Fluorine-18 labelled Ruppert – Prakash reagent ( [ 18 F ] Me 3 SiCF 3 ) for the synthesis of 18 F-trifluoromethylated compounds †

Positron emission tomography (PET) is a molecular imaging technique that non-invasively visualizes biochemical processes in vivo. This highly sensitive technique is increasingly used for the diagnosis of diseases, evaluation of the efficacy of a drug treatment, and drug discovery. PET radiotracers incorporate positron-emitting radionuclides, the most frequently used nuclide being fluorine-18. Fluorine-18 has very favourable characteristics for PET imaging, such as a convenient half-life (109.7 minutes) and a low positron energy (Emax = 0.634 MeV), making it desirable for distribution to radiochemistry laboratories and imaging centres thereby enabling multi-centre clinical trials. There is a tremendous interest in the use of fluorine-19 in drug development. This is reflected by recent Food and Drug Administration (FDA) drug approvals: in 2018 almost 1/3 of the new drugs approved contained one or more fluorine atoms. The trifluoromethyl group is a prevalent motif as it can positively influence many characteristics of a given drug, e.g. metabolic stability, lipophilicity, and protein binding affinity. Since trifluoromethyl moieties are so often found in modern pharmaceuticals, much attention has been paid to the development of synthesis strategies for introducing fluorine-18 labelled trifluoromethyl groups into PET tracers. Aromatic [F]trifluoromethylation is well established and multiple strategies have been reported. However, the reported methodologies mainly rely on [F]fluoroform as [F]trifluoromethylation agent. In organic chemistry, not many trifluoromethylation procedures report the use of [F]fluoroform, as alternative trifluoromethylation agents are generally preferred. One of the most applied nucleophilic trifluoromethylation agents is the Ruppert–Prakash reagent (Me3SiCF3). 9 It was first described in 1984 by Ruppert et al. and found its first application in 1989 by Prakash et al. We therefore envisioned that developing a synthesis strategy for F-labelled Ruppert–Prakash reagent would be very useful for the translation of CF3 functionalization reactions that have been developed in fluorine-19 chemistry to radiofluorination reactions and PET tracer synthesis. To synthesize [F]Me3SiCF3 we followed a procedure reported by Prakash et al. that reacted fluoroform with trimethylsilyl chloride in presence of potassium hexamethyldisilazide (KHMDS) as the base in toluene. Initial experiments using [F]fluoroform 1 synthesized according to our previously reported method showed successful formation of the desired product [F]Me3SiCF3 3, in addition to unreacted [F]fluoroform 1 and formation of a side product which was identified as [F]trimethylsilyl fluoride 4 (see Scheme 1). To optimize the reaction towards full conversion to [F]Me3SiCF3 3 several reaction parameters were varied.

Positron emission tomography (PET) is a molecular imaging technique that non-invasively visualizes biochemical processes in vivo. [1][2][3] This highly sensitive technique is increasingly used for the diagnosis of diseases, evaluation of the efficacy of a drug treatment, and drug discovery. 3 PET radiotracers incorporate positron-emitting radionuclides, the most frequently used nuclide being fluorine-18. 4 Fluorine-18 has very favourable characteristics for PET imaging, such as a convenient half-life (109.7 minutes) and a low positron energy (E max = 0.634 MeV), making it desirable for distribution to radiochemistry laboratories and imaging centres thereby enabling multi-centre clinical trials. [5][6][7] There is a tremendous interest in the use of fluorine-19 in drug development. 8,9 This is reflected by recent Food and Drug Administration (FDA) drug approvals: in 2018 almost 1/3 of the new drugs approved contained one or more fluorine atoms. 10 The trifluoromethyl group is a prevalent motif as it can positively influence many characteristics of a given drug, e.g. metabolic stability, lipophilicity, and protein binding affinity. [11][12][13] Since trifluoromethyl moieties are so often found in modern pharmaceuticals, much attention has been paid to the development of synthesis strategies for introducing fluorine-18 labelled trifluoromethyl groups into PET tracers. Aromatic [ 18 F]trifluoromethylation is well established and multiple strategies have been reported. [14][15][16][17][18] However, the reported methodologies mainly rely on [ 18 F]fluoroform as [ 18 F]trifluoromethylation agent. In organic chemistry, not many trifluoromethylation procedures report the use of [ 19 F]fluoroform, as alternative trifluoromethylation agents are generally preferred. One of the most applied nucleophilic trifluoromethylation agents is the Ruppert-Prakash reagent (Me 3 SiCF 3 ). 9 It was first described in 1984 by Ruppert et al. 19 and found its first application in 1989 by Prakash et al. 20 We therefore envisioned that developing a synthesis strategy for 18 F-labelled Ruppert-Prakash reagent would be very useful for the translation of CF 3 functionalization reactions that have been developed in fluorine-19 chemistry to radiofluorination reactions and PET tracer synthesis. 21 To synthesize [ 18 F]Me 3 SiCF 3 we followed a procedure reported by Prakash et al. that reacted fluoroform with trimethylsilyl chloride in presence of potassium hexamethyldisilazide (KHMDS) as the base in toluene. 22  Stirring of the precursor Me 3 SiCl together with the base KHMDS before the trapping was not necessary for [ 18 F]Me 3 SiCF 3 formation, nor was stirring during the reaction. Increase of the precursor concentration as well as the total amounts of precursor, base and solvent helped to push the reaction towards completion and shorten the reaction time with passive warming up to 5 min (see Table 1).
After having established high-yielding reaction conditions for the formation of [ 18 F]Me 3 SiCF 3 , we turned our attention to the isolation of the product for its use in subsequent [ 18 F]trifluoromethylation reactions. As Me 3 SiCF 3 has a low boiling point (54-55 1C) 9 we investigated purification by distillation. Different distillation temperatures and flow rates were tested. An overview of the conditions is given in Table 2. The distillation efficiency increased with higher flow (entry 1-7). At flow rates of 30 mL min À1 and higher, 490% of [ 18 F]Me 3 SiCF 3 could be distilled into THF. Exploring different distillation temperatures showed that temperatures Z70 1C led to the highest yields and at 50 1C the co-distilled precursor precipitated in the tubing and led to blockage during distillation (see entry 6,[8][9][10]. The radiochemical purity of the distilled [ 18 F]Me 3 SiCF 3 was determined and only minor impurities (0-2%) of [ 18 F]Me 3 SiF were found.
At higher flow rates and higher temperatures we noted a significant decrease of the volume in the first reaction vessel and the formation of a precipitate in the second vessel suggesting that precursor Me 3 SiCl (bp. 57 1C) 23 20 In this procedure, the aldehyde or ketone was reacted with Me 3 SiCF 3 under addition of catalytic amounts of TBAF in THF. After hydrolysis with aqueous HCl, the desired trifluoromethylated product was obtained. 4-Nitrobenzaldehyde 6g was selected as a model substrate and the following reaction parameters were varied: reaction time and temperature, quantities of TBAF and precursor as well as the scale of the reaction. [ 18 F]Me 3 SiCF 3 was found to readily react with 4-nitrobenzaldehyde: Reaction of an aliquot of the [ 18 F]Me 3 SiCF 3 in THF (300 mL) with 100 mmol 4-nitrobenzaldehyde 6g and 200 mmol TBAF for 5 minutes at room temperature in a total of 0.5 mL THF resulted in 39 AE 4% RCY. [ 18 F]CHF 3 was formed as a side product, likely resulting from reaction of [ 18 F]Me 3 SiCF 3 with traces of water present in THF or the TBAF solution. The 1M TBAF solution in THF was therefore stored in small batches (10-20 mL) over molecular sieves (3Å) to minimize the water content as much as possible. 23 Variation of reaction time (2.5-20 min) and temperature (0-80 1C) did not have an effect on the radiochemical yield under our conditions. The control of the TBAF concentration however was important since low concentrations (r200 mM) resulted in incomplete release of [ 18  Conditions: Trapping at À80 1C, passive warmup to rt in 5 min. a Isolated yield, average AE SD, decay corrected (dc), n = 3. b Determined by HPLC, average AE SD, dc, n = 3. beneficial for the reaction, and is likely attributed to increasing amounts of water in the TBAF solution which is responsible for promoting [ 18 F]CHF 3 formation. Next, the effect of precursor concentration was investigated and showed that the higher the concentration of precursor, the higher the radiochemical yield (64 AE 5% RCY with 400 mmol precursor; see ESI † for details). To adapt the reaction for relevance to PET tracer syntheses the scale of the reaction was reduced compared to the initial conditions. It was observed that scaling down the reaction to 250 mL and 125 mL total volume resulted in comparable to even slightly higher radiochemical yields than the initial conditions (see Table S6 in ESI †). Due to easier handling and abundance of the model precursors we moved on to explore the substrate scope with the following conditions: 250 mL THF, 200 mmol precursor, 100 mmol TBAF (ESI, † Table S6, entry 7).
Three different groups of substrates were investigated, benzaldehydes, acetophenones and benzophenones, each with no substituents, electron withdrawing (4-MeO-) and electron donating (3-NO 2 -, 4-NO 2 -) substituents. An overview of all products and the corresponding RCYs is shown in Scheme 2. Two general trends were observed: 1. Electron withdrawing groups had a positive influence on the RCY. 2. Benzaldehydes reacted very well whereas benzophenones only resulted in low RCYs (o10%). The first observation can be explained by the electron density. The trifluoromethyl group attacks the positively polarised C atom of the carbonyl group. Electron withdrawing groups reduce the electron density of the carbonyl C atom and facilitate the nucleophilic attack whereas electron donating groups exert the opposite effect. The second observation can be explained by the reaction kinetics. It has been described that the rate constants of trifluoromethylation reactions observe the following rank order: k(benzaldehyde) 4 k(acetophenone) 4 k(benzophenone); and a competing reaction is the formation of fluoroform by reaction of Me 3 SiCF 3 with trace amounts of water. 24 The reaction with benzaldehydes is likely fast enough to outcompete the [ 18 F]fluoroform formation whereas with benzophenones [ 18 F]fluoroform formation predominates.
Attempts to increase the RCY of the benzophenone reactions by further optimization of the previously varied reaction conditions (temperature, time, TBAF amount) were not fruitful. We therefore turned our attention to alternative initiators. Based on work of Johnston et al. 24 we chose two initiators, KOPh and TBAT. Reactions initiated by K + -containing initiators were expected to have a faster turnover rate than with NH 4 + and could improve the RCY of the slow-reacting benzophenones. Anhydrous TBAT was reported to result in more reproducible yields compared to TBAF, due to lower water content. Under our radiochemistry conditions KOPh resulted in very low yields for all substrate groups, and is likely due to poor solubility of KOPh in THF. However, TBAT proved to be an excellent initiator (see Scheme 2). Reactions with nitrosubstituted benzaldehydes and acetophenones resulted in Z90% RCY of the corresponding products. It is noteworthy that three of these compounds previously failed to label using [ 18 F]fluoroform. 16 Furthermore, the unsubstituted and methoxy-substituted benzaldehydes were [ 18 F]trifluoromethylated with decent to good yields (73 AE 4% and 43 AE 6%, respectively). Unsubstituted and methoxy-substituted acetophenones and benzophenones still showed low yields (r10%). On the contrary to the nitro-substituted derivatives, these substrates are reported to react very well with [ 18 F]fluoroform.
All optimization reactions were carried out using aliquots of a [ 18 3 SiCl, KHMDS and/or toluene could contaminate the desired reaction.
In conclusion, we report the first synthesis and application of fluorine-18 labelled Ruppert-Prakash reagent. [ 18 F]Me 3 SiCF 3 was synthesized with radiochemical yields of over 90% and radiochemical purities of 495% (starting from [ 18 F]fluoroform) within 20 minutes. Reaction with benzaldehydes, acetophenones and benzophenones provided a complementary substrate scope to the previously reported method using [ 18 F]fluoroform, 25 enabling the synthesis of compounds that were not previously accessible. It should be noted that the relatively high amounts of precursor (200 mmol) require further optimization for routine application in PET tracer synthesis, and will be the focus of our future work. Since Ruppert-Prakash reagent is a widely used trifluoromethylation agent in organic synthesis, the development of [ 18

Conflicts of interest
There are no conflicts to declare.