|Home | About | Journals | Submit | Contact Us | Français|
The aim of this study was to synthesize 18FDG in some consecutive runs and check the quality of manufactured radiopharmaceuticals and to determine the distribution of metallic impurities in the synthesis process.
For radiopharmaceuticals the general requirements are listed in European Pharmacopeia and these parameters have to be checked before application for human use.
Standard methods for the determination of basic characteristics of radiopharmaceuticals were used. Additionally, high resolution γ spectrometry was used for the assessment of nuclidic purity and inductively coupled plasma with mass spectrometry to evaluate metallic content.
Results showed sources and distribution of metallic and radiometallic impurities in the production process. Main part is trapped in the initial separation column of the synthesis unit and is not distributed to the final product in significant amounts.
Produced 18FDG filled requirements of Ph.Eur. and the content of radionuclidic and metallic impurities was in the acceptable range.
Positron emission tomography (PET) is a dynamically developing imaging method of nuclear medicine, which allows to diagnose metabolic changes in human body. PET diagnostic techniques use β+ emitting isotopes for labeling biologically active compounds and track their distribution in a living organism. Due to its relatively long half-life (110 min), 18F is the most commonly used radioisotope in PET and is produced in medical cyclotrons (mostly proton-deuteron 10–20 MeV machines). Nowadays, the most widely used radiopharmaceutical for diagnostic procedures using PET is 18FDG, the glucose derivative labeled with 18F, which applications are regularly reviewed1, 2, 3, 4, 5, 6, 7 and standardized.8
One of the most important aspects of working with 18FDG is a short time (about 30 min) that can be spent on quality control and release procedures, thus the speed, simplicity and reliability of developed analytical methods are critical factors. For radiopharmaceuticals, the general requirements are listed in European Pharmacopoeia9 and these parameters have to be checked before application for human use. Short-lived radiopharmaceutical preparations may be released before completion of some tests, specified in individual monographs. The aim of this study was to synthesize 18F-FDG in some consecutive runs and check the quality of manufactured radiopharmaceuticals. Several tests were performed to determine chemical and radiochemical impurities, chemical identity and other adequate parameters for parenteral formulation. During production process, trace amounts of metallic radioisotopes are produced due to radio activation on the metal target housing. In addition, the distribution of metallic impurities on synthesis and dispensing module was measured by gamma spectroscopy and level of non-radioactive metals was determined with inductively coupled plasma with mass spectrometry (ICP-MS).
18FDG was synthesized in six independent runs with standard method from mannose triflate with alkaline hydrolysis as initially proposed in Ref. 10. Cyclotron GE PETtrace840 with high yield niobium target (General Electric, Uppsala, Sweden) was the source of anionic fluorine. Standard produced activity in 18O(p,n)18F reaction with 16.5 MeV proton beam at 40–45 μA, was 4.0 ± 0.2 Ci (140.6–155.4 GBq) after 120 min of irradiation and was transferred to the GE MXFDG unit (General Electric, Liege, Belgium), where the synthesis and purification were performed. In the synthesis path, the 18F-fluoride solution was passed through an ion exchange column, which trapped anions. Cations, including some metal contaminants, were collected with the recovered enriched water. [18F]fluoride was then eluted to the reaction vessel with a mixture of potassium carbonate and Kryptofix 2.2.2, then water was removed by azeotropic distillation with acetonitrile and 18F reacted with mannose triflate. After alkaline hydrolysis, the solution was purified with sequence of C18-RP and alumina columns and eluted with water. The final product was formulated with saline, passed through a 0.22 μm filter and dispensed in automatic module DDS-Vials (Tema-Synergie, Italy). Starting materials were, ready-to-use, Ph.Eur. compliant kits, obtained from ABX (Radeberg, Germany).
Identification tests were performed as described in Ref. 9. For γ-spectrometry, high resolution germanium detector GMX-20190-P with digital signal processor (DSPEC, Ortec) and GammaVison software was used. 2 μL sample was applied on silica plate, fixed in a holder and inserted into a 5 cm Pb shielded, low-background housing. Spectra was recorded for 5 min.
Half-life was measured with Atomlab300 (Biodex, USA) dose calibrator: 300 μL (1.2–1.5 GBq) sample was crimped in penicillin vial, fixed in a standard vial holder and measured in triplicate at 20 min intervals.
Identity of manufactured 18FDG was confirmed by comparison of retention time to the certified reference standard (CRS) of main compound (ABX, Radeberg, Germany).
Radionuclidic purity and radionuclidic impurities were determined using gamma spectroscopy with a high resolution germanium detector GMX-20190-P with a digital signal processor (DSPEC, Ortec) with GammaVison software. Efficiency and energy calibration was performed with 241Am (255.162 kBq at the day of measurements), 137Cs (203.425 kBq at the day of measurements) and 152Eu (260.733 kBq at the day of measurements) sources at 13.9, 17.8, 26.47, 59.67, 121.9, 244.8, 344.37, 661.7, 788.98, 964.13, 1085.92, 1112.17 and 1408.14 keV, respectively. For sources and samples, a universal holder, fixed in 14 cm from a detector window was constructed and located in a fully shielded (5 cm Pb) low-background housing.
The spectra were recorded in 10,800 s each for final product and purification cartridges used during the synthesis of 2-[18F]FDG: ion exchange columns Accel Plus QMA Sep-Pak™, used for preconcentration and separation of 18F from target, reverse phase separation columns Sep-Pak™ C-18 RP used in a basic hydrolysis and purification process of FDG, alumina columns Sep-PakTM N Plus for ionic contaminants removal. Isotopes were identified on the basis of the characteristic γ-emissions. Only for final determination of impurities in 18FDG, the time was extended to 21,600 s. Each peak was analyzed by marking the region of interest and recording the energy, count rate and background corrected area.
The γ-ray spectra for radionuclidic purity test A was recorded immediately after synthesis and test B was performed 24 h after irradiation. Recorded activities were calibrated at the end of synthesis (EOS).
The γ-ray spectra of the residual radionuclides were collected 72 h (36 times the 18F half-life) after irradiation, because by that time the 18F activity decreased to a level comparable to longer-lived compounds and did not hamper the spectra recording.
Radiochemical purity (test A) was performed with an ion chromatography system ICS-5000+ (Thermo Scientific, former Dionex) with a pulsed amperometric detector and radiometric detector (GabiStar, Raytest, Germany). 20 μL sample was injected via a manual multiport valve. The separation was done on Thermo Scientific Carbopac PA-10 column (250 mm × 4.0 mm i.d., 10 μm), with 0.1 M NaOH (CO2-free) as a mobile phase and 1 mL/min flow rate. Data acquisition and processing was preformed with Chromeleon software.
Radiochemical purity determination (test B) was conducted with a thin layer chromatography system Bioscan MiniScan B-MS-100 (Canberra Packard) with Flow-count B-FC-100 (Canberra Packard) data processing unit. 2 μL of sample were introduced on a silica gel plate and developed on 8 cm path in a 95:5 acetonitrile–water mixture.
To ensure the quality of measurements, methods were developed and validated with certified reference standards (CRS) of main compound and impurities.
Isotonicity was determined by cryogenic osmometer (Knauer, Germany) in 50 μL samples. Measurements of pH were conducted with MulitiSeven pH-meter (Mettler-Toledo, Germany) with microelectrode vs. 4.01, 7.01 and 9.21 buffers (Hamilton, UK).
Since the 7th edition the Ph.Eur has replaced in fludeoxyglucose monograph thin-layer chromatographic (TLC) detection of Kryptofix 2.2.2 with the simple color spot test.11 Silica gel stripes, immersed in iodoplatinate reagent were used for optical comparison of color reaction versus pattern (blank, saline, sample, positive sample) with pass/fail criteria.
Head-space gas chromatography was performed on 7890A Agilent system, equipped with a J&W HP-5 column (30 m × 0.32 mm × 0.25 μm) and a flame ionization detector (FID). GC system was supplied with an H2 (30 mL/min) from CFH200 generator (Peak Scientific, UK), ensuring 99.9995% of hydrogen purity, zero-air (400 mL/min) from a Jun-Air 0F301-4B generator (Jun-Air, UK) and He (6.0, Air Products, flow rate 25 mL/min). Gas chromatographic system was operated at the following conditions: oven temperature 40 °C for 1 min, inlet temperature 150 °C, detector temperature 180 °C.
The samples were injected via 7694E Agilent head-space system. For a comfortable sample application, 5-microliter single-use capillaries (Drummond Scientific, USA) were used for transfer the radioactive sample to 10 mL penicillin vials with aluminum caps and PTFE/Si septa (Agilent). Head-space injector was set to 80 °C for 2 min, then sample was equilibrated for 0.2 min and transferred to the GC system. The loop and transfer line were heated to 105 °C and 110 °C, respectively. Chemstation software was used for operation of chromatograph, acquisition and processing of data.
Sterility test was performed by an external microbial laboratory as specified in Ref. 9. For bacterial endotoxins the gel-clot method (limit test) was used, where gel formation indicated the presence of endotoxins. The test was performed with Pyrogent™ set (Lonza, Belgium) The gel formation in sequence of samples (blank, sample, water spiked with endotoxins, spiked sample) after 1 h incubation at 37 °C was inspected manually.
Inductively coupled plasma with mass spectrometry (ICP-MS, Perkin Elmer Elan 9000) was used for the determination of Cu, Fe, Pb, Ag, Co, Mn, Cd, Zn, Cr in 18FDG and enriched water samples. Calibration curves in the range of up to 100 μg/L with four transitive points were prepared. The internal standard, 103Rh was used for minimizing interferences.
Methanol, ethanol and acetonitrile used for experiment were GC grade and received from Merck. Water used for dilution was provided from validated MilliQ system and controlled for the content of organic impurities.
18F was identified by recording the principal γ-peak at 510 ± 0.3 keV (Fig. 1a) and determination of half-life (108.8 ± 0.9 min). 18FDG was confirmed, comparing the retention times of standards, observed in the reference chromatogram with retention time of the principal signal in the radiochromatogram (Fig. 1b and c). Recorded retention times were 10.38 min for the standard and 10.55 min for the principal peak.
Gamma spectrum recorded for identification was used for determination of radionuclidic purity. Presence of any peaks with an energy different from 511 keV was checked and, except signals coming from Pb-X-rays (range 70–80 keV), none was found. That confirmed a better than 99.9% radionuclidic purity. Then 18FDG sample was left for 24 h to decay the fluorine and again impurities were tested with no significant peaks, except traces of 18F. Only about 2 Bq signals of 51 Cr, 52Mn and 56Co were observed, representing 10−8% of total 18F activity. Above mentioned results complied with radionuclidic purity tests A and B.9 Similar levels of radionuclidic impurities were reported in other papers.12, 13, 15 When impurities were not found, theoretical content was calculated by the determination of minimum detectable activities for all radionuclidic impurities found in the previous steps of the production. The activity was estimated to 3 kBq.12
More detailed study was conducted on radionuclidic impurities distribution along the production process with high resolution γ-spectrometry. Radionuclidic impurities were evaluated 72 h post-irradiation to remove dominating 18F and 13N by decay. It leaves out as well very short half-life (in the order up to tens of minutes) or metastable impurities, generated from target body and foils during the bombardment, covering, but not restricted to 50,52mMn, 60,62Cu, 54Co and technetium (92,94mTc).
Samples from the production process: of irradiated water (500 μL), ion exchange columns Accel Plus QMA Sep-PakTM, reverse phase separation columns Sep-PakTM C-18 RP used in the basic hydrolysis, the same column type used for purification process of FDG and alumina columns Sep-PakTM N Plus were used to identify radionuclides and their distribution in 18FDG routine production (Fig. 2). List of identified radionuclides, their content in the recovered 18O water and columns along the synthesis process are collected in Table 1.
The main source of radionuclide impurities in 18FDG production is proton beam and secondary neutrons interactions with Havar® foil, containing Co (42.5%), Cr (20%), Mn (1.6%), Mo (2%), Ni (13%), W (2.8%), Fe (18.1%) and traces of carbon. Theoretical investigation showed 627 nuclear reactions possible13 but practical results limited the number of isotopes to principal radionuclides and primary proton beam interaction. As dominating, (p,n) reactions on naturally occurring isotopes are generally proposed12, 14, 16, 22 with (p,α)12, 16, 22 and (n,α), (n,γ)22 supplementary paths.
The set of isotopes agreed in literature consists of 52Mn, 54Mn, 56Co, 57Co, 58Co, 96Tc, 183Re and these isotopes are the main contributors to activity captured in foil and radiochemistry setup. Some isotopes are hardware fingerprints: 109Cd for silver targets, 48V in niobium target systems with titanium foils.15 Distribution of other radionuclides was highly variable and depended on production parameters (target, proton energy and beam current), examined samples (foils, disposable cassettes, wastes, final product) and methodology of determination (equipment, time elapsed from end of bombardment (EOB) to measurement).
In this work 15 radioisotopes were identified in the QMA column. Aside from those mentioned above, 7Be, 51Cr, 55Co, 56Ni, 95mTc, 182Re, 186 Re were found. 96Tc, 56Co and 56Ni were the main contributors of residual activity with 70% share in total column activity (3.120 MBq). Only 6 isotopes were found in 18O water after passing the QMA: 51Cr, 52Mn, 54Mn, 56Co, 57Co, 58Co with domination of 58Co and significant contribution from 52Mn, 56Co and 57Co. Tc and Re isotopes were not observed, which could be easily explained by chemical properties, where negatively charged perrhenate and pertechnetate ions are immobilized on QMA cartridge (Fig. 3). Activity of 18Owater was 1.400 MBq (44% of QMA, 30% total activity) but individual activities of dominating isotopes were higher than in the QMA column, due to the formation of Mn and Co cationic complexes. Presence of these isotopes in QMA, which is an anionic exchanger, can be explained by the formation of hydroxycomplexes in close to neutral pH. A similar pattern and distribution (10% of activity in 18O water, ca. 85% trapped in QMA) were observed in Ref. 12, but reversed proportions were found in Ref. 15, confirming high variability in impurities distribution. In this work 51Cr was mainly trapped in QMA but was found as well in 18O water, with traces migrating along the production path. This could be related to complex chromium equilibria in water and organic solvents and thus the distribution inconsistency was observed as in other papers.
Interesting was the peak identified at the energy of 477.606 keV which corresponded to 7Be. The isotope was identified only in Ref. 23 without any hypothesis about its origin. In another paper,22 the 7Be occurrence was explained by proton bombardment of trace amounts of 12C, present in the Havar foil. The reaction 12C(p,3n3p)7Be was proposed as a source of 7Be, but requires relatively high, for PET cyclotrons, proton energies. Anyhow, in this case 7Be should have been observed by other authors (Table 2).
Reasonable source of 7Be could be natural 10B, found as chemical impurity in 18O water16 in reaction 10B(p,α)7Be. Boron is a ubiquitous element, common in environment and widespread in industry. The control of boron levels could not be efficient due to complex chemistry and numerous application in nuclear science and technology. Exact identification of the source is quite difficult, but it would explain why 7Be was not detected by other authors.
The spectra of the other purification columns (carbon and aluminum) along the production process show <5 Bq peaks of 51Cr, 52Mn and 56Co. This is a result of significant purification based on cationic and anionic properties of radiometallic impurities, where cations are collected in recovery water and anions are trapped in the QMA column. Residual radionuclidic impurities are separated and collected in further purification steps, resulting in that only 51Cr, 52Mn and 56Co with activity in the range of 1–2 Bq could be recorded in the final product.
Decrease of radionuclide impurities along the production cycle was reported as in other papers,12 as well as the radionuclide impurities produced differ depending on the target vessel materials and even when beam energy is different on the same target type. These results suggested the necessity of estimating the radionuclides produced for every combination of proton energy and 18FDG unit using different target vessels.
The most notable is the remarkable reduction of chemical and radionuclidic impurities when using the Nb-sputtered Havar foil as compared with the impurities generated when using a straight Havar foil. Radiometallic impurities were decreased 10–25 times.16 It did not influence patients’ safety, because the majority of contaminants were excluded from the production process but a 6.4 percent increase in the average 18FDG yield was observed.17
Another isotope, which has to be taken into consideration is β emitting 3H from 18O(p,3H)16O reaction. But other works verified that 3H is not detected in the final product and patients receiving 18FDG do not receive any extra internal exposure from tritium.15, 18
Concluding, all impurities were efficiently eliminated from the final product and met radionuclidic purity set in Ref. 9.
2-chloro-2-deoxy-d-glucose (ClDG, impurity A), 18FDG and 2-18fluoro-2-deoxy-mannose (18FDM) were identified by ion chromatography by a qualitative comparison with the reference solution, containing maximum available concentration of ClDG, FDG and FDM and determined quantitatively from 6-points calibration curves for each standard. The retention times were 9.16, 10.38 and 11.19 min, respectively, which corresponded to relative retention factors vs. FDG: 0.88 for FDM and 1.08 for ClDG. Although the regulation for ClDG requests only pass/fail criteria, the quantitative measurements were done and average value did not exceed 0.026 mg/V (limit 0.5 mg/V). 18FDG and 18FDM were determined by HPLC with radiometric detection, where the principal peak of 18FDG was observed at 10.55 min, with 18FDM signal at 9.40 min and residual 18F at 6.2 min (Fig. 1c). The average content of 18FDM was 0.49%.
Six consecutive runs were analyzed in triplicate each. Free fluoride-18 (Rf = 0.0), 18FDG (Rf = 0.55) and acylated-18FDG (Rf > 0.85) were determined quantitatively. Typical radiochromatogram is presented in Fig. 4. Average content of 18FDG was 96.80 ± 0.44%, with the lowest result 96.01% and >95% required in Pharmacopeia. Average content of impurities was 3.2% ± 0.44%.
Organic solvents were determined with headspace gas chromatography (HS-GC). In all samples acetonitrile and ethanol were determined quantitatively, with traces of methanol observed in some samples. For the determination, originally developed method was used, where the separation of organic solvents was completed in 1 min (Fig. 5) with total time of analysis below 4 min, where the pharmacopoeial method or procedures described in literature require about 20 min for a complete separation alone. Detailed results are presented in Table 3.
The values are comparable to other formulations19, 20 which are not fortified with ethanol. To check the quality of measurements, three 18F-FDG batches, available on the market, were analyzed in parallel with manufacturer's laboratory, working as commercial 18F-FDG provider, using a standard Ph.Eur. method, with 20 min of chromatographic separation. The results are presented in Table 4 and show good correlation. However, our proposed method has a better limit of detection and throughput of samples.
Heavy metal content was determined in six 18FDG samples and respective recovery water from each run. First, a quantitation limit was determined as an average blank ± 6 standard deviation. Then determination of 9 heavy metals: Cu, Fe, Pb, Ag, Co, Mn, Cd, Zn and Cr in samples was conducted. Summarized values of the quantitation limit are presented in Table 5. Presented results clearly indicate that the metal impurities are not a serious threat for the quality of the produced FDG and are significantly lower than set by the regulatory office.21 Most of the metallic impurities are concentrated in the ion exchange column in the inlet or migrate with enriched water to recovery and do not influence significantly the synthesis process. A slight increase in the concentration of copper, zinc and chromium in the final product is observed. The first two are common trace contaminants and are difficult to remove as their probable source is a saline solution used for the final formulation. Increasing concentration of chromium can be explained by the presence of elements of stainless steel in the dispensing line.
Isotonicity, pH, Kryptofix content, sterility and endotoxin test, as generally known procedures are summarized in Table 3. All parameters were according to the quality criteria for 18FDG.
All of the impurities were efficiently determined and then eliminated in the 18FDG synthesis process, and the final product was purified from main radionuclidic and metallic impurities. Final product meets the requirements set by relevant regulations.