH resin (Scheme 3). Both
2
tBu
PEG
PS-P
2
and PS-P
2
could be stored at room temperature indefinitely.
PEG
The resulting PS-P
2
2 2
(95% phosphazene-functionalized) was reacted with Np(CH ) OMs under
tBu
the conditions optimized for PS-P
2
. Unfortunately, due to its high swelling ability the microgel was
prone to trapping air bubbles disrupting the uniform flow of the substrate through the column.
Although this was partially rectified by careful degassing and recirculation of the starting material, the
tBu
PEG
2
RCY topped at 38% (Table 5, entry 1). Similar to PS-P
sequential radiofluorination of Np(CH OMs and Np(CH
F in 25% and 24% RCY correspondingly.
2
, the PS-P
resin could be reused: the
2
)
2
2 2
) OTs performed on the same column gave
1
8
2 2
the desired [ F]Np(CH )

Molecules 2013, 18
10540
PEG
Scheme 3. Synthesis of PS-P2
.
Table 5 summarizes the scope of the polymer supported bases used for radiofluorination of
tBu
PEG
Np(CH
2
)
2
OMs. PS-P
2
gave the best RCY at 69% (entry 2) followed by P-P
2
at 38% (entry 1).
1
8
−
Both resins trapped 99% of [ F]F . The polymer-supported DIPEA (entry 3) and BEMP (entry 4)
failed to trap fluoride, consistent with their lower pKa (11.4 and 27.6, correspondingly).
2 2
Table 5. Radiofluorination of Np(CH ) OMs: the scope of the polymer supported bases.
[
a]
18
−
[b]
Entry
Structure
[ F]F trapping [%] RCY [%]
1
99
38
2
99
69
[
c]
3
4
—
—
—
—
[a]
18
− 18
New column for each experiment; Trapping: [ F]F /[ O]H
2
O, 500–1100 MBq, 3 mL, flow rate 1.5 mL/min;
Drying: MeCN, 3 mL, flow rate 1.5 mL/min; Radiofluorination: Np(CH ) OMs: 52 µmol; toluene, 3 mL,
2
2
[b]
18
−
[c]
0
.55 mL/min, 90 °C; RCY: decay-corrected radiochemical yield based on [ F]F (aq)
;
below detection limit.

Molecules 2013, 18
The optimized conditions established for PS-P2 and PS-P2
10541
tBu
PEG
were used to investigate the
substrate scope. The results are presented in Table 6. The reactivity profile of both resins was similar
1
8
Et
to that observed for homogeneous radiofluorination with [ F]P
2
HF.[16] The sulfonates (Table 6,
tBu
entries 1,2, and 7) were the best substrates for on-column radiofluorination with PS-P
and
2
PEG
2
18
PS-P
, delivering tetra-O-acetyl-[ F]FDG in 52% and 36% RCY, correspondingly (Table 5, entry 7).
tBu
PEG
The halides could be radiofluorinated with PS-P
2
, but not with PS-P
2
(Table 6, entries 3–5). The
corresponding yields of FLT were only 7% and 5%, which is at the lower end of the currently reported
homogeneous radiofluorination yields (8%–50%) [20]. The total radiosynthesis time for any substrate
did not exceed 45 min.
tBu
PEG
Table 6. Radiofluorination with P-P2 and P-P2 : The substrate scope.
[
a]
[b]
[c]
Entry
Substrate
Trapping
RCP
RCY
tBu
PEG
tBu
PEG
tBu
PEG
PS-P2 /PS-P2
PS-P2 /PS-P2
[%]
PS-P2 /PS-P2
[%]
[
%]
1
2
3
4
99/99
94/100
94/100
95/99
97/96
86/96
73/-
51/38
34/24
18/-
93/-
42/-
5
6
95/95
67/-
60/-
16/-
23/-
92/100
7
8
99/98
99/97
65/92
52/36
7/5
100/100
18
−
18
[
a] New column for each experiment; Trapping: [ F]F /[ O]H
2
O, 500–5000 MBq, 3 mL, flow rate 1.5 mL/min;
Drying: MeCN, 3 mL, 1.5 mL/min; Radiofluorination: Substrate: 100 µmol; in toluene at 120 °C (aromatic
substrates), tBuOH: MeCN 5:1 at 100 °C (FLT) or MeCN at 85 °C (FDG), flow 0.55 mL/min; Total
radiosynthesis time between 35 and 45 min. [b] RCP: radiochemical purity; [c] RCY: radiochemical yield
18
−
based on [ F]F
aq)
as an average of two runs.
(
Finally, the on-column solid phase radiofluorination was applied to the automated radiosynthesis of
F]FDG on a 120 GBq scale typically used for the commercial production of this radiotracer in our
1
8
[
laboratory. The apparatus was adopted to accommodate the hydrolysis step (Figure 6).
tBu
18
−
18
Using PS-P
0% RCY and 96.7% RCP as measured by the radio-HPLC and radio-TLC. The specific activity was
measured to be higher than 11 GBq/μmol, the limit of detection of cold FDG by our HPLC.
2
we achieved 98% of [ F]F trapping efficiency obtaining the desired [ F]FDG in
4
1
8
Importantly, the resulting formulation of on-column produced [ F]FDG met the stringent GMP
1
8
quality control requirements used in our laboratory to release the commercial [ F]FDG produced by
the conventional methodology (Table 7).

Molecules 2013, 18
Figure 6. Automated radiosynthesis module for on-column radiosynthesis of [ F]FDG.
10542
1
8
18
Table 7. Automated on-column radiosynthesis of [ F]FDG on 120 GBq scale: Quality Control.
Analysis
Result
0.5%
Requirement (European Pharmacopoeia)
1
8
−
[
F]F
<5%
1
8
[
[
F]FDM
ND
<10%
1
8
F]FDG
97.8%
≥85%
1
8
Unhydrolyzed [ F]FDG
RCP
1.0%
<5%
≥95%
96.7%
1
9
[
F]FDG
<0.0167mg/mL
0.13mg/mL
Clear, colorless
≤0.0167 mg/mL
No limit
Glucose
Appearance
Clear, colorless or slightly yellow
3
. Experimental
3.1. General
Unless otherwise noted, all synthetic steps were performed under an inert atmosphere of argon.
Glassware and reaction vessels were dried in an oven at 160 °C overnight or flame-dried on the
Schlenk line before use. All reagents were used as received without additional purification; toluene and
THF were distilled from sodium benzophenone; other solvents were dried over activated molecular
tBu
sieves. All radiochemical yields were decay-corrected. Polymer-supported P2 (loading: 1.6 mmol/g),
2
2
-tert-Butylimino-2-diethylamino-1,3-dimethylperhydro-1,3,2-diazaphosphorine (BEMP, loading:
.3 mmol/g) and diisopropylaminoethyl (DIPEA, loading: 3 mmol/g) were obtained from Sigma
Aldrich; the amine-functionalized polystyrene resin with polyethylene glycol spacer (TentaGel HL,
particle size 160 µm, loading: 0.4 mmol/g, PEG: 1500–2000 Da) was purchased from Rapp Polymere

Molecules 2013, 18
10543
GmbH. Mannose triflate was purchased from Sigma Aldrich and 3-N-Boc-5′-O-dimethoxytrityl-3′-O-
nosyl-thymidine (FLT precursor) was purchased from ABX. The naphthalene halides or pseudohalides
were purchased from Sigma Aldrich or synthesized as described elsewhere.
3.2. Instrumentation
1
13
31
H, C and P-NMR spectra were recorded on a Bruker Avance II 500 instrument operating at
1
9
5
00, 126 and 202 MHz, respectively. F-NMR was recorded on a Brucer Avance DPX 250 instrument
at 235 MHz. Mass spectrometry was performed on a Bruker Esquire 4000 ion-trap (IT) spectrometer
equipped with electrospray ionization (ESI) interface. Thin-layer chromatography (TLC) was run on
pre-coated plates of silica gel 60, F254 (Merck). Radio-HPLC was performed by using a Knauer
HPLC System K501, equipped with a Knauer RI detector K2301 and CRA radioactivity detector 105
−1
S-1 on a Carbopac PA10 4_25 mm Dionex column eluted with 0.1M NaOH at 1.0 mL min . Radio-TLC
was performed with a Raytest MiniGita TLC scanner. The aqueous solutions of fluoride-18 were
1
8
18
prepared by the O(p,n) F reaction in a GE PETtrace cyclotron using a Ag or Nb target containing
1
8
~
2.5 mL of 95%–98% enriched [ O]H
2
O irradiated by a 16.5 MeV proton beam at 55 µA. For
automated radiosynthesis we used custom-made radiosynthesis robot controlled by LabView software.
3.3. Multivariate Analysis
Multivariate analysis was performed using MODDE 9.0 and SIMCA 13 software. The raw data
were mean-centered and Pareto scaled; the Hotteling T2 values were below 95% for all models.
PEG
3
.4. Synthesis of PS-P
2
Cl
Cl
TentaGel-NH
2
was reacted with P
2 4 2 4
*BF [19] by mixing the resin (200–1000 mg) with P *BF
(
3
3 eq. relative to the amine) and Et N (dry, 9 eq relative to the amine) in DCM (dry, 5 mL) in a sealed
glass ampule. The reaction mixture was then heated and shaken at 90 °C for three days. The procedure
was repeated one more time in order to ensure better coupling. The resulting resin was deprotonated by
reacting the resin with a mixture of KOMe (1 eq to amine) in MeOH (dry, 5 mL) for one hour giving
PEG
the desired PS-P
2
. The degree of resin functionalization by the phosphazene residues was estimated
1
9
18
19
using [ F/ F]fluoride trapping in the following way: an aqueous solution of [ F]NaF (42.54 mg,
1
8
−
19
−
−1
1
.013 mmol) and [ F]F (82.85 MBq) in 3.4 mL, corresponding to 0.012 mmol [ F]F MBq was
PEG
−1
passed through a column containing PS-P
2
(204.58 mg, original amine content 0.4 mmol g ) at
1
mL/min; the column was then washed with water (4 mL, flow 1 mL/min). The activity retained on
1
8
−
19
−
the column was measured at 6.49 MBq [ F]F , which corresponded to 0.078 mmol [ F]F , and
consequently, to 0.38 mmol of phosphazene residues per gram of resin (95% of the original amine
functionalization).
1
8
3
.5. [ F]Fluoride Trapping, RCP and RCY
1
8
[
F]fluoride trapping [%] was based on target water and calculated as: Trapping = 100%*
added to the column
eluted from the column
added to the column
(
Activity
– Activity
)/Activity
. The RCP was determined by

Molecules 2013, 18
10544
product
radio-TLC or radio-HPLC and calculated as: RCP = (Area
/Total Area)*100%. The RCY was
product
[18F]F− added to the column
determined by the dose calibrator and calculated as RCY=Activity
*RCP/Activity
.
1
8
18
tBu
3
.6. Manual Radiofluorination of [ F]Np(CH
2
)
2
OMs with [ F]HF and PS-P
2
(Scheme 2, B)
tBu
A PTFE tube (OD = 0.64 cm) was packed with PS-P2 (100–200 mg) and 1200 mg of glass beads
and connected to a PE vial containing 4 mL of 98% H
2 4
SO . The PE vial was connected to an Ar inlet
18
−
and Ar was bubbled through sulphuric acid at a rate of 150–300 scc/min. [ F]F
(1 mL, 200–500 MBq)
(aq)
was added and the vial was heated for 30 min at 80 °C in an ultrasound bath while being irradiated at
18
3
5 kHz under the flow of Ar. The resulting [ F]HF was carried by argon flow to the column. After 30 min
the column was placed in an oil bath at 100 °C and a solution of NpEtOMs (200 µmol) in toluene (dry,
mL) was passed through the column using a syringe pump (flow rate 10 mL/h). Toluene (dry, 5 mL)
5
was passed through the column to elute the residual product. The fluorinated product was analyzed by
radio-TLC (eluent heptane:EtOAc 80:20). Trapping = 95%, RCP = 73%, RCY = 5%. Total time of
radiosynthesis was up to 90 min.
3
.7. General Procedure for Manual Solid-Phase Radiofluorination (Scheme 2 C and Table 1)
1
8
−
An aliquot of [ F]F
aq)
(1.0 ml, 200–800 MBq) was mixed with water (3 mL) and passed through
(
tBu
the column (PTFE tube, PS-P
2
: 100 mg; glass beads: 1200 mg). MeCN (dry, 7 mL) was then passed
through the column (flow rate: 30 mL/h, duration 10 min, rt to 60 °C) followed by Ar until excess of
solvent was removed. A substrate (52–200 µmol) was dissolved in toluene (dry, 5 mL) and the solution
was passed through the column (flow rate 10 mL/h, duration 30 min) while heating the column at 90 °C.
Thereafter, toluene (dry, 5 mL) was passed through the column (flow rate 20 mL/h, duration 15 min) at
9
0 °C to elute the remaining product. The fluorinated products were analyzed by radio-TLC (eluent
heptane:EtOAc 80:20 for the naphthalene derivatives, MeCN:H
2
O 95:5 for Ac
4
-FDG, DCM:MeOH
9
:1 for hydrolyzed FLT, petroleum ether:EtOAc 3:1 for the pyridine derivative) Total time of
radiosynthesis was 60–90 min.
tBu
3
.8. General Procedure for Automated Solid-Phase Radiofluorination Using PS-P
2
1
8
−
An aqueous solution of [ F]F (0.5–3.5 mL, flow rate 0.5–30 mL/min) was passed through the
column (borosilicate glass tubing, ID 4 mm, OD 0.6 cm, length 12 cm) packed with the resin
5–500 µmol; glass beads to fill the remaining volume). To remove the bulk of water, the column was
(
flushed with He(gas) (flow rate 100 mL/min, 3 min). The column was dried by passing MeCN or
acetone (1–10 mL, flow rate 1.5–20 mL/min) through, followed by a helium flush (100 mL/min,
2
–5 min). The column was primed by passing the radiofluorination solvent (MeCN for mannose
triflate, MeCN/tBuOH 1:5 for FLT-ONs, toluene for the naphthalene derivatives and 2-nitro-3-
methoxypyridine, dry, 4 mL, flow rate 2 mL/min) through the column at room temperature followed
by the substrate (50–100 µmol) dissolved in the radiofluorination solvent (dry, 3 mL) at 60–120 °C
(
flow rate 0.55 mL/min). The radiofluorination solvent (dry, 2 mL, flow rate 0.55 mL/min) was then
passed through the column again to elute the remaining product. The reaction mixture was analyzed by
radio-TLC (eluent heptane:EtOAc 80:20 for the naphthalene derivatives, MeCN:H O 95:5 for FDG,
2

Molecules 2013, 18
10545
DCM:MeOH 9:1 for hydrolyzed FLT, petroleum ether:EtOAc 3:1 for the pyridine derivative. Total time
of radiosynthesis was 35–45 min.
PEG
3
.9. General Procedure for Automated Solid-Phase Radiofluorination Using PS-P2
The apparatus and column was degassed before use by passing through MeCN (5 mL, 1.5 mL/h)
1
8
−
followed by water (5 mL, 1.5 mL/h). An aqueous solution of [ F]F (3.5 mL, flow rate 1.5 mL/min)
was passed through the column (borosilicate glass tubing, I.D. 4 mm, OD 0.6 cm, length 12 cm)
packed with the resin (200–500 µmol). Residual water was removed by passing through MeCN (5 mL,
flow rate 1.5 mL/min). The column was primed by passing the radiofluorination solvent (MeCN for
mannose triflate, MeCN/tBuOH 1:5 for FLT-ONs, toluene for the naphthalene derivatives and 2-nitro-
3
-methoxypyridine, dry, 4 mL, flow rate 1.5 mL/min) through the column at room temperature
followed by the substrate (50-100 µmol) dissolved in the radiofluorination solvent (dry, 3 mL) at
0–120 °C (flow rate 0.55 mL/min). The radiofluorination solvent (dry, 2 mL, flow 0.55 mL/min) was
6
then passed through the column again to elute the remaining product. The reaction mixture was analyzed
by radio-TLC (eluent heptane:EtOAc 80:20 for the naphthalene derivatives, MeCN:H O 95:5 for FDG,
2
DCM:MeOH 9:1 for hydrolyzed FLT, petroleum ether:EtOAc 3:1 for the pyridine derivative. Total
time of radiosynthesis was up to 45 min.
3
.10. Washing Procedure for Back-to-Back Radiofluorination
After each trapping-radiofluorination sequence the column was washed with dry MeCN (50 mL,
flow rate 1.5 mL/min, rt) and dried with a flow of He (flow rate: 100 mL/min, 20 min).
1
8
tBu
2
3
.11. Large Scale Automated [ F]FDG Synthesis using P-P
Column
Mannose triflate together with the solvents and reagents used for hydrolysis of the product was
acquired from ABX Advanced Biochemical Compounds (Radeberg, Germany) and are of
pharmaceutical grade quality manufactured according to GMP standards. The cyclotron irradiated
1
8
−
target water containing [ F]F (3.0 mL, flow rate 1.5 mL/min, 119 GBq) was passed through the
tBu
column (borosilicate glass, PS-P
2
: 100 µmol; glass beads to fill the remaining volume) followed by
He(gas) (flow rate 100 mL/min, 2 min). MeCN (dry, 5 mL, flow 1.5 mL/min) was passed through the
column followed by He(gas) (100 mL/min, 3 min) in order to dry the column completely (trapped
1
8
−
1
17 GBq of [ F]F , 98% trapping efficiency). MeCN (dry, 4 mL, flow 2 mL/min) was passed through
the column at room temperature followed by mannose triflate (25 mg, 52 µmol) dissolved in MeCN
dry, 3 mL) at 85 °C (flow 0.55 mL/min). MeCN (dry, 2 mL, flow 0.55 mL/min) was then passed
(
through the column to elute the remaining product. The two fractions were mixed with water (total
volume 50 mL) and passed through a tC C-18 Sep Pak cartridge (preconditioned using 5.0 mL EtOH
1
8
and 5.0 mL water) to trap the unhydrolyzed [ F]FDG. The cartridge was washed with water (5 mL).
NaOH (2.0 M, 0.5 mL) was added to the cartridge in order to hydrolyze the compound (duration
5
min) and the hydrolyzed product was eluted with water (5 mL) to a product vial containing saline
(
9 mg/mL NaCl, 0.64% EtOH, total volume 14 mL) and a buffer solution (citrate buffer, 6 mL,
1
8
pH 1.3). Full QC analysis confirmed that the product was [ F]FDG (49 GBq, RCP 97%, RCY 41%).

Molecules 2013, 18
10546
4
. Conclusions
tBu
In conclusion, we have demonstrated that polymer supported phosphazenes PS-P
and the novel
2
PEG
18
−
PS-P
2
allowed for efficient extraction of [ F]F from the target water and subsequent
radiofluorination of a broad range of substrates directly on the resin, thus simplifying radiosynthesis
1
8
−
and QC by avoiding [ F]F elution, azeotropic evaporation and kryptand addition. The highest
radiochemical yields were obtained with aliphatic sulfonates (69%) and bromides (42%); the total
radiosynthesis time for any substrate did not exceed 45 min. We found that the RCY and RCP were
controlled by the amount of resin, reaction temperature, and column packing effects. The same resin
could be used at least 3 times with the same substrate and at least two times with two different
substrates. The incorporation of the compact phosphazene resin column into a remotely controlled
programmable dispensing system resulted in a solid phase radiosynthesis module which was used for
1
8
automated radiosyntheses, including FLT and a large scale production of [ F]FDG. Although at this
stage of development the solid phase RCY are generally ~10%–30% behind those obtained
1
8
homogeneously, the large scale production of GMP-quality [ F]FDG with 40% RCY and 97% RCP
clearly demonstrates the potential of solid phase radiofluorination methodology. We believe that the
1
8
−
combination of compact form factor, streamlined [ F]F recovery and processing, and column
reusability will make the solid phase radiofluorination an attractive radiochemistry platform for the
emerging dose-on-demand instruments for bedside production of PET radiotracers.
Supplementary Materials
Supplementary materials can be accessed at: http://www.mdpi.com/1420-3049/18/9/10531/s1.
Acknowledgments
This research was supported by Hevesy Laboratory. We thank Michael Jensen, Palle Rasmussen,
and Gregory Severin for discussions.
Conflicts of Interest
The authors declare no conflict of interest.
References
1.
2.
3.
4.
5.
Paans, A.M. J.; van Waarde, A.; Elsinga, P.H.; Willemsen, A.T.M.; Vaalburg, W. Positron emission
tomography: The conceptual idea using a multidisciplinary approach. Methods 2002, 27, 195–207.
Skotland, T. Molecular imaging: Challenges of bringing imaging of intracellular targets into
common clinical use. Contrast Media Mol. Imaging 2012, 7, 1–6.
Ametamey, S.M.; Honer, M.; Schubiger, P.A. Molecular imaging with PET. Chem. Rev. 2008,
1
08, 1501–1516.
Collins, S.A.; Hiraoka, K.; Inagaki, A.; Kasahara, N.; Tangney, M. PET Imaging for gene cell
therapy. Curr. Gene Ther. 2012, 12, 20–32.
Marik, J.; Bohorquez, S.M.S.; Williams, S.-P.; van Bruggen, N. New imaging paradigms in drug
development: The PET imaging approach. Drug Discovery Today 2011, 8, e63–e69.

Molecules 2013, 18
10547
6
.
Coenen, H.H.; Elsinga, P.H.; Iwata, R.; Kilbourn, M.R.; Pillai, M.R.A.; Rajan, M.G.R.;
Wagner, H.N., Jr.; Zaknun, J.J. Fluorine-18 radiopharmaceuticals beyond [18F]FDG for use in
oncology and neurosciences. Nucl. Med. Biol. 2010, 37, 727–740.
7
8
9
1
1
1
.
.
.
Sanchez-Crespo, A. Comparison of Gallium-68 and Fluorine-18 imaging characteristics in
positron emission tomography. Appl. Radiat. Isotopes 2013, 76, 55–62.
Keng, P.Y.; Esterby, M.; van Dam, R.M. Emerging technologies for decentralized production of
PET tracers. Positron Emiss. Tomogr. Clin. Res. Asp. 2012, 153–182.
Nutt, R.; Vento, L.; Ridinger, M. In Vivo molecular imaging biomarkers: Clinical pharmacology’s
new “PET”? Clin. Pharmacol. Ther. 2007, 81, 792–795.
0. Ming-Wei W. Wei-Yu L.; Kan L.; Masterman-Smith, M.; Kwang-Fu S.C. Microfluidics for
positron emission tomography probe development. Mol. Imaging 2010, 9, 175–191.
1. Kim, D.W.; Choe, Y.S.; Chi, D.Y. A new nucleophilic fluorine-18 labeling method for aliphatic
mesylates: Reaction in ionic liquids shows tolerance for water. Nucl. Med. Biol. 2003, 30, 345–350.
2. Lemaire, C.F.; Aerts, J.J.; Voccia, S.; Libert, L.C.; Mercier, F.; Goblet, D.; Plenevaux, A.R.;
Luxen, A.J. Fast production of highly reactive no-carrier-added [18F]fluoride for the labeling of
radiopharmaceuticals. Angew. Chem. Int. Ed. 2010, 49, 3161–3164.
1
1
3. Mulholland, G.K.; Mangner, T.J.; Jewett, D.M.; Kilbourn, M. R. Polymer-supported nucleophilic
radiolabeling reactions with [18F]fluoride and [11C]cyanide ion on quaternary ammonium resins.
J. Label. Compound. Radiopharm. 1989, 26, 378–380.
4. Toorongian, S.A.; Mulholland, G.K.; Jewett, D.M.; Bachelor, M.A.; Kilbourn, M.R. Routine
production of 2-deoxy-2-[18F]fluoro-d-glucose by direct nucleophilic exchange on a quaternary
4
-aminopyridinium resin. Int. J. Rad. Appl. Instrum. B 1990, 17, 273–279.
1
1
1
5. Mathiessen, B.; Jensen, M.; Zhuravlev, F. [18F]Fluoride recovery via gaseous [18F]HF. J. Label.
Compound. Radiopharm. 2011, 54, 816–818.
6. Mathiessen, B.; Jensen, A.T.I.; Zhuravlev, F. Homogeneous nucleophilic radiofluorination and
fluorination with phosphazene hydrofluorides. Chem-Eur. J. 2011, 17, 7796–7805.
7. Jadhav, V.H.; Jang, S.H.; Jeong, H.-J.; Lim, S.T.; Sohn, M.-H.; Chi, D.Y.; Kim, D.W.
Polymer-supported pentaethylene glycol as a facile heterogeneous catalyst for nucleophilic
fluorination. Org. Lett. 2010, 12, 3740–3743.
1
8. Wold, S.; Sjöström, M.; Carlson, R.; Lundstedt, T.; Hellberg, S.; Skagerberg, B.; Wikström, C.;
Öhman, J. Multivariate design. Anal. Chim. Acta 1986, 191, 17–32.
1
9. Schwesinger, R.; Schlemper, H.; Hasenfratz, C.; Willaredt, J.; Dambacher, T.; Breuer, T.; Ottaway, C.;
Fletschinger, M.; Boele, J.; Fritz, H.; et al. Extremely strong, Uncharged auxiliary bases; Monomeric
and polymer-supported polyaminophosphazenes (P2–P5). Liebigs Ann. 1996, 1996, 1055–1081.
0. Niedermoser, S.; Pape, M.; Gildehaus, F.J.; Wängler, C.; Hartenbach, M.; Schirrmacher, R.;
Bartenstein, P.; Wängler, B. Evaluation of an automated double-synthesis module: Efficiency and
reliability of subsequent radiosyntheses of FHBG and FLT. Nucl. Med. Biol. 2012, 39, 586–592.
2
Sample Availability: Not available.
©
2013 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
http://creativecommons.org/licenses/by/3.0/).
(