3.1. Production of [18F]PSMA-1007 via a Two-Step Procedure on Trasis AllInOne Radiosynthesiser
It was already known from our preclinical experiments that [18
F] 6-fluoropyridine-3-carboxylic acid (F-Py-TFP) exhibits a bad reactivity towards Precursor 2 (conjugation yield of [18
F]F-Py-TFP to the amino group of glutamic acid was actually always low compared to other amino groups) [26
]. We assume that the amino-group in Precursor 2 is deactivated, most probably by lactamisation. An alternative precursor with protected carboxylic acid(s), at least in the terminal position, should lead to an improvement of the yield, however also necessitating an additional step for deprotection. Therefore, our preliminary experiments focused on improving the reaction conditions. The initial radiolabeling reaction delivering [18
F]F-Py-TFP proved to be optimal under the conditions reported by Olberg et al. [27
]. However, we slightly increased the reaction temperature to 50 °C and have chosen a reaction time of 10 min. It turned out that the yield of the conjugation to Precursor 2 could be increased under dry conditions using DIPEA as the base (pyridine, triethylamine and Kryptofix 2.2.2/K2
were also tested, but led to decreased yields), which was then used for full automation. Since the yield obtained by this procedure was sufficient for the first-in-man studies, no alternative precursor was considered at this stage.
The activity trails recorded by the five sensors inside the synthesiser are shown in Figure 4
. The position of the detector probes, as well as their colour code are indicated in Figure 6 (see below in Materials and Methods) by the radiation symbols and the respective coloured fields below.
The blue line indicates the activity trapped on the QMA-cartridge. As expected, the activity is efficiently eluted to the reaction vessel using TBAHCO3
-solution (green trail). The green trail (activity in Reaction Vessel 1) shows some noisy behaviour during the drying process, which is caused by the changes of the activity per volume in Vessel 1 occurring throughout the process (geometric factors). About 20% of the radioactivity is left in Reaction Vessel 1 after labelling and extraction of the mixture, which is a behaviour for no carrier added [18
typically observable in glass reaction vessels [28
]. Roughly 30% of the activity from Reaction Vessel 1 is then trapped on the MCX cartridge (purple line). Actually, we expected a somewhat higher value of adsorption of approximately 45% at this stage. This deviation may be caused by sputtering of the reaction mixture during the drying process, resulting in uncomplete contact of the reaction solvent with the activated [18
-complex in Reaction Vessel 1. Further, a small loss of activity is caused by the first elution of the cartridge with 500 µL acetonitrile. However, it was known from preliminary experiments that this fraction elutes with unidentified small particles presumably originated from the cartridge material. Considering the following cartridge drying, final HPLC purification and sterile filtration, there was no risk for contamination of the final injection solution at all; however, with respect to the risk of clogging the cassette at one of those barriers, we decided to discard this fraction anyway. Finally, [18
F]F-Py-TFP is eluted from the cartridge into Reaction Vessel 2 (grey line) for the second step of the labelling procedure (conjugation to Precursor 2). Obviously, there is no unexpected behaviour except for the “peak” at the end of the grey activity trail, which is also caused by geometric factors.
Before HPLC purification, the basic reaction mixture in Reaction Vessel 2 containing the acidic [18
F]PSMA-1007 had to be acidified and diluted. Therefore, addition of 6 mL water containing 10 µL TFA proved to be sufficient. All major impurities including 6-[18
F]Fluoronicotinic acid (formed by hydrolysis of [18
F]F-Py-TFP), non-reacted Precursor 2 and [18
F]F-Py-TFP were effectively separated by the final semi-preparative HPLC (Figure 5
). However, it turned out that two minor radiochemical impurities are formed during the reaction (altogether approximately 3%), which could not be separated by HPLC. We believe that addressing this problem would at least require a solvent change to an acetonitrile/acidified water mixture. Furthermore, measures to decrease injection volume or even a solvent change before injection aiming towards higher resolution of the semi-preparative HPLC could be necessary, which would add substantially to the complexity of the process. Therefore and because of the low levels of impurities, we decided to accept those side products. In none of the batches produced by the described method were significant chemical impurities observed.
In summary, we successfully and reliably produced 24 batches of [18F]PSMA-1007 for first-in-man PET/CT studies. Although yields for the reaction presented here are low, the general feasibility of a two-step radiofluorination with the prosthetic group [18F]F-Py-TFP using an AllInOne module has been demonstrated. Impurities arising from [18F]F− activation during the initial labelling step are effectively separated by the cartridge extraction process before subsequent coupling of the prosthetic group. Anyhow, for precursors showing a better reactivity towards [18F]F-Py-TFP, we estimate that on a daily basis, a multi-dose batch production of the respective radioligands is feasible. Furthermore, the procedure proved to be excellent for the setup of respective new libraries of radiotracer variants bearing the 6-[18F]Fluoropyridine-3-carboxy moiety as a radiolabel-bearing subunit, as well as the preliminary (“bridging”) synthesis procedure during clinical translation.
3.2. Production of [18F]PSMA-1007 by Direct One-Step Synthesis on the GE Tracerlab FX FN Module, GE TRACERlab MX, NEPTIS Mosaic-RS and IBA SYNTHERA+
Although the used synthesisers are quite different, procedures for the production of [18F]PSMA-1007 injection solution by direct radiofluorination are still quite comparable on the selected systems and therefore discussed together.
The reaction proved to be reproducible, delivering the product in good radiochemical yields between 25% and 80% after cartridge separation with slightly higher yields on the IBA SYNTHERA+. The higher yield on the SYNTHERA+ is most probably due to shorter fluidic pathways causing fewer losses in tubings and manifolds. On all modules, the product was obtained in excellent synthesis times well below 55 min. Upscaling of the synthesis using start activities of approximately 90 GBq resulted in activity yields of up to 49 GBq and was finally accomplished on IBA SYNTHERA+. During upscaling, no affection of the radiochemical yield was observed. However, it should be noted that the addition of sodium ascorbate as a stabiliser to avoid radiolysis in the final formulation is necessary when product activities of more than 20 GBq are produced (threshold activity concentration of approximately 1 GBq/mL). Using sodium ascorbate addition stability was proven over a time period of 8 h, which is a typical shelf-life for fluorine-18 radiopharmaceuticals (in prior experiments, 100 mg were used).
Importantly, the cartridge separation is the crucial step for the quality of the final product. During this separation process, multiple subtle washing steps have to be applied. After fixation of the product and impurities from the crude reaction mixture, the more hydrophilic side products are removed by washing with 5% EtOH solution in a first step. Subsequently, the product is fractionally eluted with 30% EtOH (25% EtOH in the case of IBA SYNTHERA+). The first 30% EtOH fraction (3 mL) is still contaminated with significant amounts of impurities and therefore has to be discarded. Furthermore, the volume of the second fraction (5 mL) was adjusted (limited) for avoiding the introduction of more lipophilic impurities. Using this approach, impurities were only present in trace amounts well below the limits according to recent Eur. Ph. monographs (important note: acceptance criteria were chosen based on typical monographs for fluorine-18 radiopharmaceuticals (e.g., [18F]FET Monograph No. 07/2015:2466 Eur. Ph.) and the monograph on radiopharmaceutical preparations (Monograph No. 07/2016:0125)). Therefore, we recommend using these specifications when applying the radiosynthesis in a PET radiopharmacy. Furthermore, one should be aware that the product quality can be further improved by careful adjustment of the volumes for washing and eluting the cartridges. Application of typical quality control procedures for the release of radiopharmaceuticals (see below) revealed that all produced batches could have been released for clinical use without restriction of any kind.
Principally, the product can also be purified by suitable semipreparative HPLC procedures. However, the additional time including separation using the mobile phase and reformulation of the product can be estimated with approximately 45 min, which equals 25% product loss owing to decay only.
In summary, we developed a precursor and a unique synthetic procedure for the highly economic production of [18F]PSMA-1007 injection solution, [18F]PSMA-1007 being the next generation 18F-tracer for the diagnosis and noninvasive staging of PSMA-positive prostate cancer. The produced batches meet all acceptance criteria according to recent Ph. Eur. Upscaling was successfully conducted to batch sizes of approximately 50 GBq with proven stability over 8 h. Thus, the clinical routine even in larger hospitals can be sustainably supplied on a daily basis by single batches of [18F]PSMA-1007 obtained by the novel one-step radiofluorination procedure transferable to commercially available radiosynthesisers.
3.3. Quality Control
3.3.1. Acceptance Criteria
Acceptance criteria were chosen in compliance with the general texts and monographs of the current European Pharmacopoeia and are summarised in Table 1
. Most of the QC methods are standard procedures for skilled personnel and, thus, do not need to be discussed here. However, special emphasis should be given to the chemical purity of the product. Limits were also chosen comparable to existing monographs for fluorine-18-labelled radiopharmaceuticals and are 0.1 mg/Vmax
for PSMA-1007 (19
F-carrier), not more than 0.1 mg/Vmax
for a single unknown impurity assuming the same extinction coefficient like PSMA-1007, the sum of all unknown impurities including PSMA-1007 not more than 0.5 mg/Vmax
and a disregard limit of 0.03 mg/Vmax
for any unknown impurity detected in analytical HPLC. We recommend a minimal radiochemical purity of 95% of the total activity based on TLC for free fluoride-18 and HPLC for any other radiochemical impurity for the release of [18
3.3.2. Discussion of Quality Control
With respect to the one-step procedure with cartridge separation, a tolerable amount of side-products will occur in the final formulation. Using the limits as recommended, all produced batches from the direct radiolabeling method fulfill the release criteria.
In all productions (one- and two-step), the amount of carrier PSMA-1007 was in the range of 1–10 µg/mL. For research purposes, we calibrated in the range of 0.5–20 µg/mL; however, considering the chosen Vmax of 10 mL/patient, a calibration curve covering a 3–10-µg/mL range is sufficient with respect to the limit of non-radioactive PSMA-1007 content and the disregard limit. The standard wavelength of 254 nm using the aromatic absorption and/or 220 nm should be applied. Since measurements using the standard wavelengths proved to be sufficient, no additional UV-spectrum for the determination of absorption maxima was recorded.
A test for osmolality using freeze point reduction could not be applied due to the ethanol content of the final formulation. Anyhow, intravenous injectability of the final injection solution is guaranteed by the application of 0.9% saline as the main component (70%) in the chromatographic separation and final dilution (1:3) with isotonic PBS, resulting in a final ethanol concentration of 7.5% v/v.
In the case of the two-step procedure, additional tests for tBuOH and DIPEA by GC were applied. In all tests, contents were below LOQ.
It was shown that TBA can be precisely quantified using the TLC spot test presented above (Figure 3
). This represents a quick and low-cost method for analysis of TBA as compared to the liquid chromatography currently described in Ph. Eur. monographs. The TLC test is currently under evaluation and hopefully will be published soon in the Ph. Eur. as an alternative to the HPLC test method.