[18F]fluoride Activation and 18F-Labelling in Hydrous Conditions—Towards a Microfluidic Synthesis of PET Radiopharmaceuticals

18F-labelled radiopharmaceuticals are indispensable in positron emission tomography. The critical step in the preparation of 18F-labelled tracers is the anhydrous F-18 nucleophilic substitution reaction, which involves [18F]F− anions generated in aqueous media by the cyclotron. For this, azeotropic drying by distillation is widely used in standard synthesisers, but microfluidic systems are often not compatible with such a process. To avoid this step, several methods compatible with aqueous media have been developed. We summarised the existing approaches and two of them have been studied in detail. [18F]fluoride elution efficiencies have been investigated under different conditions showing high 18F-recovery. Finally, a large scope of precursors has been assessed for radiochemical conversion, and these hydrous labelling techniques have shown their potential for tracer production using a microfluidic approach, more particularly compatible with iMiDEV™ cassette volumes.


Introduction
Positron emission tomography (PET) imaging represents a powerful technique for molecular diagnostic and therapeutic procedures.The development of suitable PET imaging agents that can be readily labelled with positron-emitting radionuclides is essential for the detection, characterisation and staging of diseases.Molecular imaging using 18 F-radiolabelled smallmolecule probes as imaging agents is successfully used in cardiology, neurodegenerative disease, inflammation diseases, bacterial infection detection and oncology [1].
The nucleophilic substitution reaction is the most frequently used strategy for 18 Fradiolabelling (Figure 1).
Table 1.Overview of various 18 F-radiolabelling methods via nucleophilic substitution in hydrous conditions.When putting the emphasis on the simplicity of the whole 18 F-radiofluorination process, hydrous cryptate-mediated 18 F-fluorination and the use of tetrabutylammonium salts stand out from the drying-free radiofluorination methods.In addition, kryptofix 2.2.2 (K 222 ) and tetrabutylammonium salts are widely used in the manufacturing of radiopharmaceuticals, and thus, they are accompanied by a large panel of analytical standardised methods described in the European or US Pharmacopoeia.This is advantageous to establish quality control if a final injection to humans is envisaged.The use of these hydrous elution methods is also supported by earlier results in microfluidic PET tracer production, where small amounts of water (<10%) have been well tolerated [33,34].

Entry
In this article, we have exploited these drying-free approaches (Table 1, entries 6 and 7) to evaluate the elution efficiency of fluorine-18 trapped on several types of anion exchangers embedded on the innovative iMiDEV TM microfluidic platform with various amounts of water.We have then assessed the influence of water content on the radiofluorination reaction on seven different precursors of radiotracers and prosthetic reagents.This study represents the first step toward the implementation of full radiosynthesis on the iMiDEV TM microfluidic platform.

Results
This study aims to develop a method for [ 18 F]fluoride activation at the microfluidic scale suitable for labelling various radiotracers using the iMiDEV TM microfluidic platform.Due to the limited volume that can be used on microfluidic cassettes, an optimisation of the trapping of fluorine-18 on anion exchange beads (AEX), hydrous release and radiofluorination has been performed as described in Figure 2.

Results
This study aims to develop a method for [ 18 F]fluoride activation at the microfluidic scale suitable for labelling various radiotracers using the iMiDEV TM microfluidic platform.Due to the limited volume that can be used on microfluidic cassettes, an optimisation of the trapping of fluorine-18 on anion exchange beads (AEX), hydrous release and radiofluorination has been performed as described in Figure 2.

[ 18 F]fluoride Recovery Studies
Efficient [ 18 F]fluoride recovery from strong anion exchange cartridges and its reactivity in the subsequent nucleophilic substitution reaction with versatile precursors is of the utmost importance for developing a robust radiolabelling procedure.Based on a compilation of literature data, we were prompted to evaluate two promising methods on a microvolume scale.The first method reported by Kniess et al. [22] consists of using a kryptofix/K2CO3 complex (30-60 µmol/15-30 µmol) in a solvent system containing water (2-3%) in acetonitrile (v/v, 1 mL).The second method developed by Kim et al. [12] uses a low water content (5% v/v) of the concentrated tetrabutylammonium bicarbonate (40%) in different organic solvents (CH3CN, DMF, DMSO, 1 mL).Based on these results, our aim was to optimise the elution protocols for volumes and conditions that could be easily translated onto the microfluidic platform iMiDEV™ where volumes are limited.
The first part of this study aimed to select the best AEX resins among different commercial sources and the most efficient hydrous elution method.The schematic workflow is shown in Figure 2. Three types of commercially available AEX beads from Waters TM , Eichrom and S*Pure were used.First, the data obtained from the suppliers on the bead mass in each commercial cartridge was compared to the measured bead mass after opening the cartridge (Table S3 of Supplementary Materials).A discrepancy between the packaging information and the measured data was observed.To standardise the trapping-elution experiments and minimise the bead mass, 25 mg (corresponding to 50µL bed volume) of each type of resin was filled in an empty SPE cartridge. Aqueous

[ 18 F]fluoride Recovery Studies
Efficient [ 18 F]fluoride recovery from strong anion exchange cartridges and its reactivity in the subsequent nucleophilic substitution reaction with versatile precursors is of the utmost importance for developing a robust radiolabelling procedure.Based on a compilation of literature data, we were prompted to evaluate two promising methods on a microvolume scale.The first method reported by Kniess et al. [22] consists of using a kryptofix/K 2 CO 3 complex (30-60 µmol/15-30 µmol) in a solvent system containing water (2-3%) in acetonitrile (v/v, 1 mL).The second method developed by Kim et al. [12] uses a low water content (5% v/v) of the concentrated tetrabutylammonium bicarbonate (40%) in different organic solvents (CH 3 CN, DMF, DMSO, 1 mL).Based on these results, our aim was to optimise the elution protocols for volumes and conditions that could be easily translated onto the microfluidic platform iMiDEV™ where volumes are limited.
The first part of this study aimed to select the best AEX resins among different commercial sources and the most efficient hydrous elution method.The schematic workflow is shown in Figure 2. Three types of commercially available AEX beads from Waters TM , Eichrom and S*Pure were used.First, the data obtained from the suppliers on the bead mass in each commercial cartridge was compared to the measured bead mass after opening the cartridge (Table S3 of Supplementary Materials).A discrepancy between the packaging information and the measured data was observed.To standardise the trapping-elution experiments and minimise the bead mass, 25 mg (corresponding to 50 µL bed volume) of each type of resin was filled in an empty SPE cartridge.
Aqueous [ 18 F]fluoride (50-200 MBq) was first trapped on the AEX resin, and the 18 O-enriched water was discarded into the waste vial.The cartridge was then dried with an airflow.Excellent trapping efficiencies were observed for all three types of beads despite the reduced beads quantity.A trapping efficiency of 92 ± 3% (n = 3) was measured for beads from Waters TM , 95 ± 4% (n = 3) from Eichrom and 97 ± 2% (n = 3) for those from S*Pure.Residual water was eliminated by passing anhydrous acetonitrile through the cartridges, followed by air drying.The 18 F-activity loss for the washing step was ≤5% for all experiments.[ 18 F]fluoride was gradually eluted in the opposite direction to the trapping using 0.1 mL fractions of the eluent (total elution volume 0.5 mL).Eluents used in the manual tests were 1) kryptofix/K 2 CO 3 eluent (60 µmol/30 µmol) containing 3% (v/v) water in CH 3 CN and 2) 5% water (v/v) TBAB 40% solution in acetonitrile or DMSO.
The activity of the cartridge and the eluate were measured after each elution fraction.Table S4 of Supplementary Materials and Figure 3 show the results of the QMA-CO 3 elution profile using the kryptofix-based method on different types of QMA-CO 3 beads.With the K 222 -based method, QMA-CO 3 beads supplied by Eichrom have shown a faster start of the elution (from the first 0.1 mL).However, overall elution efficiency (EE) was similar to QMA-CO 3 beads supplied by Waters TM .The fractions between 0.1 mL and 0.4 mL (6 bed volumes) contained the highest radioactivity concentrations (around 83% of the total activity).Accordingly, based on the results obtained using AEX resin supplied by S*Pure, the K 222based method allowed us to obtain 68 ± 8% elution efficiency with the maximum activity concentrated between 0.2 and 0.5 mL (around 63% of the total activity in 6 bed volumes).
Molecules 2024, 29, x FOR PEER REVIEW 5 of 16 cartridges, followed by air drying.The 18 F-activity loss for the washing step was ≤5% for all experiments.[ 18 F]fluoride was gradually eluted in the opposite direction to the trapping using 0.1 mL fractions of the eluent (total elution volume 0.5 mL).Eluents used in the manual tests were 1) kryptofix/K2CO3 eluent (60 µmol/30 µmol) containing 3% (v/v) water in CH3CN and 2) 5% water (v/v) TBAB40% solution in acetonitrile or DMSO.
The activity of the cartridge and the eluate were measured after each elution fraction.Table S4 of Supplementary Materials and Figure 3 show the results of the QMA-CO3 elution profile using the kryptofix-based method on different types of QMA-CO3 beads.With the K222-based method, QMA-CO3 beads supplied by Eichrom have shown a faster start of the elution (from the first 0.1 mL).However, overall elution efficiency (EE) was similar to QMA-CO3 beads supplied by Waters TM .The fractions between 0.1 mL and 0.4 mL (6 bed volumes) contained the highest radioactivity concentrations (around 83% of the total activity).Accordingly, based on the results obtained using AEX resin supplied by S*Pure, the K222based method allowed us to obtain 68 ± 8% elution efficiency with the maximum activity concentrated between 0.2 and 0.5 mL (around 63% of the total activity in 6 bed volumes).4 represent the elution efficiencies obtained with 5% H2O TBAB40% in acetonitrile.An average recovery of 73.8 ± 3.4% and 70.8 ± 7.4% have been measured for Waters TM and Eichrom AEX beads, respectively.Only half of the trapped activity was eluted under the same conditions when S*Pure beads were used.As for the kryptofix approach, the first 0.1 mL of eluate (2 bed volumes) did not contain any 18 F-anions.
Significantly lower elution efficiency was observed with 5% water TBAB40% when ac- Table S5 of Supplementary Materials and Figure 4 represent the elution efficiencies obtained with 5% H 2 O TBAB 40% in acetonitrile.An average recovery of 73.8 ± 3.4% and 70.8 ± 7.4% have been measured for Waters TM and Eichrom AEX beads, respectively.Only half of the trapped activity was eluted under the same conditions when S*Pure beads were used.As for the kryptofix approach, the first 0.1 mL of eluate (2 bed volumes) did not contain any 18 F-anions.This first study led us to select QMA-CO3 beads from the Waters™ and the kryptofixbased hydrous elution method to be transferred to the iMiDEV TM microfluidic platform.In the second part of this study, different elution conditions were evaluated using a microfluidic cassette in which reactor 1 (R1, dedicated to the trapping and concentration of [ 18 F]F − ) was filled with QMA-CO3 beads from Waters TM .The trapping efficiency of [ 18 F]fluoride in QMA-CO3 beads using the microfluidic cassette was 99.0 ± 0.9% (n = 20).Activity loss during the resin washing step with CH3CN was 0.8 ± 1.0% (n > 50, decay corrected).
Preliminary tests with direct trapping and the elution approach gave poor elution efficiencies when using small elution volumes; thus, we chose to continue only with the reversed trapping-elution approach (Figure 2).Elution results using the kryptofix-based method are presented in Figure 5 and Table S6 of Supplementary Materials.To evaluate the importance of water for the elution efficiency, 1% to 5% water-containing elution solutions were studied.Throughout the data, the elution efficiency showed an increasing trend when the water content was increased (Figure 5).With 2% or more water in the eluent, EE was over 97% when ≥20 bed volumes (≥1 mL) of eluent were used.Decreasing the water content to 1% dropped the EE to 90.8 ± 1.2% with ≥20 bed volumes of eluent.When decreasing the eluent volume to 200 µL (4 bed volumes), the EE variance increased, but the EE stayed satisfactory when the water content was at least 2%.With 200 µL of 1% water containing eluent, the EE variance was the highest and EE was only up to 80%.Decreasing the eluent volume to 150 µL (3 bed volumes) gave similar results as 200 µL.Further reduction of the eluent volume to 100 µL (2 bed volumes) had a drastic decreasing effect on EE, and elution was not reproducible with any amount of water as can be seen in Figure 5. Significantly lower elution efficiency was observed with 5% water TBAB 40% when acetonitrile was replaced by DMSO as a solvent, allowing only 11.5% (n = 1) 18 F-recovery from AEX cartridges.Therefore, all following tests were performed with phase-transfer catalysts (PTC) in acetonitrile solution.
This first study led us to select QMA-CO 3 beads from the Waters™ and the kryptofixbased hydrous elution method to be transferred to the iMiDEV TM microfluidic platform.In the second part of this study, different elution conditions were evaluated using a microfluidic cassette in which reactor 1 (R1, dedicated to the trapping and concentration of [ 18 F]F − ) was filled with QMA-CO 3 beads from Waters TM .The trapping efficiency of [ 18 F]fluoride in QMA-CO 3 beads using the microfluidic cassette was 99.0 ± 0.9% (n = 20).Activity loss during the resin washing step with CH 3 CN was 0.8 ± 1.0% (n > 50, decay corrected).
Preliminary tests with direct trapping and the elution approach gave poor elution efficiencies when using small elution volumes; thus, we chose to continue only with the reversed trapping-elution approach (Figure 2).Elution results using the kryptofix-based method are presented in Figure 5 and Table S6 of Supplementary Materials.To evaluate the importance of water for the elution efficiency, 1% to 5% water-containing elution solutions were studied.Throughout the data, the elution efficiency showed an increasing trend when the water content was increased (Figure 5).With 2% or more water in the eluent, EE was over 97% when ≥20 bed volumes (≥1 mL) of eluent were used.Decreasing the water content to 1% dropped the EE to 90.8 ± 1.2% with ≥20 bed volumes of eluent.When decreasing the eluent volume to 200 µL (4 bed volumes), the EE variance increased, but the EE stayed satisfactory when the water content was at least 2%.With 200 µL of 1% water containing eluent, the EE variance was the highest and EE was only up to 80%.Decreasing the eluent volume to 150 µL (3 bed volumes) gave similar results as 200 µL.Further reduction of the eluent volume to 100 µL (2 bed volumes) had a drastic decreasing effect on EE, and elution was not reproducible with any amount of water as can be seen in Figure 5.

[ 18 F]fluoride Labelling Studies
To investigate the applicability of these two different [ 18 F]fluoride activation approaches for the subsequent radiolabelling, the obtained eluates have been tested in manual radiolabelling of different precursors at a microvolume scale.Reaction volumes were adjusted to be compatible with the iMiDEV™ microfluidic platform (volume of the reaction chamber R2 < 290µL).Figure 6 (Table S1 of Supplementary Materials) represents the results of the hydrous nucleophilic 18 F-fluorination of various radiotracer and prosthetic reagent precursors.Aliphatic and heteroaromatic precursors have been selected for a large validation scope.
[ 18 F]FDG, representing the gold standard of PET imaging, was the first selected radiopharmaceutical to be tested.The synthesis of [ 18 F]FDG has been widely studied with various synthesisers, from conventional to chip [35].The intermediate molecule [ 18 F]FTAG has been synthesised from only 5 mg of mannose triflate precursor under hydrous conditions with an excellent RCC of 82.2 ± 5.8% and 89.9 ± 2.4% using kryptofix/K 2 CO 3 (with 3% of water) and TBAB (with 5% of water) activation methods, respectively (Table S1, entries 1, 2 of Supplementary Materials).

[ 18 F]fluoride Labelling Studies
To investigate the applicability of these two different [ 18 F]fluoride activation approaches for the subsequent radiolabelling, the obtained eluates have been tested in manual radiolabelling of different precursors at a microvolume scale.Reaction volumes were adjusted to be compatible with the iMiDEV™ microfluidic platform (volume of the reaction chamber R2 < 290µL).Figure 6 (Table S1 of Supplementary Materials) represents the results of the hydrous nucleophilic 18 F-fluorination of various radiotracer and prosthetic reagent precursors.Aliphatic and heteroaromatic precursors have been selected for a large validation scope.Figure 6.Chemical structures of different radiopharmaceuticals and radiopharmaceutical intermediates were obtained in this study under drying-free nucleophilic 18 F-fluorination conditions.RCC rates as measured from radio-TLC (n = 3).
[ 18 F]FDG, representing the gold standard of PET imaging, was the first selected radiopharmaceutical to be tested.The synthesis of [ 18 F]FDG has been widely studied with various synthesisers, from conventional to chip [35].The intermediate molecule [ 18 F]FTAG has been synthesised from only 5 mg of mannose triflate precursor under hydrous conditions with an excellent RCC of 82.2 ± 5.8% and 89.9 ± 2.4% using kryptofix/K2CO3 (with 3% of water) and TBAB (with 5% of water) activation methods, respectively (Table S1, entries 1, 2 of Supplementary Materials).

Discussion
In this study, two hydrous F-18 activation methods based on (1) cryptate-mediated (kryptofix) 18 F-fluorination or (2) using concentrated TBAB solution have been evaluated based on the literature data (Table 1) along with three different anion exchangers.The manual 18 F-elution tests have shown that the three types of beads gave trapping efficiencies of fluorine-18 higher than 92%.The elution efficiency was around 20% superior for Waters TM and Eichrom beads compared to S*Pure beads.A comparison of elution approaches showed that the K 222 -based method gave better results than the TBAB-based method.The elution efficiency is 20% higher for K 222 independently from the bead type.An optimal volume of elution of 0.5 mL has been determined, while the first 0.1 mL may be discarded due to the very low amount of radioactivity it contains.The best beads and elution method combination was the QMA-CO 3 beads from Waters TM eluted with 0.5 mL of a K 222 solution containing 3% water in acetonitrile.The kryptofix-based elution method was transferred to the iMiDEV TM microfluidic platform, but the elution volume had to be reduced because the reactor content (R2) on the iMiDEV TM cassette cannot exceed 286 µL.Nevertheless, fractioned elution showed that 70% and even 90% of the radioactivity could be eluted within 0.2 and 0.3 mL of PTC solution, respectively.These observations are encouraging for the transfer on the iMiDEV TM microfluidic platform.
On the iMiDEV™ cassettes, reactor 1, dedicated to the trapping and concentration of [ 18 F]F − (R1, volume 50 µL), was filled with QMA-CO 3 beads from Waters TM due to their optimal trapping and elution efficiencies.Only the hydrous K 222 -based method has been evaluated using a large panel of elution volumes from 0.1 to 1 mL and water content from 1 to 5%.Trapping efficiency was almost quantitative, reaching 99% without any loss during the washing and drying steps of the beads.Elution efficiency increased with increasing elution volume and water content.Only half of the radioactivity could be eluted with 0.1 mL of elution solution, whatever the water content, along with a high variance in elution efficiency.This result is in accordance with the manual elution study in which only very low amounts of radioactivity could be eluted using the first 0.1 mL.Moreover, Mallapura et al. discussed in a previous study that the cassette and vials clamping step on the docking plate of the iMiDEV™ platform may conceal small volumes in vials, leading to variable dead volumes affecting the subsequent elution step [41].With higher volumes (0.15, 0.2 and 1 mL), elution efficiency increased and reached 95 to 100%, with the highest water content (5%).Nevertheless, a volume of 1 mL is not relevant regarding the volume of R2 (286 µL) on the iMiDEV™ cassette.This set of elution efficiency measurements showed that a minimum of 3 bed volumes of chamber R1 (i.e., 0.15 mL) of elution solution containing at least 2% water leads to satisfactory elution efficiencies.This is a suitable volume considering the needed precursor volume (0.15 mL) and reactor R2 volume on the iMiDEV™ cassette.To compare with earlier microelution studies with AEX beads, we have reached a similar elution efficiency using 3-4 bed volumes of 2-5% water-containing kryptofix eluent, what has earlier been reached using 4 bed volumes of fully aqueous basic eluents [34,42].
We focussed then on the evaluation of the radiochemical conversion (measured by radio-TLC) of various precursors representing most of the possible options for nucleophilic radiofluorination.Aliphatic radiofluorination was performed on five precursors displaying different leaving groups, such as a tosylate (precursors of [ 18 F]DPA-714, [ 18 F]fallypride and [ 18 F]F-Me-OTs), a chlorine atom (precursor for [ 18 F]LBT-999) or a triflate moiety (precursor for [ 18 F]FTAG).Satisfactory to good RCCs were observed for all precursors despite the PTC solution used.Damont et al. [43] described RCC ranging from 50 to 70% for [ 18 F]DPA-714, and Gao et al. [44].reported the best d.c.RCC, reaching 50% for [ 18 F]fallypride, which is fully in accordance with our observations.For [ 18 F]LBT-999, our RCC was comparable with the n.d.c.radiochemical yields reported by Vala et al. [45].For the aliphatic radiofluorination, the RCCs tend to slightly decrease when 5% water compared to 2% water in K 222 solution was used, which is not the case when using 5% water in TBAB 40% solutions, leading to the highest [ 18 F]DPA-714 and [ 18 F]FTAG RCCs.For the prepara-tion of [ 18 F]F-Me-OTs, several reports indicated that the presence of water increased the production yield of [ 18 F]FCH 2 OTs while limiting the formation of the [ 18 F]tosylfluoride by-product [36][37][38].In batch chemistry, Neal et al. optimised the amount of water to 5% and obtained [ 18 F]FCH 2 OTs in 83% d.c.yield.They also observed that conversion was higher using kryptofix rather than TBAB, a difference in PTC performance that we have not observed [38].Rodnick et al. introduced 13% water in their cyrptate-based solution and obtained the desired intermediate in 38% RCC [36].The preparation of [ 18 F]FCH 2 OTs was also implemented on the Avdion TM microfluidic device.Pascali et al. showed that a significant amount of water (optimised to 20%) was necessary to promote the formation of the desired radioactive intermediate (n.d.c.RCY of 44%) and to avoid the production of [ 18 F]FOTs [36].We observed in our study that adding water (5 to 10%) to the reaction mixture does not hamper the radiofluorination, and our RCCs are also higher than the ones previously reported.
Heteroaromatic radiofluorination was performed on four various precursors of prosthetic reagents.[ 18 F]FPyOBn, starting from a quaternary ammonium precursor, was obtained with the highest RCCs, probably due to the strong activation of the pyridine ring provided by the benzyl ester.The content of water did not seem to have any influence on the conversion rate.The activated ester, [ 18 F]FPyNHS, was obtained with modest and even low RCC when the content of water in the eluent was increased from 2 to 5%, respectively.Partial hydrolysis of the activated ester due to the presence of water in basic conditions may be the reason for such low RCCs.Richard et al. reported comparable conversion yields for [ 18 F]FPyOBn but drastically higher conversions for [ 18 F]FPyNHS.Note that reactions were conducted in standard anhydrous conditions at lower temperatures [39].Finally, the preparation of azide-containing reagent, [ 18 F]FPyZIDE, was evaluated starting either from a nitro or a quaternary ammonium precursor.Nitro-precursor gave only low RCCs at 130 • C. In a previous report, Kuhnast et al. [46] showed that the radiofluorination of such a 2-nitro-3-alkoxypyridine analogue gave very high yields (>90%) at 165 • C in a 3 min reaction.In this study, we chose 130 • C as the maximum temperature to be compatible with the future transfer to the iMiDEV TM platform.The quaternary ammonium precursor gave good RCCs when using 2% water in the PTC solution.Again, in this latter case, increasing the amount of water drastically decreased the RCC.Overall, 2% water in K 222 solution was well tolerated by most of the tested precursors, and the increase of water content to 5% adversely affected the radiochemical conversion rates.When TBAB solution was used as PTC, the presence of water seemed to be better tolerated and gave even better RCCs compared to 5% water in K 222 solutions.F]FPyOBn reference) were synthetised at SHFJ as previously described [39,40].
Sterile water for injections (WFI) Macoflex N from Macopharmas; 0.9% Sodium chloride (NaCl) solution for injections was purchased from Baxter ® .Pure H 2 O (18.2 MΩ) was produced with a purification system (ARIUM ® MINI, Göttingen, Germany).Each solution was freshly prepared and ultrasonicated until the complete homogenisation of the eluent prior to use.Water content was checked using the Karl Fisher method.

Preparation of 5% H 2 O TBAB 40% Solution
In a vial, 50 µL of concentrated TBAB (40%) was added to 1 mL of acetonitrile.The resulting freshly prepared solution was vortexed to ensure thorough mixing.

[ 18 F]fluoride Production
Aqueous [ 18 F]fluoride was produced via the 18 O(p,n) 18 F nuclear reaction from 18 Oenriched H 2 O on a PET Trace cyclotron from General Electric (Windsor, CT, USA) at the CURIUM TM facility (Nancy, France) and a Cyclone-18/9 Cyclotron (IBA) at the SHFJ (Orsay, France).

General Manual [ 18 F]fluoride Elution Method
Quaternary methyl ammonium carbonate (QMA-CO 3 ) anion-exchange (AEX) SPE cartridges were purchased from Waters TM (Sep-Pak ® Accell Plus QMA Carbonate, 37-55 µm particle size, Guyancourt, France), Eichrom (QMA-S-BC, 55 µm particle size, Lisle, IL, USA) and S*Pure (Maxi-Clean QMA Carbonate, 50 µm particle size, Singapore), and the packing material was reduced to 25 mg to mimic the mass available in the R1 reactor of the microfluidic cassette iMiDEV TM .Aqueous [ 18 F]fluoride (0.5-0.75 mL) was passed through an anion-exchange cartridge (without preconditioning) followed by 5 mL air, and the activity of the cartridge was determined with a dose calibrator (CRC ® -25R, Capintec, Inc., Florham Park, NJ, USA).The cartridge was rinsed with 4 mL anhydrous acetonitrile to eliminate the residual water and dried with 15 mL air.[ 18 F]fluoride was fractionally eluted from the reverse direction using two different elution solutions (3% K 222 /K 2 CO 3 /CH 3 CN/H 2 O or 5% water TBAB 40% ) followed by 5 mL of air per fraction.Elution fractions were 0.1 mL, and the total elution volume was 0.5 mL.The activity of the eluate fraction and AEX cartridge (residual [ 18 F]fluoride) was determined with a Capintec dose calibrator.Activity data was used to calculate % fluoride recovery.

General [ 18 F]fluoride Elution Method Using iMiDEV TM
The microfluidic-based platform iMiDEV™ (PMB-Alcen, Peynier, France) uses disposable microfluidic cassettes (PMB-Alcen, France) for radiotracer production.Characteristics of the device and the microfluidic cassettes have been described earlier [47].In brief, the cassettes have four independent chambers, where chamber R1 is dedicated to radionuclide concentration and chamber R2 is dedicated to the labelling reaction.R1 has a volume of 50 µL (dimensions W 5.5 mm, L 10 mm, H 1 mm), and R2 has a volume of 286 µL (dimensions W 12 mm, L 36 mm, H 1 mm).In this work, to study elution in microfluidic conditions, we used cassettes where chamber R1 was filled with approximately 25 mg QMA-CO 3 anion-exchange beads (Waters™).R1 is filled during the manufacturing of the cassettes, as previously described [48].Briefly, filling is done through a hole on top of chamber R1; the initial filling is based on gravity, and to make the bead distribution around the chamber even, compressed air is pushed through, and ultrasonic vibrations are used.
Aqueous [ 18 F]fluoride (2 mL) was passed through the nonpreconditioned QMA beads, followed by a stream of He for drying for 2 min at 1.2 bar.Beads were washed with CH 3 CN (4 mL in vial C/D), followed by a stream of He for drying for 3 min at 2 bar.[ 18 F]fluoride was then eluted with K 222 /K 2 CO 3 /CH 3 CN eluent solution (0.1 to 1 mL in vial A/B), followed by a stream of He for 0.5 min at 0.2 bar.Elution efficiency (EE) and activity loss during the washing step were calculated using the radioactivity sensor data.A radioactivity sensor is placed at the outlet of the R1 reactor.Cassette architecture and the elution pathways are presented in the supporting information (Figures S1 and S2 of Supplementary Materials).S1 of Supplementary Materials.The corresponding mixture was heated according to commonly used labelling conditions.At the end of the labelling, the mixture was cooled down, and an aliquot of the crude reaction solution was used for the quality control tests.The radiochemical conversion is determined by radio-TLC analysis of a small aliquot from a reaction solution and identity using radio-HPLC analysis (Table S2 of Supplementary Materials).Each labelling was done in triplicate.

Quality Control Analysis
Radio-TLCs were performed either with a Mini GITA TLC scanner controlled by GinaX 10.4.5 software (Elysia S.A., Angleur, Belgium) or a Mini-Scan and Flow-Count radioactive detection system (Bioscan) and Chromeleon software 7.2.10.ES (Thermo Fischer, Courtaboeuf, France).An aliquot of the reaction mixture (5 µL) was analysed by radio-TLC on a silica-gel-coated aluminium plate.Elution conditions and Rf are detailed in the table below (Table 3).

Conclusions
In this study, we have optimised several conditions to allow the implementation of hydrous radiofluorination approaches at the microfluidic scale on the iMiDEV™ platform.We first tested manually different anion exchangers to trap fluorine-18 and two PTC solutions to release it prior to radiofluorination.We observed that QMA-CO 3 beads from Waters TM gave the best results combined with a cryptate-based PTC solution for the recovery of fluorine-18.Fractioned elution showed that 90% of radioactivity could be released within 0.3 mL, which is compatible with the iMiDEV TM microfluidic platform.These results were confirmed on the microfluidic platform with even better elution efficiencies using only 0.15 to 0.2 mL of PTC solution.We then evaluated the hydrous radiofluorination of seven different aliphatic and heteroaromatic precursors with different leaving groups.Two standard PTC solutions containing various amounts of water were tested.Good to high radiochemical conversions were observed for the aliphatic radiofluorination whatever the PTC solution used.For the aromatic radiofluorination, results are more contrasted, probably due to the presence or absence of an electron-withdrawing group on the pyridine ring and to the temperature that we have limited to 130 • C. Altogether, the results show that most of the tested precursors tolerate the presence of water, but an increase in water led to a decrease in the radiofluorination radiochemical conversions.This study opens access to a variety of radiopharmaceuticals using hydrous radiofluorination approaches, which could be integrated into new emerging microfluidic platforms.

Figure 2 .
Figure 2. Schematic workflow of the manual [ 18 F]fluoride elution in reverse mode and subsequent hydrous labelling.

Figure 2 .
Figure 2. Schematic workflow of the manual [ 18 F]fluoride elution in reverse mode and subsequent hydrous labelling.

Figure 3 .
Figure 3. [ 18 F]F-elution profile using different types of anion exchange cartridges and the kryptofixbased method.

Figure 3 .
Figure 3. [ 18 F]F-elution profile using different types of anion exchange cartridges and the kryptofixbased method.

Figure 4 .
Figure 4. [ 18 F]fluoride elution profile using different types of anion exchange cartridges and tetrabutylammonium-based method.

Figure 4 .
Figure 4. [ 18 F]fluoride elution profile using different types of anion exchange cartridges and tetrabutylammonium-based method.

Molecules 2024 , 16 Figure 5 .
Figure 5. [ 18 F]fluoride elution profile using the kryptofix-based elution method in a microfluidic cassette varying the water content and eluent volume.

Figure 5 .
Figure 5. [ 18 F]fluoride elution profile using the kryptofix-based elution method in a microfluidic cassette varying the water content and eluent volume.

Figure 6 .
Figure 6.Chemical structures of different radiopharmaceuticals and radiopharmaceutical intermediates were obtained in this study under drying-free nucleophilic 18 F-fluorination conditions.RCC rates as measured from radio-TLC (n = 3).

Table 1 .
Overview of various18F-radiolabelling methods via nucleophilic substitution in hydrous conditions.

Table S5 of
Supplementary Materials and Figure 18 4.2.Eluent Preparation 4.2.1.Preparation of K 222 /K 2 CO 3 /CH 3 CN/H 2 O Solutions Preparation of different stock solutions of K 222 /K 2 CO 3 /acetonitrile/water was performed using amounts and volumes described in Table 2.