A Gas Phase Route to [18F]fluoroform with Limited Molar Activity Dilution

Positron emission tomography (PET) is an important imaging modality for biomedical research and drug development. PET requires biochemically selective radiotracers to realize full potential. Fluorine-18 (t1/2 = 109.8 min) is a major radionuclide for labeling such radiotracers but is only readily available in high activities from cyclotrons as [18F]fluoride ion. [18F]fluoroform has emerged for labeling tracers in trifluoromethyl groups. Prior methods of [18F]fluoroform synthesis used difluoro precursors in solution and led to high dilution with carrier and low molar activity (Am). We explored a new approach for the synthesis of [18F]fluoroform based on the radiosynthesis of [18F]fluoromethane from [18F]fluoride ion and then cobaltIII fluoride mediated gas phase fluorination. We estimate that carrier dilution in this process is limited to about 3-fold and find that moderate to high Am values can be achieved. We show that [18F]fluoroform so produced is highly versatile for rapidly and efficiently labeling various chemotypes that carry trifluoromethyl groups, thereby expanding prospects for developing new PET radiotracers.

Positron emission tomography (PET) is an increasingly important molecular imaging modality for drug development 1,2 , biomedical research 3 , and medical diagnosis [4][5][6] . The value of PET for imaging molecular targets in living animal 7 and human 8 subjects derives from the development of biochemically specific radiotracers (i.e., radiotracers that are each capable of imaging a single targeted protein, such as a low density neuroreceptor). One of the most useful and widely used radionuclides for labeling such radiotracers is the short-lived positron-emitter, fluorine-18 (β + = 97%, t 1/2 = 109.8 min) 9,10 . Nowadays, fluorine-18 can be produced in very high activities (~500 GBq) as aqueous [ 18 F]fluoride ion with moderate to high molar activity (A m ; where A m is defined 11 as the ratio of the radioactivity of a compound to its mass at a specified time), typically in the 40-400 GBq/μmol range. Therefore, there has been a surge in the development of methods for the late-stage labeling of PET radiotracers with [ 18 F]fluoride ion. However, these methods have been confined mostly to labeling monofluorocarbon (C−F) groups 12,13 .
Substitution of a methyl, chloro, or another substituent in a drug-like molecule with a trifluoromethyl (CF 3 ) group can lead to better pharmaceutical properties and improved metabolic stability [14][15][16][17] . Consequently, a CF 3 group regularly appears in many new drugs and drug candidates [18][19][20][21][22] . Prominent examples include fluoxetine (1; Prozac), celecoxib (2; Celebrex), and leflunomide (3; Arava) ( Fig. 1). Because of the role of PET in drug development and a frequent requirement to label drugs and new radiotracers with a positron-emitter, academic groups have pursued the development of methods for labeling CF 3 groups with fluorine-18 23 (Fig. 2). Generally, however, the molar activities that are needed for radiotracers to be used for PET imaging of low-density protein targets are at the high end of the achievable range or ideally even higher. Here we explored the radiosynthesis of [ 18 F]fluoroform according to a different strategy involving initial installation of the fluorine-18 followed by subsequent gas phase difluorination. We find that carrier dilution with this method is limited to about 3-fold. We

Results
Production of [ 18 F]f luoroform. We found that initial conditioning of a newly installed CoF 3 column by heating it once to 320 °C while sealed under helium resulted in optimal yields of [ 18 F]fluoroform in subsequent use at lower temperatures. Conditioning of the column before a run and subsequent regeneration are described in Supplementary  A single heat-conditioned CoF 3 column could be used for a series of [ 18 F]fluoroform productions (Fig. 4). Yield increased appreciably after the first run and was well maintained over at least 12 subsequent runs. The average yield of [ 18 F]fluoroform from [ 18 F]fluoromethane was 35 ± 11% (n = 77) from six different CoF 3 columns operated at least a dozen times each. HPLC showed that the only radioactive contaminant was occasionally a very low amount of unchanged [ 18 F]fluoromethane ( Supplementary Fig. S7). The six CoF 3 columns produced [ 18 F]fluoroform with 98 ± 3% purity (n = 77). This good re-usability implies that the CoF 3 is not rapidly and completely decomposed to CoF 2 and fluorine at 280 °C. The overall process for producing [ 18 F]fluoroform from [ 18 F]fluoride ion required 60 minutes from the end of a cyclotron irradiation and was thus much less than one half-life of fluorine-18.  From the A m values and measurements of radioactivity entering and leaving the CoF 3 column, we calculated that in the absence of GC purification the average number of moles of carrier fluoroform produced was 2.01 ± 1.56−fold greater than the number of moles of fluoromethane introduced into the CoF 3 column. When GC purification was used, this ratio became closer to unity (0.69 ± 0.41−fold) (Fig. 5B, Supplementary Table S4). The latter finding is consistent with our observation that recovery of radioactivity from the CoF 3 column was 34%, implying that the rest (66%) was retained on the CoF 3 column. The retained activity was not identified but is clearly not [ 18 F]fluoromethane or [ 18 F]fluoroform because we had found earlier that no radioactivity adheres to the CoF 3 column in the conversion of [ 11 C]methane into [ 11 C]fluoroform 33 .

Investigation of carrier dilution in [ 18 F]fluoroform synthesis. Of major interest was the
To explain our observations on carrier dilution and yield, and the radioactivity retained on the CoF 3 column, we postulate that there is exchange of 18 F between [ 18 F]fluoroform and the co-produced two equivalents of HF (Fig. 2), and that all the radioactive HF adheres to the CoF 3 column. No radioactivity was ever detected in the depicted HF trap of the apparatus, which is now regarded as redundant. According to our postulate, the yield of [ 18 F]fluoroform from [ 18 F]fluoromethane at equilibrium is expected to be 33% and the carrier dilution 3-fold, which within likely experimental errors, accords with our observations.   www.nature.com/scientificreports www.nature.com/scientificreports/ Most of our runs to produce [ 18 F]fluoroform were performed at varying periods up to several hours after the end of radionuclide production. To bench-mark comparisons, all estimated A m values were decay-corrected to the end of radionuclide production. The maximal molar activity of the [ 18 F]fluoride ion available to us was 336 GBq/µmol and on average was 150 ± 73 GBq/µmol (n = 10). We found that [ 18

Cu(I)-mediated trifluoromethylations with [ 18 F]fluoroform.
We tested the reactivity of the Cu(I) derivative of the [ 18 F]fluorofrom from the new method of radiosynthesis on several model substrates with various methods (Fig. 6). We first confirmed the known reactivity of [ 18 F]CuCF 3 towards iodoarenes 27 (Fig. 7). The very high www.nature.com/scientificreports www.nature.com/scientificreports/ yield of [ 18 F]9a from this method (82 ± 13%) was very similar to that which we obtained from an iodo precursor, and far exceeded that previously obtained by van der Born et al. from the same reaction (4 ± 2%) 29 . For other examples (11a, 11d, 11f), our yields were very high (>91%) and in accord with those previously reported 29 . The more labile BrCH 2 and AcO substituents were less well tolerated, giving moderate yields under non-optimized conditions. Nonetheless, these examples ([ 18   www.nature.com/scientificreports www.nature.com/scientificreports/ Finally, the treatment of commercially available 'wet' diazonium salts 12a-12e with [ 18 F]CuCF 3 gave [ 18 F]11j, and [ 18 F]13a-[ 18 F]13d, respectively, in good to high yields (Fig. 7). The yield of [ 18 F]11j (74 ± 9%) was comparable to that from the use of boronic acid as precursor (85 ± 13%). The yields of [ 18 F]13a-[ 18 F]13c exceeded 86% and compare well with the yields of these labeled compounds from the use of arylboronic acids or aryl iodides as precursors 28 . This new method therefore appeared highly effective for the simple one-pot conversion of arylamines into [ 18 F]trifluoromethylarenes.

Discussion
[ 18 F]f luoroform was readily produced in useful yield and with limited carrier dilution from cyclotron-produced [ 18  synthesis may be capable of delivering higher molar activities than so far reported by using much higher levels of starting radioactivity and by limiting the amount of difluorocarbene formation. The radiochemical pathway in our new method for producing [ 18 F]fluoroform clearly avoids any possibility for carrier dilution from difluorocarbene formation. The radiosynthesis apparatus is considered amenable to automation and remote control to ensure radiation protection for personnel. With this method, the labeling of PET radiotracers at a trifluoromethyl group with usefully high A m becomes possible. Although the overall yield of [ 18 F]fluoroform appears modest, the speed, broad scope, and generally high efficiency seen in the many examples of labeling reactions augurs well for useful application of this new method. This is especially so given that very high activities of [ 18 F]fluoride ion can be produced on modern cyclotrons (>400 GBq). With this method, we now envisage access to an enhanced range of useful and exciting radiotracers for PET based on adapting the known richly diverse chemistry of fluoroform [37][38][39][40] and its derivatives 41-51 for unprecedented 18 F-labeling at trifluoromethyl groups. These radiotracers may include chemotypes never previously labeled with fluorine-18.

Materials and Methods
Sources of materials are detailed in Supplementary Information. Fig. 3  was flushed out of the vial with nitrogen gas (20 mL/min) and into Porapak Q (80-100 mesh; 1 g) contained in a first U-shaped stainless-steel tube (0.069 in i.d.) cooled with liquid argon (−186 °C). The transfer generally required 5 min. The sealed trap was then removed from the cooling bath and measured for radioactivity at RT (20-26 °C) with a dose calibrator. The [ 18 F]fluoromethane was then released into a stream of helium gas (20 mL/min) from the Porapak Q trap through Sicapent (phosphorus pentoxide) and then through a heated column (280 °C) of CoF 3 (19 g) for a period of 7 to10 min. The generated [ 18 F]fluoroform was passed through a trap cooled in dry-ice/MeCN (−41 °C) and finally into a glass V-vial containing DMF (0.6-0.8 mL) that was cooled also in a dry-ice/MeCN bath. A second U-shaped stainless-steel tube containing Porapak Q (80-100 mesh) was connected to the outlet of the V-shaped glass product vial to retain any breakthrough of radioactive material for measurement.
[ 18 F]CuCF3 synthesis. CuBr (5 µmol, 0.7 mg) was added to 1-mL glass vial and moved to a glove box (dry nitrogen atmosphere). t-BuOK in DMF (0.3 M, 50 µL) was added to the vial, which was then septum-sealed and removed from the glove box. [ 18 F]f luoroform in DMF (50-300 µL) was added to the vial, mixed, and left at RT for 1 min. A solution of Et 3 N·3HF in DMF (1.64% v/v, 5 mL) was then added. The mixture was mixed thoroughly and allowed to stay at RT for another minute before use in labeling reactions.