Production of [¹⁸F]DPA-714, [¹⁸F]fallypride and [¹⁸F]LBT-999 using iMiDEV, a fully automated microfluidic platform: towards clinical radiopharmaceutical production

Abstract

Background

Positron emission tomography is widely used to study biological processes without disrupting normal physiological functions. Traditional radiotracer synthesis and industrial market is focused on producing large batches of ¹⁸F-labelled tracers, especially [¹⁸F]FDG. Accessibility to smaller quantity of diverse radiopharmaceuticals is a key to enable a more personalised approach in nuclear medicine. A novel microfluidic module, iMiDEV™, has earlier been shown to be a versatile labelling platform as it has been used in the production of Na[¹⁸F]F and various ¹¹C- and ⁶⁸ Ga-labelled tracers. In the current study our aim was to utilise iMiDEV™ in the synthesis of fluorine-18-labelled radiotracers, specifically [¹⁸F]DPA-714, [¹⁸F]LBT-999 and [¹⁸F]fallypride.

Results

[¹⁸F]DPA-714, [¹⁸F]LBT-999 and [¹⁸F]fallypride have been produced in up to 24%, 12% and 11% radiochemical yield, respectively, using the microfluidics based iMiDEV™ labelling platform. Activity yields at the end of synthesis were 3.6 GBq, 2.1 GBq and 2.3 GBq, respectively. All individual synthesis steps were studied for efficient activity transfer and labelling and the optimised synthesis sequence was fully automated.

Conclusion

In this paper, we have demonstrated fully automated production of different ¹⁸F-tracers of clinical relevance with moderate to good yields using microfluidic iMiDEV™ platform. Our work is a step towards more personalised, dose-on-demand manufacturing of PET radiopharmaceuticals.

Introduction

Positron emission tomography (PET) is an important diagnostic tool offering profound insights into various diseases' molecular and physiological mechanisms. Its non-invasive nature allows for the interrogation of biological processes without disrupting normal physiological functions. PET holds immense potential in disease diagnosis and staging, therapy selection, treatment monitoring, and drug development, both in research and clinical settings. All the development of PET also leads to increasing demand for new methods in radiochemistry (Rong et al. 2023).

Currently, PET radiotracer synthesis predominantly occurs in centralised nuclear medicine facilities, posing challenges for clinics requiring immediate access to radiotracers, particularly those utilising short-lived radionuclides like carbon-11 (¹¹C, T_½: 20.3 min) or fluorine-18 (¹⁸F, T_½: 109.8 min). To address these limitations, innovative microfluidic approaches have emerged as promising solutions, offering localised on-demand production of radiotracers tailored to individual patients (Pascali et al. 2013; Mc Veigh and Bellan 2024). Crucial to the success of microfluidic techniques is the development of fully automated synthesis protocols using microfluidic platforms. Development of fully automated and robust labelling systems makes the radiopharmaceutical production accessible also for “non expert” PET centres regardless the availability of highly trained personnel.

While fluorine-18-based radiopharmaceuticals provide valuable diagnostic information across various medical domains, traditional synthesis methods can be still found cumbersome and cost-inefficient. Technological advancements seek to overcome these challenges by integrating microfluidic techniques into radiosynthesisers. Recent studies have showcased the potential of microfluidic modules, such as iMiDEV™, to achieve comparable radiochemical yields using significantly reduced precursor amounts (Mallapura et al. 2022, 2023). Utility of iMiDEV™ platform has been originally shown with one step production of Na[¹⁸F]F⁶, and [⁶⁸Ga]Ga-citrate (Ovdiichuk et al. 2022), and in the production of more complex radiotracers such as [⁶⁸Ga]Ga-PSMA-11 (Ovdiichuk et al. 2023), [⁶⁸Ga]Ga-FAPI-46 and [⁶⁸Ga]Ga-DOTA-TOC (Mallapura et al. 2023), and ¹¹C-labelled tracers [¹¹C]flumazenil, L-[¹¹C]deprenyl, L-[¹¹C]methionine and [¹¹C]choline using either [¹¹C]CH₃I or [¹¹C]CH₃OTf as ¹¹C-labelling precursor (Mallapura et al. 2022, 2024).

In-house preparation of radiopharmaceuticals is a highly regulated field and production must be done in accordance with increasing quality and good manufacturing practice (GMP) requirements (Gillings et al. 2021, 2022; Hendrikse et al. 2022; Moya et al. 2024). Previously, the use of microfluidic technologies in PET tracer production has been associated with certain risks, including engineering challenges and the integration of all the synthesis steps into one device, or time-consuming cleaning cycles of permanent devices and parts to which radiopharmacy prefers single use cassettes today (Lebedev et al. 2013; Rensch et al. 2013). To overcome these issues, in this study, we use a disposable microfluidic cassette based approach with the compact iMiDEV™ microfluidic platform. Automated synthesis modules, based on sterile and single use disposable cassettes and reagents, are the most preferred configuration in a clinical routine environment, despite their higher cost, as they reduce the risk of cross contamination, reduce the radiation exposure to the operator and improve reproducibility. It should also be noted that, where relevant, both hardware and software should be in compliance with GMP and good automated manufacturing practice 5 (GAMP5).

In this study, our aim is to utilise an innovative microfluidic cassette-based iMiDEV™ module for the synthesis of fluorine-18-labelled radiotracers, specifically [¹⁸F]DPA-714, [¹⁸F]LBT-999 and [¹⁸F]fallypride, to evaluate its suitability for single-dose or dose-on-demand (DOD) (Pascali et al. 2016; Arima et al. 2013; Knapp et al. 2020; Matesic et al. 2017) production. Our focus lies in implementing and automating various radiosyntheses inside the microfluidic cassette, with the goal of assessing synthesis feasibility and efficiency. Through this investigation, we aspire to contribute to the advancement of microfluidic radiochemistry synthesis platforms, enabling enhanced accessibility and efficiency in PET radiotracer production for clinical application.

Materials and methods

Chemicals and consumables

N,N-Diethyl-2-(2-(4-(2-tosyloxy-1-ethoxy)phenyl)-5,7-dimethylpyrazolo[1,5-a]pyrimidin-3-yl)acetamide (TsO-DPA-714, > 98%), N,N-diethyl-2-(2-(4-(2-fluoro-1-ethoxy)phenyl)- 5,7-dimethylpyrazolo[1,5-a]pyrimidin-3-yl)acetamide (DPA-714), were purchased from Pharmasynth (Estonia). (S)-2,3-Dimethoxy-5-[3-[[(4-methylphenyl)-sulfonyl]oxy]propyl]-N-[[1-(2-propenyl)-2-pyrrolidinyl]methyl]benzamide (TsO-fallypride, > 90%), (S)-5-(3-fluoropropyl)-2,3-dimethoxy-N-[[(2S)-1-(2-propenyl)-2-pyrrolidinyl]methyl]-benzamide (fallypride) were purchased from ABX (Germany). LBT-999 (8-((E)-4-fluoro-but-2-enyl)-3-beta-p-tolyl-8-aza-bicyclo[3.2.1]octane-2-beta-carboxylicacid methyl ester) and its chloro-precursor (8-((E)-4-chloro-but-2-enyl)-3-beta-p-tolyl-8-aza-bicyclo[3.2.1]octane-2-beta-carboxylicacid methyl ester, > 95%) were obtained from ERAS Labo (France).

Anhydrous acetonitrile (˃99.8%), K₂CO₃ (99.99%), Kryptofix^® K2.2.2 (4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo-[8.8.8]-hexacosane, K₂₂₂), ammonium acetate, trifluoroacetic acid (TFA) and dimethyl sulfoxide (DMSO) were purchased from Sigma Aldrich (France). Acetonitrile (HPLC grade) was purchased from CARLO ERBA Reagents (France). 0.9% Sodium chloride (NaCl) solution for injections was obtained from B Braun (France). ¹⁸O-enriched H₂O (> 97%-enriched) was purchased from CortecNet (France).

K₂₂₂/K₂CO₃/acetonitrile solutions containing 1–5% H₂O were prepared as previously described in Ovdiichuk et al. (2024).

The microfluidic cassettes and the iMiDEV™ radiosynthesiser were supplied by PMB-Alcen (Peynier, France). Solid phase extraction beads were filled in the microfluidic cassette by the supplier. 1.1 mL glass vials (12*32 mm) were obtained from Waters (MA, United States). 4 mL and 15 mL crimp neck glass vials were acquired from SCHOTT (Germany). 11 mm and 13 mm aluminium seals with septa (polytetrafluoroethylene, PTFE/butyl/PTFE) and 20 mm cap stopper (butyl) were obtained from Fisher Scientific (Germany). 13 mm aluminium seals with septa (PTFE/rubber) were supplied by Thermo Scientific (Germany).

Isotope calibrator (Capintec, Mirion Technologies) calibrated for fluorine-18 was used for radioactivity measurements.

iMiDEV™ radiosynthesiser and the microfluidic cassette

The characteristics of iMiDEV™ microfluidic platform and the disposable microfluidic cassettes (Figure S1) have been described earlier (Ovdiichuk et al. 2021). Briefly, the batch-type cassettes consist of microfluidic channels and 4 independent chambers R1-R4, simplified presentation in Fig. 1. Positive intrinsic negative (PIN) diode radiation detectors are located on top of each chamber. Chambers R1, R2 and R4 were used in this study, R1 (50 µL—approximately 25 mg beads) was dedicated for fluorine-18 trapping and activation, R2 (286 µL) for the labelling reaction and R4 (200 µL—approximately 100 mg beads) for the solid phase extraction (SPE) based formulation of the product. As presented in our earlier work on microelution (Ovdiichuk et al. 2024), R1 was filled with QMA-CO₃ anion-exchange beads (supplied by Waters or Eichrom). R4 was filled with C18 beads (Waters). No preconditioning of beads was performed. As internal quality control of the SPE beads density in R1 and R4, the supplier measured the back pressure in the chamber by pushing compressed air through the beads at 1000 mbar. The back pressure value represents a measure of the flow through beads and gives a value that allows comparison between cassettes and therefore repeatable elutions.

Fig. 1

Simplified presentation of iMiDEV™ microfluidic cassette used in nucleophilic ¹⁸F-labelling

The module has integrated HPLC purification system which consists of Azura P 6.1L pump and Azura 2.1S UV detector (Knauer, Germany). Purification conditions are detailed in Table 1.

Table 1 HPLC purification conditions

General ¹⁸F trapping and elution techniques

[¹⁸F]fluoride trapping on the QMA beads in R1 was performed using two different approaches, (1) direct pathway or (2) reversed pathway (Fig. 2a). In the direct pathway, trapping, washing and elution are performed on the same direction. In the reversed pathway, trapping and washing are performed from the opposite end of the R1 compared to the elution.

Fig. 2

Simplified presentation of (a) trapping and (b) elution techniques using the microfluidic cassette

Elution was performed using three different approaches (Fig. 2b):

• 1) Subsequent flow from eluent vial towards precursor vial, and subsequent transfer of resulted mixture to reaction chamber R2.

• 2) Simultaneous flow from eluent and precursor vials towards reaction chamber R2.

• 3) Pulse-flow where flow is directed from eluent vial and precursor vial in short consecutive pulses towards reaction chamber R2. Pulse flow parameters, individual pulse length and pressure, and total amount of pulses, were optimised for optimal filling of R2. Optimisation was done by following visually the filling of R2 using different parameter combinations.

Cassette preparation

Cassette preparation consists of filling the reagent vials A-I. Reagents used in direct trapping/elution approach are described in supporting information Table S1. Reagents used in reversed approach for optimised radiosynthesis are described in Table 2.

Table 2 Preparation of vials for [¹⁸F]DPA-714, [¹⁸F]LBT-999 and [¹⁸F]fallypride synthesis

General [ ¹⁸ F]fluorination protocol using iMiDEV™

During the radiosynthesis, all radioactive waste in each step was collected separately and measured with isotope calibrator to determine the losses in each step.

Aqueous [¹⁸F]fluoride was produced via the ¹⁸O(p,n)¹⁸F nuclear reaction from ¹⁸O-enriched H₂O on a Cyclone-18/9 Cyclotron (IBA). Full synthesis sequence information is presented in Supporting information Table S2. [¹⁸F]fluoride trapping on the non-preconditioned QMA beads in R1 was performed either by direct pathway or reversed pathway. Briefly, aq [¹⁸F]fluoride (2 mL) was passed through the QMA beads followed by drying with He flow. Beads were washed with MeCN, followed with He flow. [¹⁸F]Fluoride was eluted with K₂₂₂/K₂CO₃/2% water/MeCN eluent solution towards R2 using subsequent flow, simultaneous flow, or pulse-flow. For manual radiolabelling study, eluent was collected outside the hotcell and used for labelling in a macro-sized vial as described earlier in Ovdiichuk et al. (2024)

Using iMiDEV, radiolabelling was conducted at 90 °C or 120 °C (temperature instruction was set to 120 °C or 150 °C, respectively, as earlier study has shown 30 °C difference between the measured and set values for the temperature inside chamber R2 (Ovdiichuk et al. 2023)). Counter pressure (2000 mbar), meaning pressurising the cassette without opening the valves towards waste, was directed towards chamber R2 from vials C (microfluidic valve, MFV 10 open) and E (MFVs 15, 16, 18, 19 open) to prevent the leak of pressurised reaction solution through the microfluidic valves during the heating. After 10 or 20 min of heating, the temperature was set to 20 °C. Counter pressure was released, and the microfluidic channels depressurised. When reaction chamber reached < 40 °C, the solution was pushed to HPLC inlet loop (1 mL loop for purification with analytical column or 2 mL with semipreparative column) with HPLC dilution solution. In the initial tests to study the radiolabelling efficiency inside the microfluidic cartridge, reaction mixture was collected after the reaction at HPLC injection waste vial.

HPLC purification was initiated automatically. After liquid was detected by an optical sensor located before the HPLC inlet loop, the reaction mixture was directed into the HPLC inlet loop. Purification was started automatically when the optical sensor detected gas after the liquid transfer was completed. The delay for liquid to reach the loop after passing the optical sensor was studied separately before performing the full synthesis. HPLC collection was followed with UV and radioactivity detectors placed at the column outlet. Collection of the product fraction was started and stopped automatically by the software when radioactivity was detected at the column outlet. Collection loop (2 mL loop for analytical column or 5 mL for semipreparative column) was emptied towards HPLC collection vial (vial H) containing the dilution solution to lower the organic solvent concentration. This solution was pushed through R4 and the tracer was trapped in C18 beads followed by drying with He flow. Tracer was eluted from C18 with EtOH (vial F) towards vial I. Product was collected by directing flow from vial I through the formulation chamber to collection vial followed by stream of He. Cleaning of the device after synthesis is described in supporting information Table S3.

Trapping efficiency is calculated by measuring the collected waste during the trapping process with an isotope calibrator and comparing it to the radioactivity sent to the device (isotope vial was measured before and after sending the activity to the device). Activity loss during R1 washing step is calculated by measuring the collected waste during the washing step. Elution efficiencies (EE) from R1 and R4, and transfer efficiency from R2 to HPLC are calculated using the iMiDEV™ radioactivity detector data (example given in the supporting information Figure S2). Total radioactivity loss during a synthesis is calculated by comparing the collected waste to initial radioactivity. Radiochemical yields (RCYs) are calculated from isolated product. Molar activity (A_m) values are decay-corrected to start of synthesis (SOS).

Quality control

Aliquot of the formulated product solution was used for the quality control tests. Radiochemical purity (RCP) was analysed by radio-TLC or radio-HPLC, and identity by radio-HPLC, see supporting information Figures S3–S8. Radiochemical conversion (RCC) is determined by radio-TLC from reaction solution or by semipreparative HPLC. Radionuclidic purity, radionuclidic identity, pH, chemical purity (K₂₂₂) and residual solvents were analysed with QC1 (Trasis, Belgium), an integrated quality control system. All tests were carried out according to methods meeting the expectations of the European Pharmacopeia.

Chromatographic methods

Radio-thin layer chromatography (Radio-TLC) was performed with a Mini-Scan and Flow-Count radioactive detection system (Bioscan) and Chromeleon software (Thermo Scientific). An aliquot of the reaction mixture (5–10 µL) was applicated on silica gel coated aluminium plate and eluted with EtOAc ([¹⁸F]DPA-714, [¹⁸F]LBT-999) or 9:1 MeCN/H₂O ([¹⁸F]fallypride).

Analytical high pressure liquid chromatography (HPLC) was performed on the following system: Waters Alliance 2690 equipped with a UV spectrophotometer (Photodiode Array Detector, Waters 996 (Waters)) and a Berthold LB509 radioactivity detector; column: analytical Symmetry-M^® C-18, 50 × 4.6 mm, 5 µm (Waters); solvent A: H₂O containing Low-UV PIC^® B7 reagent (20 mL for 1000 mL), solvent B: H₂O/MeCN: 30:70 (v/v) containing Low-UV PIC^® B7 reagent (20 mL for 1000 mL); UV detection at λ = 254 nm/220 nm. Detailed analytical HPLC and TLC conditions are in Table 3.

Table 3 Conditions for analytical HPLC and TLC

Results

[ ¹⁸ F]Fluoride trapping and activation optimisation on microfluidic cassette

Different trapping (direct or reversed) and elution techniques (subsequent, simultaneous or pulse flow) inside the microfluidic cartridge were analysed and results are collected in supporting information Table S4. During this preliminary study, trapping and elution process was optimised regarding to eluent vial size, used eluent volume and eluent water content as specified in Table S4. Direct and reversed trapping and elution approaches showed excellent trapping efficiencies (> 95%) and high elution efficiencies (> 85%) with all different elution techniques. However, using direct approach, 67.3 ± 18.1% (n = 19, subsequent elution technique) or 88.7 ± 6.4% (n = 2, simultaneous elution technique) of the initial activity was lost inside the microfluidic cartridge during the synthesis process. With reversed approach, 54.4 ± 7.6% (n = 2) or 61.6 ± 21.0% (n = 2) of the initial activity was lost, respectively. Changing elution method to pulse flow elution with reversed trapping, most of the activity could be recovered from the cassette. Radioactivity loss was 35.8 ± 17.7% (n = 19) with Waters QMA beads and 21.8 ± 17.2% (n = 10) with Eichrom QMA beads in chamber R1. After full synthesis optimisation, including pulse flow optimisation and R1 rinsing after elution, activity loss during the synthesis, when using Eichrom QMA beads in R1, has dropped to 2.4 ± 0.3% (n = 3).

In the preliminary study, R1 back pressure variation was observed to affect the elution results inside the microfluidic cassette. The effect of the QMA beads size and substrate type on the R1 back pressure was studied and the results are collected in supporting information Table S5.

Radiolabelling and purification optimisation

[¹⁸F]DPA-714 radiolabelling conditions were optimised in manual radiolabelling study and results are collected in supporting information Table S6. Up to 76% radiochemical conversion (RCC, based on radioTLC analysis) was obtained with 2% H₂O containing eluent after 10 min at 120 °C using 1:1 MeCN + 2% H₂O/DMSO mixture as reaction solvent and 20 µmol/mL precursor solution concentration. Decreasing the reaction temperature to 95 °C led to only 27% conversion and changing the reaction solvent from MeCN + 2% H₂O/DMSO to only acetonitrile, dropped the RCC to 7%. The optimal reaction conditions of [¹⁸F]DPA-714 (10 min at 120 °C) were used for [¹⁸F]LBT-999 and [¹⁸F]fallypride without separate optimisation study.

With iMiDEV™, [¹⁸F]DPA-714 was produced with 62 ± 7% (n = 4) RCC (based on semiprep HPLC) using the optimal conditions (20 min at 120 °C, 1:1 MeCN + 2% H₂O/DMSO and 20 µmol/mL precursor). In the preliminary labelling studies with iMiDEV™, 20–50% of radioactivity was lost during the labelling reaction when chamber was heated at 120 °C (T_set) for up to 10 min (Fig. 3a). This was hypotesised to be caused by leak of pressurised reaction mixture through the microfluidic valves. To study this, cassette was pressurised around the reaction chamber R2 during the heating step, and no leak of radioactivity was observed even when heating R2 at 150 °C (T_set) for 20 min (Fig. 3b).

Fig. 3

Radioactivity detector data during the radiolabelling reaction in R2 without (a) and with (b) counter pressure. One radioactivity unit nA corresponds to approximately 1000 MBq

[¹⁸F]DPA-714 reaction solution was purified in analytical conditions or semipreparative conditions. Using analytical conditions, RCP of the final product was ≥ 99.9%, RCY (decay-corrected) 18 ± 11% and A_m 22 ± 11 GBq/µmol (n = 6). Using semi-preparative purification conditions, RCP was ≥ 99.9%, RCY 12 ± 5% and the A_m 82 ± 18 GBq/µmol (n = 6). Use of analytical column for [¹⁸F]DPA-714 purification was not sufficient to separate chemical impurities from the product fraction which can be seen from the lower molar activity when using analytical column for HPLC purification. Formulation step in C18 beads in R4 has not been a limiting factor and thus did not need further optimisation.

[¹⁸F]LBT-999 and [¹⁸F]fallypride were produced in iMiDEV™ microfluidic cassette with 24 ± 5% (n = 4) and 26 ± 2% (n = 2) RCC and both tracers were purified only using semipreparative conditions. Notable difference to [¹⁸F]DPA-714 is that both tracers needed ascorbic acid as stabiliser to reach sufficient radiochemical purity (> 95%).

Full radiosynthesis in the microfluidic cassette and quality control

Radioactivity trends from a [¹⁸F]DPA-714 synthesis are presented in Fig. 4. Results from optimised [¹⁸F]DPA-714, [¹⁸F]LBT-999 and [¹⁸F]fallypride production are collected in Table 4.

Fig. 4

Radioactivity data from detectors on chambers R1, R2, R3 and R4 during a [¹⁸F]DPA-714 synthesis

Table 4 Results of [¹⁸F]DPA-714, [¹⁸F]LBT-999 and [¹⁸F]fallypride production

Discussion

[¹⁸F]Fluoride trapping and activation, radiolabelling, HPLC purification and product formulation were successfully done using the disposable microfluidic cassette. All individual steps, including [¹⁸F]fluoride trapping method, elution technique and subsequent reaction chamber filling, reaction conditions, purification method and formulation, were studied for efficient activity transfer and labelling.

[ ¹⁸ F]Fluoride trapping and activation optimisation on microfluidic cassette

In our previous study on [¹⁸F]fluoride activation in hydrous conditions, we described efficient elution from QMA resin when using K₂₂₂/K₂CO₃/MeCN eluent containing 1–5% water in iMiDEV™ microfluidic cartridge (Ovdiichuk et al. 2024). In the current study, we studied in further detail the trapping method, implemented the previously studied elution conditions to radiotracer synthesis done solely in the microfluidic cartridge, and observed following factors to affect the elution efficiency (EE): (1) trapping/elution direction: direct or reversed, (2) elution technique: subsequent, simultaneous or pulse flow, (3) beads packing in chamber R1, and (4) chamber R2 filling.

Optimisation study showed that the best labelling results and the highest radioactivity transfer to R2 are reached when using reversed trapping combined with pulse flow elution. The trapping and elution process was optimised over time regarding to eluent vial size, used eluent volume and eluent water content, thus the results between different trapping and elution methods are not directly comparable. Notable difference of subsequent and simultaneous elution method to pulse flow elution was the used eluent vial size, used eluent volume and eluent water content. Subsequent and simultaneous elution tests were done using 4 mL vial with higher eluent volume and 3% H₂O containing eluent compared to pulse flow elution tests, which were done with 1.1 mL vial, with less volume and eluent containing only 2% H₂O. As observed earlier with iMiDEV™ microfluidic cartridge when producing ¹¹C-tracers by Mallapura et al. (2022) precursor volume needed was dependant of the vial used. Larger volume (250 µL vs 100 µL) was needed to have reproducible reaction when using large vials compared to small vials (4 mL vs 0.3 mL) due to the large variation in dead volume with 4 mL vials (Mallapura et al. 2022). Our results support this, with 1.1 mL vials, the elution step was reproducible with only 150–200 µL of eluent. This is also in accordance with our previous results which showed good EE with ≥ 150 µL eluent when 1.1 mL vial was used (Ovdiichuk et al. 2024).

Another variant affecting the elution efficiency inside the microfluidic cassette was the QMA beads density in R1. In our previous study we observed the Waters beads to be sligthly superior compared to the Eichrom QMA beads in terms of elution efficiency (Ovdiichuk et al. 2024), however, when conducting the entire radiosynthesis inside the microfluidic cartridge, Eichrom QMA beads have showed better EE and higher reproducibility. We found this to be related to the back pressure in the chamber R1 (780–810 mbar with Eichrom beads vs. 800–920 mbar with Waters beads). Difference in the back pressure can be partially explained by the size range of the QMA beads (Waters 37—55 µm, Eichrom beads 55 µm, stated value) and also beads substrate type (Waters irregular silica 2024; Eichrom spherical silica 2024). Back pressure variation has been earlier observed to affect the radiochemical yield and total synthesis time with other resin types such as HLB and CM, (Mallapura et al. 2024) and C18 (Mallapura et al. 2022). Mallapura et al. hypothesised that unifying the beads size would ensure consistent back pressure and repeatable results (Mallapura et al. 2023). To further study the effect of different size beads on the elution, two different batches of microfluidic cassettes filled with sieved Eichrom QMA beads (1) 50 < × < 63 µm, 760–770 mbar back pressure and 2) 63 < × < 80 µm, 730–740 mbar) were tested. Trapping and elution efficiencies were similar in both sieving batches but the pulse flow elution parameters needed optimisation for each batch to result in optimal filling of reaction chamber R2 and to have reproducible synthesis results. To conclude, after strict control of the resin inside R1, we have significantly increased the reproducibility of our ¹⁸F-labelling process. This improvement is valuable when considering implementing iMiDEV™ platform for automated clinical radiopharmaceutical production.

Final limiting factor for the elution efficiency was filling of reaction chamber R2 and mixing with precursor. Use of low viscosity organic solvents, like acetonitrile is challenging due to facile formation of gas bubbles inside R2 during the elution and mixing process. These gas bubbles can significantly limit the amount of liquid that can be transferred to R2, and thus the amount of eluent pushed through QMA beads. This issue is particulalry challenging for devices working at the microfluidic scale but absent in batch technology synthesisers. Carefully controlled pulse flow elution was found to be most efficient way to prevent the formation of gas bubbles and optimise the filling of R2.

Radiolabelling optimisation and full radiosynthesis inside the microfluidic cassette

Synthesis sequence was optimised using [¹⁸F]DPA-714 as a model molecule. After optimisation of each individual step for [¹⁸F]DPA-714, including [¹⁸F]fluoride activation, transfer of ¹⁸F and precursor to the reactor, radiolabelling reaction, transfer of reaction solution to HPLC, collection and dilution of product fraction and formulation, the same sequence was used for production of two other radiotracers [¹⁸F]LBT-999 and [¹⁸F]fallypride. Only modifications to the sequence were done on the purification conditions (standard conditions for each tracer Javed et al. 2014; Dollé et al. 2007) and sodium ascorbate was added to each step after the labelling reaction as a stabiliser against radiolytic dissociation. We have successfully demonstrated the utility of the microfluidic labelling platform iMiDEV™ in the production of popular ¹⁸F-labelled radiotracers via nucleophilic ¹⁸F-labelling.

[¹⁸F]DPA-714 and [¹⁸F]LBT-999 have been earlier produced only using conventional radiosynthesisers. [¹⁸F]DPA-714 has been produced with up to 50% RCY (decay corrected) within 50 min synthesis time (Kuhnast et al. 2012) and recent optimisations have resulted in up to 70% RCYs allowing multi-centre studies (Cybulska et al. 2021). [¹⁸F]LBT-999 has been earlier produced with 16% RCY (non-decay corrected) in 90 min (Dollé et al. 2007) and recently with up to 35% RCY (decay corrected) after 50 min synthesis (Vala et al. 2020). Even if we observe lower RCYs in our study, the produced activities are largely sufficient for a DOD approach and we were able to lower the precursor amount making the synthesis more cost-efficient.

Microfluidic labelling approaches have been studied by several groups but none of these applications has spread to clinical radiopharmaceutical production. [¹⁸F]Fallypride has earlier been produced with NanoTek microreactor apparatus with up to 88% RCC (reaction at 170 °C), however this approach did not include the [¹⁸F]fluoride activation, which was done in traditional manner using azeotropic distillation (Matesic et al. 2017; Lu et al. 2009). An example of fully microfluidic approach for radiotracer production utilised a microcartridge for [¹⁸F]fluoride concentration and microdroplet reactor to produce [¹⁸F]fallypride in 25% RCY within 35 min synthesis time (Wang et al. 2020). In earlier studies, it has also been noticed that lowering the starting activity (< 1 GBq) significantly increases the achievable RCY up to 70% (Javed et al. 2014; Wang et al. 2020). Using conventional radiosynthesisers, [¹⁸F]fallypride has been produced with up to 55% RCY (non-decay corrected) in 58 min (Huhtala et al. 2019). Even if in our approach [¹⁸F]fallypride has been produced with lower RCY, we have demonstrated fully automatic microfluidics based radiolabelling process suitable for clinical radiopharmaceutical production. Most of the other microfluidic approaches have shown potential for automation but not yet to be mature enough to be used in clinical production. Due to the simplicity of the cassette removal and device cleaning, iMiDEV™ platform allows production of different tracers one after another without cross-contamination risk.

Conclusion

In conclusion, we have demonstrated fully automated production of different ¹⁸F-tracers of clinical relevance with moderate to good yields using microfluidic iMiDEV™ platform. Produced radiotracers meet the Ph. Eur. QC requirements. The developed synthesis process can be potentially applied to variety of different tracers requiring one-step labelling process with minimal process optimisation. Our work is a step towards more personalised DOD manufacturing of radiopharmaceuticals. This proof-of-concept study with nucleophilic ¹⁸F-labelling is also one more example of the versatility of iMiDEV™ microfluidic cassette and labelling platform which has earlier been used in the production of Na[¹⁸F]F and various ¹¹C- and ⁶⁸Ga-labelled tracers.