Determination of Total Radiochemical Purity of [18F]FDG and [18F]FET by High-Performance Liquid Chromatography Avoiding TLC Method

The goal of this work was to present two high-performance liquid chromatography (HPLC) method that could be applied for the determination of the total radioactive purity of 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) and O-(2-[18F]fluoroethyl)-L-tyrosine ([18F]FET). The separation of [18F]fluoride ions, [18F]FET and [18F]FET intermediate was accomplished on LiChrosper RP-18, 250 × 4 mm, 5 µm (Merck) analytical column. For mobile phase 10 mM potassium dihydrogen phosphate buffer at pH7 (A) and acetonitrile (B) was used: 0–2 min: 15% B; 2–12 min: 85% B; 12–15 min: 15% B, respectively. Analysis of [18F]FDG was performed using LiChrosper 100 NH2, 250 × 4.5 mm, 5 µm (Merck) analytical column. The initial mobile phase composition was 10 mM KH2PO4 buffer (pH7) and acetonitrile (15:85, v/v) and the acetonitrile ratio was decreased to 15% at 2 min after the sample injection and held for 5 min. Complete elution of [18F]fluoride ions from stationary phases could be achieved by adding 10 mg/mL K[19F]F to radioactive samples in a ratio 1:1 during the sample preparation. Recovery of [18F]fluoride ions ranged from 99.5 to 100.6%. The validation of the developed methods showed good results for linearity (r2 = 0.9981–0.9996), specificity (RS = 3.7–10.2), repeatability (%Area RSD% = 1.2–4.3%) and limit of quantitation (LOQ = 1.6–4.5 kBq). During the cross-validation similar radiochemical purity values were obtained by the novel HPLC methods and thin layer chromatography performed according to the recommendations of the Ph. Eur. monographs.


Introduction
Positron emission tomography (PET) is a frequently used imaging technique in the clinical diagnosis of cancer. Substances labeled with radioactive isotopes are used as radiotracers for PET examinations to visualize pathophysiological processes. Predominantly, 18 F isotope is applied for the production of radiopharmaceuticals for human use. The workhorse of PET imaging is [ 18 [1].
The synthesis of 18 F-labeled radiopharmaceuticals is basically performed using one-or two-step procedure. Radiofluorination of precursors bearing sulfonate leaving groups proceeds via S N 2 nucleophilic substitution mechanism in polar aprotic solvents. The active pharmaceutical ingredient could be obtained by hydrolysis of the intermediate radioactive compound containing protective functional groups. As a rule, the final solution beside the main component might contain free [ 18 F]fluoride ions and the intermediate compound. The amount of these impurities fundamentally determine the radiochemical purity of the final product [2].
The application of radiopharmaceutical batches for human PET examinations could be started after the determination of radiochemical purity, which is a critical parameter of the quality control system. For this purpose, the separation of radiochemical components and assessment of their radioactivity values are generally performed by chromatographic methods. Several high-performance liquid chromatography procedures could be found in the literature for PET radiopharmaceuticals. In the case of [ 18 F] FET, mainly reverse stationary phase (RP) as well as biner mixtures of buffer solutions and organic solvents are applied for the determination of radiochemical purity [3][4][5][6][7][8][9][10]. Orlovskaya [3]. On the other hand, Mueller et al. using the same eluent mixture and LiChrospher RP-18e stationary phase (Merck) could not identify unreacted [ 18 F]fluoride in the purified final solution [4]. Wang et al. reported that besides the main [ 18 F] FET peak one unidentified radiochemical impurity could be detected applying Gemini C18 column (Phenomenex) with an eluent of 12% Ethanol and 88% 50 mM NaH 2 PO 4 buffer (pH 6.8) [5]. In the case of [ 18 F]FDG, mostly ion chromatography is used for the determination of radiochemical purity [11]. Due to this method [ 18  and RP-HPLC are tested for the separation of radiochemical compounds. Since the analysis of [ 18 F]fluoride ions in radiopharmaceutical samples is a great challenge for radioanalytical chemists [18], special emphasis is given to elution profile optimization and investigation of the recovery of [ 18 F]fluoride ions.

Chemicals
Acetonitrile was supplied from VWR. Sodium hydroxide, potassium fluoride, potassium dihydrogen phosphate and phosphoric acid were obtained from Sigma Aldrich. Fluoroethyl-L-tyrosine, 3,4-dimethoxy-L-phenylalanine, 2-deoxy-2-fluoro-D-glucose, 2-deoxy-2-chloro-D-glucose and 2-deoxy-2-fluoro-D-mannose was purchased from ABX (Radeberg, Germany). All chemicals used for the experiments were HPLC grade and applied without additional purification. Water used for dilution was provided by a Milli-Q purification system and controlled for the content of organic impurities.

H[ 18 F]Fluoride
Production of 18

[ 18 F]FDG and Acetylated [ 18 F]FDG
The Hamacher process was applied for the synthesis of [ 18 F]FDG and the intermediate product [19]. [ 18 F]fluoride ions were extracted from target water onto SepPak Light QMA Cartridge (Waters) and subsequently eluted with a mixture of 15 mg Kryptofix 222 (Merck) in 0.8 mL acetonitrile (Sigma) and 3.5 mg K 2 CO 3 (Sigma) in 0.25 mL water. The effluent was delivered to the reaction vessel and azeotropically evaporated to dryness. In the next step, 20 mg of 1,3,4,6-tetra-O-acetyl-2-O-trifluoromethanesulfonylβ-D-mannopyranose precursor (mannose triflate, ABX Advanced Biochemical Compounds, Germany) in 1 mL of anhydrous acetonitrile was added to the reaction vessel. The nucleophilic substitution reaction was carried out in 1 min at 85 °C. The produced acetylated [ 18 F]FDG was used directly for chromatographic method optimization or for further synthesis of [ 18 F]FDG which was proceeded with the evaporation step and subsequent removal of protective acetyl groups by hydrolysis with HCl (1 M, 1 mL) in 3.5 min at 121 °C. C18 Plus and Alumina N SepPak Cartridges (Waters) were used for the purification of the reaction mixture. The pH was adjusted using AG11A8 ion retardation resin (Bio-Rad). The purified solution was diluted with 14 mL water. Finally, the solution was filtered through 0.22 μm Millex-GS Syringe-Driven Filter Unit (Millipore). The synthesis time was 23 min and up to 130 GBq radioactivity was obtained. The production of [ 18 [8]. [ 18 F]fluoride ions were harvested from target water using SepPak Light QMA Cartridge and the subsequent elution of the radioactivity was accomplished with 0.075 mol/L tetrabutylammonium-bicarbonate solution. The mixture was azeotropically dried and 10 mg of the precursor was added in 2 mL acetonitrile to the reactor. The substitution was performed at 110 °C in 5 min. The produced [ 18 F]FET intermediate was used directly for HPLC method development or for further synthesis of [ 18 F]FET which proceeded with the evaporation of acetonitrile and subsequent removal of protective groups by hydrolysis with HCl (0.2 M, 3 mL) in 8 min at 120 °C. SepPak Light Alumina N, SepPak WAX and SepPak HLB Cartridges were used for purification of the raw reaction mixture. The product was eluted from cartridges with 12 mL ethanol and water mixture (1:19, v/v). Finally, the product was filtered through 0.22 μm Millex-GS Syringe-Driven Filter Unit (Millipore). The synthesis time was 55 min and up to 50 GBq radioactivity was obtained.
Thin-layer chromatography measurements were carried out on miniGita Star device supplied with a GMC counting tube and controlled by miniGita control software (raytest).

Methods
The analysis of [ 18 F]FET samples according to the recommendation of Ph. Eur. monograph [17] was performed using the following reverse phase HPLC method. LiChrosper RP-18, 250 × 4 mm, 5 µm (Merck) was applied for the stationary phase. The elution was performed using carbon dioxide-free water (A) and acetonitrile (B) for the mobile phase. The separation was accomplished according to the following gradient steps. At the beginning, the content of mobile phase A was 90% and at the point of 10 min it was started to reduce to 5% A within 10 min and held for a further 10 min. The injected volume was 20 µL and the flow rate was adjusted to 1 mL/min. The following anion-exchange HPLC method was used for the determination of radiochemical purity of [ 18 F]FDG which was in compliance with Ph. Eur. monograph [16]. Dionex CarboPack PA10 250 × 4 mm, 10 µm (Thermo Scientific) was applied for the separation of radiochemical components. For the mobile phase 0.1 mmol/L sodium hydroxide was used at 1 mL/min flow rate. The injected volume was 20 µL.

Sample Preparation
Solutions of radioactive reference compounds extracted from cyclotron or automated synthesizer were directly injected into HPLC system or applied to TLC plate. During the investigation of recovery of [ 18 F]fluoride ions 10 mg/ mL potassium fluoride solution was added to the radioactive sample with 1:1 ratio. The radioactivity values of samples for the validation procedure were adjusted by dilution with water or by injection of samples at a different time interval. The radioactivity concentration of the samples was up to 1 GBq/mL. The measurements were performed in triplicate despite linearity tests.

HPLC Method Development for Analysis of [ 18 F]FET
The separation and detection of identified radiochemical components of [ 18 [17]. Despite the fact that [ 18 F]fluoride ions could be eluted from LiChrosper RP-18 column using gradient elution with water and acetonitrile mixture, a poor peak shape was obtained and the recovery was less than 85%. This is the reason why an additional TLC method is usually applied for the determination of the total radiochemical purity of radiopharmaceuticals. To overcome this problem in this work the optimization of chromatographic conditions was carried out for the development of an HPLC method for analysis of [ 18 F]FET and avoiding TLC procedure.

Optimization of Mobile Phase pH for Effective Elution of [ 18 F]Fluoride Ions
The retention and peak shape of [ 18 F]fluoride ions performing RP-HPLC analysis highly depends on the pH of the buffer component of the eluent [18]. As a first step of the method development, the effect of pH of mobile phase was examined on elution parameters of [ 18 F]fluoride ions using LiChrosper RP-18, 250 × 4 mm, 5 µm (Merck) analytical column. 10 mM buffer solution of potassium dihydrogen phosphate was used as eluent A and acetonitrile as eluent B. The elution was accomplished according to the following gradient profile. At the beginning mobile phase A ratio was 85% and at the point of 10 min after the injection it was started to reduce to 15% within 10 min and held for further 10 min. The injected volume was 20 µL and the flow rate was adjusted to 1 mL/min. The pH of phosphate buffer was adjusted to 3, 4, 5, 6 and 7. After the analysis of samples, no chromatographic peak was detected at pH 3 and 4. Apparently, [ 18 F]fluoride ions were adsorbed on column material in these cases since the pK a is 3,17 for hydrogen fluoride and the ion suppressed form of [ 18 F]fluoride has a higher affinity to the stationary phase. At pH5 a broad peak was appeared at about 5 min retention time and with a tailing factor of 3.2 (Fig. 1). Increasing the pH of the buffer solution to 6 and 7 the retention time was decreased to 3.3 and 2.2 min and the tailing factor decreased to 2.1 and 1.3, respectively. Consequently, it is recommended to adjust the pH of the buffer component of the mobile phase to 7 for effective elution of [ 18 F]fluoride ions from the stationary phase.

Enhancement of [ 18 F]Fluoride Ions Recovery by Co-injection with K[ 19 F]
It is well known from the literature, that [ 18 F]fluoride ions tend to retain in the stationary phase at appropriate pH [18]. However, complete elution of radioactivity from the HPLC system is important for accurate determination of the radiochemical purity. In our case, the recovery of [ 18 F]fluoride ions was assessed by comparison of decay corrected peak areas obtained from the analysis of samples injected onto the analytical column and a procedure in which the sample bypassed the HPLC column and flowed directly through the radioactivity detector [20]. In accordance with the Ph. Eur. recommendations, using water and acetonitrile mixture for mobile phase the obtained [ 18 F]fluoride recovery was less than 85% (

Chemical Purity Test of [ 18 F]FET
Applying UV detection the developed HPLC method is also eligible for simultaneous determination of chemical purity of [ 18 F]FET since the system suitability test could be performed successfully according to the requirements of Ph. Eur. monograph [17]. Analyzing a reference solution of 3,4-dimethoxy-L-phenylalanine and fluoroethyl-L-tyrosine with a concentration of 250 and 50 µg/mL, respectively, the compounds could be separated with a resolution of 4.1 which is higher than the minimum limit of 2.0 (Supplementary information, Fig. S4). During the validation of the chemical purity test linearity was verified in the concentration range of 0.01-300 µg/mL and 0.01-60 µg/mL for 3,4-dimethoxy-L-phenylalanine and fluoroethyl-L-tyrosine, respectively. The obtained regression coefficients (r 2 ) were in the range of 0.9985 and 0.9990. The repeatability of peak area and retention time of components was determined by injection of 6 replicates of the reference solution with a concentration of 250 and 50 µg/mL of 4-dimethoxy-L-phenylalanine and fluoroethyl-L-tyrosine, respectively. The calculated relative standard deviations (RSD % ) were in the range from 0.2 to 0.7%. Additionally, intermediate precision was in the range of 0.1 and 1.1% which was obtained from 6 injections of the reference solution performed on different days. The accuracy was determined as a recovery of adjusted 60 µg/mL concentration of FET which ranged from 96.1 to 97.3% (n = 6). Limit of quantitation of 3,4-dimethoxy-L-phenylalanine and fluoroethyl-L-tyrosine was 0.02 and 0.05 µg/mL, respectively. Robustness was examined at small changes in optimal parameters of flow rate, eluent ratio, pH and ionic strength of mobile phase. The obtained resolutions of 3,4-dimethoxy-Lphenylalanine and fluoroethyl-L-tyrosine peaks were in the range of 2.8 and 4.6 (Supplementary information, Table S2).
The HPLC analysis of [ 18 F]FET batches revealed that no relevant non-radioactive impurities could be detected in samples (Supplementary information, Fig. S5).

Analysis of [ 18 F]FDG by HILIC Method
The total radiochemical purity of [ 18 F]FDG could not be determined exclusively using anion-exchange HPLC method of the Ph. Eur monograph due to the hydrolysis of acetylated [ 18 F]FDG under the basic chromatographic conditions [16]. Additionally, the authors have found that applying 0.1 mmol/L sodium hydroxide as mobile phase [ 18 F] fluoride ions could be eluted from Dionex CarboPack PA10 250 × 4 mm, 10 µm (Thermo Scientific) column with the recovery of up to 97%. To perform the determination of the total radiochemical purity of [ 18 F]FDG a novel method was developed based on hydrophilic interaction liquid chromatography.

Influence of pH of Mobile Phase on the Elution of [ 18 F]Fluoride Ions
The HILIC method was developed based on aminopropyl modified LiChrosper 100 NH 2 , 250 × 4.5 mm, 5 µm (Merck) analytical column. The investigation of [ 18 F]fluoride elution was a critical part of the chromatographic optimization procedure. Interestingly, using water (A) and acetonitrile (B) mixture as mobile phase in 85:15 ratio at 1 mL/min flow rate was not appropriate for elution of [ 18 F]fluoride ions as it was observed in the case of LiChrosper RP-18, 250 × 4 mm, 5 µm (Merck) column. To solve this problem mobile phase A was changed to 10 mM potassium dihydrogen phosphate solution. Measurements were accomplished at isocratic conditions and the injected volume was 20 µL. The pH of the buffer solution was adjusted to 3, 4, 5, 6 and 7. After the injection of samples, it was found that [ 18 F]fluoride was not eluted from the column at pH 3 and 4. On the other hand, a broad peak was detected at 6.4 min retention time and with a tailing factor of 3.0 at pH5 (Fig. 3). Increasing the pH of the buffer solution to 6 and 7 the retention time was decreased to 3.8 and 3.2 min and the tailing factor decreased to 2.8 and 2.3, respectively. Consequently, phosphate buffer should be applied instead of water and the pH is recommended to maintain at 7 for effective elution of [ 18 F]fluoride ions from the stationary phase.

Effect of K[ 19 F] Additive on the Recovery of [ 18 F] Fluoride Ions
The determination of recovery of [ 18 F]fluoride ions during the elution from LiChrosper 100 NH 2 stationary phase was accomplished by comparison of decay corrected peak areas obtained from two subsequent measurements in the presence After the optimization of gradient elution, the developed method was validated. Linearity was verified in the radioactivity concentration range of 5-902 MBq/mL (Supplementary information, Fig. S6-S8). The regression coefficients were obtained between 0.9981 and 0.9995. The repeatability was determined by injection of 6 replicates of reference solution containing the radiochemical ingredients. The relative standard deviation of %Area for [ 18

Analysis of 2-Deoxy-2-Fluoro-D-Glucose (FDG) and 2-Deoxy-2-Chloro-D-Glucose (ClDG)
The developed HILIC-HPLC method was tested for applicability in the determination of chemical purity of [ 18 F]FDG. Non-radioactive compounds were identified by a refractive index detector. It was found that using isocratic elution at mobile phase composition of 15:85 (v/v) of 10 mM KH 2 PO 4 buffer (pH7) and acetonitrile FDG and ClDG could be separated with a resolution of 1.8 at 50 µg/mL of excipient concentration (Supplementary information,  Fig. S9). Unfortunately, 2-deoxy-2-fluoro-D-glucose and 2-deoxy-2-fluoro-D-mannose (FDM) co-eluted during the HPLC analysis. During the validation of the method a linearity test was performed using the concentration range of 10-100 µg/mL of FDG and ClDG. The obtained regression coefficients (r 2 ) were in the range of 0.9910 and 0.9986. The repeatability of peak area and retention time of components were evaluated by injection of 6 replicates of the reference solution at 50 µg/mL concentration of FDG and ClDG. The calculated relative standard deviations (RSD % ) were in the range from 0.1 to 3.7%. Additionally, intermediate precision was in the range of 0.1 and 4.3% which was obtained from 6 injections of the reference solution performed on different days. The recovery of FDG and ClDG concentrations were in the range of 90.2 to 112.8%. Limit of quantitation of FDG and ClDG was 40.6 and 44.3 µg/mL, respectively. Robustness was examined at small changes in optimal parameters of flow rate, eluent ratio, pH and ionic strength of mobile phase. The obtained resolutions of FDG and ClDG peaks were in the range of 1.7 and 2.4 (Supplementary information, Table S4). The HPLC analysis of [ 18 F]FDG batches revealed that concentrations of FDG and ClDG were below the limit of quantitation.

Conclusions
The determination of total radiochemical purity of [ 18  difference in polarity of the measured components gradient elution was applied to achieve reasonable peak resolution and measurement time. In the case of [ 18 F]fluoride ions 10 mM potassium dihydrogen phosphate buffer at pH7 is recommended to use as part of the mobile phase to obtain a good peak shape. Additionally, at least 10 mg/mL K[ 19 F] F should be added to the radioactive sample for complete elution of [ 18 F]fluoride ions from the stationary phase. The developed HPLC methods were validated, and good results were obtained for linearity (r 2 > 0.998), specificity (R S > 3.7) and repeatability (RSD % < 4.3). The application of proposed high-performance liquid chromatography methods could be extended to other 18 F-radiopharmaceuticals for the determination of total radiochemical purity and avoiding additional TLC method.