Introduction

The quality assurance of positron-emitting tracers used in positron emission tomography (PET) is performed in accordance with guidance documents such as United States Pharmacopeia/National Formulary (USP/NF) and European Pharmacopeia (EP). Although slight differences among the documents were discussed previously [1], basic requirements include characters, radionuclidic identity, radionuclidic purity, radiochemical purity, chemical purity, pH, residual solvents, bacterial endotoxin, and sterility. Most tests can be finished before the release of PET tracers; however, tests such as those for sterility and endotoxins, especially in the case of 11C-tracers, and radionuclidic purity depending on the measurement methods are completed after the release.

Regarding to radionuclidic purity, γ-ray spectrometry is required for the detection and quantification of impurities. For the preparation of fludeoxyglucose (18F) injection (18F-FDG), sodium fluoride (18F) injection (18F-NaF), and ammonia (13N) injection (13N-NH3), the radionuclides 18F and 13N are usually produced by proton irradiation of 18O-H2O and 16O-H2O, respectively, and it is well known that these aqueous target solutions contain very small amounts of γ-emitters with a longer half-life than 18F (109.8 min) [2 and references therein]. Most of these impurities are excluded from the final injections, but some impurities potentially remained in the preparation processes without purification using distillation or high-performance liquid chromatography (HPLC). Their levels may be much lower than that of 18F or 13N of interest, and the criteria for radionuclidic purity of PET tracers such as 18F-FDG and 18F-NaF required by the USP/NF (no less than 99.5 %) [3, 4] and EP (minimum 99.9 %) [5, 6]. Radionuclidic purity is also required for 11C-labeled compounds such as 11C-flumazenil and 11C-raclopride; however, it is hardly considered that these tracers derived from 11C-CO2 via the chemical reactions and HPLC purification may contain radionuclides other than 11C. The requirement for radionuclidic identity and purity for PET tracers synthesized chemically from gaseous materials seems to be substantially meaningless from a scientific point of view. However, systematic studies on what kinds of radionuclides and radiochemical species are included in the radiosynthesis processes of many PET tracers have not been reported.

The aim of this study was to clarify the radionuclides included in the radiosynthesis processes from target materials into final products. We focused on PET tracers labeled with four conventional radionuclides, 11C, 13N, 15O, and 18F, and investigated radionuclides and radiochemical species in the target materials, positron-labeled precursors, and five PET tracers. The target materials included 11C-, 15O-, and 18F-labeled gases and 13N- and 18F-target solutions. The radionuclides investigated were short half-life positron emitters and longer half-life γ-emitters. We discuss the significance of examinations of radionuclidic identity and purity in the quality control of PET tracers.

Materials and methods

General

We used a 20 MeV cyclotron (CYPRIS HM-20), target and synthesis system developed by Sumitomo Heavy Industry (Tokyo, Japan). Target folders and materials used and the expected nuclear reactions are summarized in Supplementary Tables 1 and 2, respectively. The production of positron emitters and their labeled compounds were according to the standard specifications of Sumitomo Heavy Industry. Because the radionuclidic identity and purity could be changed depending on several parameters such as irradiation time, beam current, and target materials, we set the integrated beam currents to be suitable for routine production of each PET tracers in clinical use, and changed depending on the analyses to accurately evaluate or to avoid unnecessary radiation dose. The detailed information is summarized in Supplementary tables.

Production of 11C-CO2, 11C-methyl iodide, and 11C-methylated compounds

11C-CO2 was produced by proton irradiation of N2 containing 0.5 % O2 at a pressure of 0.8 MPa.

11 C-CO 2 gas preparation

After 15-min irradiation with 10–50 μA, 11C-CO2 target gas was passed through a coiled stainless steel tube [0.5 mm inner diameter (i.d.) × 90 cm length] immersed in liquid Ar (boiling point −186 °C) to trap 11C-CO2 (boiling point −196 °C). After returning the stainless steel tube to room atmosphere, the 11C-CO2 gas was recovered in a Tedlar® bag for about 60 s with a 30 ml/min N2 flow to measure radioactivity decay.

Syntheses of 11 C-methyl iodide and three 11 C-methylated tracers

After 15-min irradiation with 5–50 μA, 11C-CH3I and 11C-methylated compounds were prepared using a multipurpose synthesizer CFN-MPS100 (Sumitomo Heavy Industries). 11C-CO2 gas was purged with a 30 ml/min N2 flow into 0.1 ml of 0.1 M LiAlH4 in tetrahydrofuran (ABX, Radeberg, Germany). After removal of tetrahydrofuran using a 200 ml/min N2 flow heated to 180 °C, 0.5 ml of HI was added and the solution was heated for 45 s. The 11C-CH3I produced was recovered in a Tedlar® bag for about 60 s with a 30 ml/min N2 flow to measure radioactivity decay. Radiosyntheses of 11C-methionine, 11C-ITMM, and 11C-CB184 are described in the Supplementary material.

Production of 15O-labeled gases and H2O

15O-Gas was produced by deuteron irradiation of N2 containing 0.5 % CO2 or 2 % O2 at a pressure of 0.29 MPa. After 10-min irradiation with 10–30 μA, 15O-gas under continuous irradiation was transferred to a 15O-gas/water synthesis system CYPRIS-G (Sumitomo Heavy Industries) with a 300 ml/min flow to produce the three 15O-gases described below.

15 O-CO 2 production

Target 15O-gas containing 0.5 % CO2 was passed through a crushed carbon granules (size, between 1.0 and 3.35 mm) column (9.8 mm i.d. × 150 mm length) at 400 °C to convert traces of 15O-CO to 15O-CO2.

15 O-O 2 production

Target 15O-gas containing 2 % O2 was passed through three columns of crushed carbon granules (size, between 1.0 and 3.35 mm, 9.8 mm i.d. × 150 mm length), soda lime (9.0 mm i.d. × 150 mm length) and molecular sieve 5A 1/16 (Wako Pure Chemical Industries, Tokyo, Japan; 9.8 mm i.d. × 150 mm length) at room temperature. Traces of 15O-CO and 13N-N2O were removed by these columns.

15 O-CO production

Target 15O-gas containing 2 % O2 was passed through two columns of charcoal (9.8 mm i.d. × 150 mm length) heated at 1000 °C and soda lime (9 mm i.d. × 150 mm length) at room temperature. Traces of 15O-CO2 are removed using the soda lime.

15 O-H 2 O synthesis

15O-O2 gas produced above was passed through a palladium black (about 20 mg, Wako Pure Chemical Industries) column (9.0 mm i.d. × 2 mm length) heated at 130 °C and a molecular sieve 5A 1/16 column (9.0 mm i.d. × 70 mm length) at room temperature. This gas at a flow rate of 500 ml/min is mixed with H2 at a flow rate of 7 ml/min and passed over a Pd wire (Aldrich 267163, Atlanta, GA) column (9.0 mm i.d. × 15 mm length) heated at 180 °C, and then bubbled into a sterile vial containing physiological saline.

Production of 18F-F2 and 4-10B-borono-2-18F-fluoro-l-phenylalanine (18F-FBPA)

18F-F2 was produced by deuteron irradiation (30 min for analysis and 120 min for production of 18F-FBPA) of neon gas containing 0.6 % F2 gas at a pressure of 0.3 MPa (Supplementary Table 1). Radiosynthesis of 18F-FBPA is described in the Supplementary data.

Production of 13N-ammmonium and 13N-NH3

13N-Ammonium solution was produced by proton irradiation of natural water containing 10 mM ethanol at a pressure of 1.8 MPa [7].

13 N-NH 3 production

13N-NH3 was produced using a 13N-NH3 synthesizer N100 (Sumitomo Heavy Industries). After 10-min irradiation with 10–50 μA, 13N-ammonium solution was passed through a Sep-Pak Accel Plus CM Plus Short cartridge (Waters, Milford, MA). The cartridge was washed with 10 ml water for injection twice, and the 13N-ammonium was eluted with 12 ml physiological saline for injection.

Production of 18F-fluoride, 18F-NaF, and 18F-FDG

18F-Fluoride solution was produced by proton irradiation of enriched 18O-water (18O: ≥98.0 atom %, Taiyo Nippon Sanso, Tokyo, Japan) at a pressure of 1.8 MPa.

18 F-NaF production

18F-NaF was prepared using a CFN multipurpose synthesizer CFN-MPS100 (Sumitomo Heavy Industries). After 3–45 min irradiation with 5–20 μA, 18F-fluoride solution was passed through a Sep-Pak Accel Plus QMA Plus Light cartridge (Waters). The cartridge was washed with 10 ml water for injection, and 18F-fluoride was eluted with 1 ml physiological saline for injection and diluted with 14 ml physiological saline.

18 F-FDG production

After 7–45-min irradiation with 20–50 μA, 18F-FDG was prepared using an 18F-FDG synthesizer F300 (Sumitomo Heavy Industries).

Determination of positron emitters in target materials and radiolabeled compounds

Immediately after the end of bombardment (EOB), gaseous and aqueous target materials were recovered in a Tedlar® bag for gases and in a small glass vial for liquids, respectively, and the radioactivity decay was successively measured at first short intervals (15, 20, 30, and 60 s), then intermediate intervals (2, 5, 10, and 20 min) and finally long intervals (0.5, 1, and 2 h) using a radioisotope calibrator CRC®-712 (Capintec, Ramsey, NJ). Immediately after the recovery of 11C-CO2 and 11C-methyl iodide gases and synthesis of radiolabeled compounds, the radioactivity decay was also successively measured, and percentages of radionuclides were decay-corrected at the EOB, at the time of recovery, or the end of synthesis (EOS).

Gas chromatography

Radiolabeled gases were analyzed using a gas chromatography (GC) system mounted in a 15O-gas/water synthesis system CYPRIS-G (Sumitomo Heavy Industries). Radiolabeled gases were analyzed on a Porapak Q column (50/80 mesh, 3 mm i.d. × 4 m length, Shimadzu, Kyoto, Japan) and a molecular sieves 13X column (30/60 mesh, 3 mm i.d. × 4 m length, Shimadzu) at 55 °C with a He flow at 55 kPa pressure (about 110 ml/min). CO2 and N2O were absorbed using the molecular sieves 13X.

Ion chromatography

13N-ammonium and 18F-fluoride target solutions, 13N-NH3, 18F-NaF, and 18F-FDG were successively analyzed 3–5 times using an ion chromatography (IC) system (Prominence HIC-SP, Shimadzu) immediately after EOS (2–8 min) up to 73 min to assign radionuclides of radioactive peaks. For analysis of anion a Shim-pack IC-SA2 (4.6 mm i.d. × 250 mm length, Shimadzu) was used at 30 °C by elution with of 12 mM NaHCO3/0.6 mM Na2CO3 = 1/1 at a flow rate of 1.0 ml/min using an electric conductivity detector (suppressor method) and radioactivity monitor. For analysis of cations a Shim-pack IC-C4 (4.6 mm i.d. × 150 mm length, Shimadzu) was used at 40 °C by elution with of 3.5 mM oxalic acid/1 mM 18-Crown-6 ether = 1/1 at a flow rate of 1.0 ml/min by a non-suppressor method.

Detection of long half-lived gamma emitters in positron emitting compounds

For analyses of γ-emitters of PET tracers we set the integrated beam currents to be suitable for routine clinical use: 13N-NH3, 10-min irradiation with 20 μA; 18F-FDG, 45-min irradiation with 50 μA; 11C-methionine, 15-min irradiation with 10 μA; 18F-FBPA, 120-min irradiation with 25 μA. The irradiation condition of 18F-fluoride target solution and 18F-NaF was same as that of 18F-FDG. 13N-ammonium target solution was produced by 30-min irradiation with 50 μA. Target materials and radiolabeled compounds were diluted with H2O to a 100 ml total volume, and the presence of impure metal ions emitting γ-rays was measured using a high-purity Ge detector at FUJIFILM RI Pharma (GMX20190-P, SEIKO EG&G, Tokyo, Japan) for 18F-target materials and 18F-labeled compounds and at Tokyo Nuclear Services (GEM20200, SEIKO EG&G) for others. The radionuclides evaluated included 15 γ-emitters: 22Na, 48V, 51Cr, 52Mn, 54Mn, 55Co, 56Co, 57Co, 58Co, 57Ni, 67Ga, 93mMo, 95Tc, 96Tc, and 181Re. Each sample was measured twice for 1 and 10 h at 2–3 and 3–11 days, respectively, after EOB.

Results and discussion

Positron-emitting radionuclides and radiochemicals in 11C-CO2 target gas and 11C-labeled compounds

The total radioactivity decay curve of 11C-CO2 target gas (n = 3) is plotted in Fig. 1. There were three phases on a log scale. The decay line between 30 and 400 min (Fig. 1b) was well fitted to the decay of 11C (t 1/2 = 20.4 min). After subtracting 11C-radioactivity extrapolated to time zero at EOB from total radioactivity, the residual radioactivity decay line between 10 and 40 min fitted well to the decay of 13N (t 1/2 = 9.97 min) (Fig. 1c). When further subtracting 13N-radioactivity as above, the decay line between 1 and 4 min was slightly lower than the decay line of 14O (t 1/2 = 70.6 s) (Fig. 1d). However, when considering the ratios of isotope: 14N vs 15N vs 16O in the target material and the cross section of each isotope: 14N(p, n)14O vs 15N(p, n)15O vs 16O(p, d)15O, we concluded that Fig. 1d showed the decay of 14O (t 1/2 = 70.6 s) rather than that of 15O (t 1/2 = 2.04 min). Then, from the radioactivities of 11C, 13N, and 14O at EOB the ratios of each radioisotope were calculated as 11C, 47.3 %: 13N, 28.5 %: and 14O, 24.2 % (Supplementary Table 3). Radioactivity decay curves plotted on a log scale for 11C-CO2 gas, 11C-CH3I, 11C-methionine, 11C-ITMM, and 11C-CB184, showed only one component 11C (Supplementary Table 3).

Fig. 1
figure 1

Radioactivity decay curves of 11C-CO2 target gas. a Total radioactivity decay (square) on a linear scale. b Total radioactivity decay (square) on a log scale indicates a half-life of 11C. The 11C-radioactivity was extrapolated to time zero and subtracted from total radioactivity. c The residual radioactivity decay (triangle) on a log scale indicates a half-life of 13N. The 13N-radioactivity was further subtracted from the residual radioactivity. d The final radioactivity decay (circle) on a log scale indicates a half-life of 14O

Full size image

11C-CO2 target gas was analyzed by GC (n = 4). Radioactive peaks were detected in the retention times of O2/N2/CO, (2.4 min), CO2, (5.4 min) and N2O (6.5 min) on Porapak Q and those of O2, (1.5 min), N2, (2.2 min), and CO (5.3 min) on Molecular Sieve 13X (Fig. 2). When considering radionuclides detected by measurement of radioactivity decay, we assigned that CO2 (average 60.4 %) and CO (1.0 %) were labeled with 11C, N2 (33.6 %) and N2O (0.1 %) were with 13N, and O2 (4.4 %) was 14O (Supplementary Table 4).

Fig. 2
figure 2

Gas chromatograms of 11C-CO2 target gas. 11C-CO2 target gas was directly applied to a gas chromatography system mounted in a 15O-gas/water synthesis system CYPRIS-G (Sumitomo Heavy Industries). Left side analysis using a Porapak Q column; right side analysis using a Molecular Sieve 13X column. Upper row thermal conductivity; lower row radioactivity

Full size image

We did not analyze 11C-CO2 gas condensed in a stainless steel tube in liquid Ar by GC. However, the radioactivity decay of the 11C-CO2 gas showed a single component. 13N-N2 and 11C-CO in the 11C-CO2 target gas were not trapped and 14O-O2 was hardly trapped at liquid Ar temperature when considering the difference of boiling points: N2, −196 °C; CO, −192 °C; Ar, −186 °C; O2, −183 °C. A trace of 13N-N2O (0.1 %, boiling point of N2O, −88 °C) might lower than the detection limit during the condensation process of the 11C-CO2 target gas. Thus, we concluded that radionuclidic pure 11C-CO2 was used for the 11C-labeling of PET tracers after the condensation process of the 11C-CO2 target gas at liquid Ar temperature.

Positron-emitting radionuclides and radiochemicals in 15O-gases and 15O-H2O

Radioactivity decay of 15O-CO2, 15O-O2, and 15O-CO (n = 4 for each) and 15O-H2O (n = 3) was measured for about 130 min until reaching to background levels (Supplementary Fig. 1 for 15O-CO2). All three 15C-gases showed a biphasic decay curve with t 1/2 = 10 and 2 min. Contamination by 13N in 12 preparations of three 15C-gases was 0.5–0.7 % of total radioactivity at EOB (Supplementary Table 5). 15O-H2O decayed away with t 1/2 = 2 min.

15O-Gases were analyzed using GC (Supplementary Fig. 2). 15O-CO2 (n = 6), 15O-O2 (n = 3), and 15O-CO (n = 6) included 0.5 % 15O-CO plus 0.7 % 13N-N2, 0.6 % 13N-N2, and 0.7 % 13N-N2, respectively, as radioactive impurities (Supplementary Table 6). Similar radioactive impurities in each of the 15O-gases were found by several groups [8 and references therein]; however, our method could exclude radiolabeled N2O using Molecular Sieve 13X.

We did not analyze 15O-H2O by GC. However, radioactive contamination in the 15O-H2O cannot be considered, because inert 13N-N2 was completely removed in the synthesis process of 15O-H2O using 15O-O2.

Detection of a radionuclidic impurity 13N2 in a range of 0.5–0.8 % of total radioactivity in all three 15O-gases indicates that criteria are necessary for both radionuclidic and radiochemical impurities in quality assurance of 15O-gases.

Positron-emitting radionuclides and radiochemicals in 18F-F2 target gas and 18F-FBPA

Radioactivity decay of 18F-F2 target gas (n = 3) showed a biphasic decay curve (Supplementary Fig. 3). The decay line up to 1300 min fitted well to the decay of 18F, and that between 1.5 and 4 min fitted well to the decay of 23Ne (37.2 s). The amount of 23Ne (83.2 %) was much larger than that of 18F-F2 (16.8 %) at EOB (Supplementary Table 7). 23Ne is inert for chemical reactions. Radioactivity decay of 18F-FBPA showed only a component of 18F. In the present study, we added F2 at 0.6 % in Ne which was sufficient to prevent adsorption of 18F-F2 on the inner surface of the target and tubing for transfer to the synthesis system. We consider that examination of radionuclidic impurity could be omitted in the quality assurance of 18F-labeled PET tracers derived from 18F-F2.

Positron-emitting radionuclides and radiochemicals in 13N-ammonium target solution and 13N-NH3

Radioactivity decay of an 13N-ammonium target solution (n = 3) showed a three-phase curve (Supplementary Fig. 4). The decay line between 100 and 1200 min fitted well to the decay of 18F, and those between 10 and 70 min and between 3 and 8 min fitted well to the decay of 13N and 15O, respectively, and the ratios of 13N, 15O, and 18F were 58.3, 41.1, and 0.6 %, respectively, at EOB (Supplementary Fig. 4; Table 8). It should be noted that during the recovery of 13N-ammonium target solution in a glass vial by He pressure, radioactive gases such as 13N-N2 [7] could be removed, if they were present in the target solution. We detected 15O and 18F as contaminants in the 13N-ammonium target solution as reported previously [9–11]. On the other hand, Tornai et al. found 17F (t 1/2 = 64.5 s) in addition to 15O in an 13N-ammonium target solution produced by proton irradiation with an incident energy of 21 MeV, but not with an incident energy of 10.7 MeV [12].

13N-NH3 (n = 3) showed a biphasic decay curve with t 1/2 = 10 and 110 min, and the ratio of 18F was 0.005 % of total radioactivity at EOS (Supplementary Fig. 5; Table 8). It was pointed out that the elapsed time when the ratio of 18F in each of three 13N-NH3 preparations exceeded 0.1 % of 13N was 41–51 min after EOS.

13N-Ammonium target solution and 13N-NH3 were analyzed using IC (n = 3). In the first of successive analyses of the 13N-ammonium target solution on an anion-exchange column, more than four minor radioactive peaks, in addition to a main radioactive peak in the void volume, were detected (Fig. 3a). The retention times of the three peaks were consistent with those of fluoride (3.9 min), nitrite (4.3 min), and nitrate (8.9 min). Based on previous studies [7, 11, 13, 14], the nitrite and nitrate were probably labeled with 13N. Candidate compounds for other minor radioactive compounds (void-1 and unknown-2) may be 13N-NH2OH [14] and 15O-H2O [12]. We detected 15O in the 13N-ammonium target solution by measuring radioactivity decay (Supplementary Table 8); however, we could not assign 15O-labeled components in successive analyses. IC may barely detect low levels of 15O-labeled components. The percentage of the main peak in the void volume containing 13N-ammonium was 96.5 % at EOS, when the contribution of 15O as a radioactive contaminant was ignored (Supplementary Table 9). 13N-NH3 showed a radioactive peak on an anion-exchange column.

Fig. 3
figure 3

Ion chromatograms of 13N-ammonium target solution (a) and 13N-NH3 (b). Left side analysis on an anion-exchange Shim-pack IC-SA2 column; right side analysis using a cation-exchange Shim-pack IC-C4 column. Upper row conductivity; lower row radioactivity

Full size image

In successive analyses of the 13N-ammonium target solution on a cation-exchange column, three radioactive peaks were detected for these periods (Fig. 3a). The peak area corresponding to the retention time of ammonium (5.6 min) decreased in accordance with a half-life of 13N, but the areas of other two peaks (retention time, 1.8 and 2.1 min) decreased more slowly, indicating that each of two included 13N- and 18F-chemicals. The 2.0 min component contained of 18F-fluoride, as described later. The percentage of the 13N-ammonium peak was 92.1 % at EOS on the premise there were no 15O-labeled components (Supplementary Table 9).

In 13N-NH3, only one radioactive peak was detected in the void volume and at the retention time of ammonium on anion- and cation-exchange columns, respectively (Fig. 3b). The contaminant 18F, probably 18F-fluoride, detected by measuring radioactivity decay (0.005 % of total radioactivity), could not be monitored using a radioactivity detector even in the last of successive analyses (60–65 min at EOB) probably because of the very low level available to detect.

In quality assurance of 13N-NH3, 18F-fluoride is noticed as a radionuclidic impurity but not radiochemical impurity. A limited time of 13N-NH3 for clinical use may be at longest a three half-life period (30 min) after the EOS. Up to that time the ratio of 18F increases, but is a very low level (about 0.05 % of total radioactivity) that is acceptable for clinical use.

Positron-emitting radionuclides and radiochemicals in 18F-fluoride target solution, 18F-NaF and 18F-FDG

A radioactivity decay curve of 18F-fluoride target solution (n = 3) showed three decay lines of 18F, 13N, and 17F (t 1/2 = 64.5 s), and the ratios of 18F, 17F, and 13N were 87.2, 12.0, and 0.8 %, respectively, at EOB (Supplementary Fig. 6; Table 10). Radioactivity decay of 18F-NaF (n = 4) showed two components 18F (99.5 %) and 13N (0.5 %) (Supplementary Fig. 7), and 18F-FDG (n = 3) had a component of 18F (Supplementary Table 10). Contamination with 17F probably disappeared during the preparation periods of 18F-NaF (8–11 min) and 18F-FDG (about 30 min). The elapsed times when the ratio of 13N in each of the four 18F-NaF preparations became less than 0.1 % of 18F were calculated to be 21–32 min after EOS. In another 18F-FDG sample we measured radioactivity decay until complete decay for over 3 days, and confirmed that no radioactivity with a longer half-life than 18F was detected using measurement with a radioisotope calibrator.

18F-Fluoride target solutions, 18F-NaF and 18F-FDG were analyzed by IC (Supplementary Fig. 8). Short half-life 17F (t 1/2 = 64.5 s) was hardly detected by IC because of the period to the start analysis and the retention time of 18F-fluoride. On an anion-exchange column 18F-fluoride target solution and 18F-NaF showed a minor radioactive peak of nitrate (8.9 min, 0.6 %) and two minor radioactive peaks corresponding to nitrite (4.3 min, 0.2 %) and nitrate (1.5 %), respectively, in addition to 18F-fluoride. 18F-FDG showed a main peak (2.5 min) and minor unknown and fluoride peaks. Because the 18F-fluoride increased slightly in successive analyses (data not shown), it was probably derived from radiolysis of 18F-FDG [15].

On a cation-exchange column, a main radioactive peak for all 18F-fluoride target solution, 18F-NaF, and 18F-FDG, was eluted at 2.0 min, and a minor radioactive peak was detected in the 18F-fluoride target solution (1.8 min) and 18F-FDG (2.3 min). In successive analyses, the 1.8-min peak area decreased faster than a half-life of 18F, indicating that the peak included 13N- and 18F-chemicals. The percentages of each component on both ion-exchange columns are summarized in Supplementary Table 11.

In clinical use of 18F-NaF, a preparation time is very short and 13N was detected in it as radionulidic impurities. It is preferable that quality assurance or release of 18F-NaF for clinical use is delayed for appropriate time after the EOS until 13N-radioactivity decays to a negligible level.

Detection of long half-life γ-emitters in positron-emitting compounds

The presence of impure metal ions emitting γ-rays in aqueous target materials is well known [2 and references therein]. We measured long half-life γ-emitters in aqueous target solution and related compounds using a high-purity Ge detector (Supplementary Table 12).

In the 13N-ammonium target solution (average yield = 13.5 GBq, n = 3), 11 of 15 γ-emitters investigated were detected in the order of 10−6–10−9 compared with 13N of interest: 51Cr, 52Mn, 54Mn, 55Co, 56Co, 57Co, 58Co, 57Ni, 95Tc, 96Tc, and 181Re. Most of these nuclides were removed in the synthesis process of 13N-NH3 (average yield = 7.4 GBq, n = 3), and very small amounts of five γ-emitters (51Cr, 52Mn, 55Co, 56Co, and 58Co) in the order of 10−13–10−14 compared with 13N of interest remaining in 13N-NH3. Theoretically, 13N-ammonium target solution was expected to contain the same γ-emitters as 18F-fluoride target solution described below, because we used the same target system: a Nb-body target with a Havar foil. A finding that two γ-emitters, 67Ga and 93mMo, were not detected in 13N-ammonium target solution was possibly because of the shorter irradiation times. In the production of 13N-NH3, extremely small amounts of γ-emitters were left, because we limited the irradiation time to a practical level for clinical use of 13N-NH3.

In 18F-fluoride target solution (average yield = 93.8 GBq, n = 3), 13 of 15 γ-emitters were detected in the order of 10−8–10−10 compared with 18F of interest: 51Cr, 52Mn, 54Mn, 55Co, 56Co, 57Co, 58Co, 57Ni, 67Ga, 93mMo, 95Tc, 96Tc, and 181Re. Avila-Rodriguez et al. found 11 γ-emitters other than 54Mn and 67Ga in 18F-fluoride target solution using a Nb-body target with a Havar foil [2]. Undetected two γ-emitters 22Na and 48V were included as potential radionuclides which were present in the production process of PET tracers depending on the target body and foils [the report on the waste of short half-life radionuclides by Japan Radioisotope Association (May 2003, in Japanese)]. For example, 48V was detected using a silver-body target with titanium foil [16]. In the synthesis processes of 18F-NaF (average yield = 78.3 GBq, n = 3), six γ-emitters were under detection, two (54Mn and 56Co) were reduced, and the other five (51Cr, 93mMo, 95Tc, 96Tc, and 181Re) were not reduced. In 18F-FDG (average yield = 63.4 GBq, n = 3), nine γ-emitters were below the level of detection, two (56Co and 181Re) remained, and two (95Tc and 96Tc) were reduced.

We examined contaminant by γ-emitters in PET tracers derived from gaseous target materials: 11C-CO2 gas via 11C-CH3I and 11C-CH3OTf, and 18F-F2. Selected PET tracers were 11C-methionine and 18F-FBPA. The former was prepared without HPLC separation. Both tracers showed no γ-emitters. These findings were reasonably well expected, because it was hardly considered that radiolabeled metal contaminants were derived from gaseous target materials.

Radionuclidic identity and purity in the quality control of PET tracers

In the present study, we investigated radionuclides and radiochemical species in radiosynthesis processes from target materials to PET tracers. In gaseous and aqueous target materials, a couple of positron emitters were found. However, regarding to radionuclidic purity the tracers produced by chemical reactions: 11C-labeled methionine, ITMM and CB184, 15O-H2O, 18F-FBPA, and 18F-FDG had only one positron emitter of interest. On the other hand, 15O-gases, 18F-NaF, and 13N-NH3 which were produced by passing through catalytic columns or ion-exchange cartridges without chemical reactions, contained minor positron impurities in addition to the positron emitter of interest. Regarding to positron impurities, it was noted that measurement of radioactivity decay was more sensitive than IC. IC could not detect 18F in 13N-NH3 and 15O-components in 13N-target solution. The radionuclidic purity of each of the three 15O-gases was less than 99.5 %; however, an impurity, that was inert 13N-N2, was reasonably acceptable for clinical use, because it interfered in neither measurement of aimed functions/imaging quality nor the welfare of subjects. The radionuclidic impurities of 18F-NaF: 13N-nitrite and 13N-nitrate decreased rapidly to be less than 0.1 % of 18F by 21–32 min after EOS as discussed above. Thus, 30–60 min after EOS, the radionuclidic purity of 18F-NaF satisfied the acceptance criteria for 18F-NaF required by USP/NF (greater than 99.5 %) and EP (minimum 99.9 %). The amounts of γ-emitters found in 18F-FDG, 18F-NaF, and 13N-NH3 were extremely low compared with acceptance criteria for the radionuclidic purity required by USP/NF and EP. The ratios of radionuclidic impurities observed in PET tracers could be slightly changed depending on the integrated beam currents; however, we considered that examinations of the radionuclidic identity and purity were not necessarily for quality assurance of PET tracers in routine production when the radionuclides were produced using a specific cyclotron with fixed energy of proton or deuteron beam (CYPRIS HM-20 in this study) and fixed target and synthesis systems, and when once the radionuclides in target materials and final PET tracers were reasonably analyzed from a scientific point of view. Regarding 15O-gases, radiochemical impurities 13N2 in 15O-O2 and 15O-CO corresponded with radionuclidic impurity, and the ratios of radionuclidic impurity in three 15O-gas species were 0.5–0.8 % of total radioactivity. The criteria of radionuclidic purity for PET tracers such as these 15O-gases should be determined based on quality for the measurement of the aimed functions and welfare of subjects, but not in the same way as other tracers such as 18F-FDG and 18F-NaF.

Conclusion

Gaseous target materials used for the production of PET tracers had considerable levels of several short half-life positron emitters in addition to the positron of interest, and aqueous target materials had extremely low levels of many longer half-life γ-emitters in addition to positron emitters. Five tracers produced from gaseous target materials via chemical reactions had no radionuclidic impurities, whereas 18F-FDG, 18F-NaF, and 13N-NH3 produced from aqueous target materials had extremely low levels of several γ-emitters as well as impure positron emitters. Three 15O-labeled gases had impure positron-emitting 13N-N2 (slightly over 0.5 %).