High-yield, standardised automated radiosynthesis process for routine clinical insulinoma PET with [⁶⁸Ga]Ga-NODAGA-exendin-4

Abstract

Background

Insulinoma is a rare pancreatic neuroendocrine tumour characterised by inappropriate insulin secretion and recurrent hypoglycaemia. Owing to their small size and intrapancreatic localisation, insulinomas are frequently difficult to detect using conventional anatomical imaging techniques. Functional imaging based on positron emission tomography has therefore gained increasing interest, particularly through targeting of the glucagon-like peptide-1 receptor, which is highly overexpressed in most insulinomas. Among available tracers, gallium-68-labelled NODAGA-exendin-4 has demonstrated excellent diagnostic performance. However, its routine clinical implementation remains limited by the absence of a harmonised, robust and transferable radiosynthesis protocol compatible with hospital radiopharmacy practice. The aim of this work was to optimise and harmonise the automated preparation of gallium-68-labelled NODAGA-exendin-4 by systematically evaluating critical synthesis parameters and validating the optimised process across multiple commercially available gallium-68 generators. Using an automated synthesis module, key variables including precursor amount, labelling temperature and duration, formulation additives, workflow sequence and quality control conditions were investigated. The optimised protocol was subsequently validated using generators from three different manufacturers.

Results

Optimisation studies demonstrated that formulation-related parameters, particularly the post-labelling addition of polysorbate 20, resulted in improved radiochemical conversion and reduced residual activity within the synthesis cassette. An optimal precursor amount of 20 µg was identified as a compromise between radiochemical yield and clinical injectability constraints. Radiolabelling at 95 °C for 12 min ensured high conversion while reducing overall synthesis time. Validation runs showed excellent reproducibility, with non-decay-corrected yields ranging from 60 to 73% and decay-corrected yields reaching up to 93%, independent of generator type. Radiochemical purity consistently exceeded 95%, and all quality control parameters complied with established specifications.

Conclusions

This study establishes a simplified, efficient and generator-independent automated synthesis of gallium-68-labelled NODAGA-exendin-4. By addressing key translational and regulatory constraints, the proposed protocol provides a practical foundation for the routine clinical implementation of glucagon-like peptide-1 receptor imaging in patients with suspected insulinoma.

Background

Insulinoma is a rare tumour of the β-cells of the pancreas that causes excessive insulin production and is clinically characterised by recurrent hypoglycaemic episodes. It affects between 1 in 250,000 and 1 in 1,000,000 people in the general population (Dr Run 2021). Surgical resection represents the gold standard treatment and offers the only curative option for most patients (McAuley et al. 2005). Conventional imaging modalities, including computed tomography (CT), magnetic resonance imaging (MRI), and endoscopic ultrasound, are routinely used for tumour localisation (Christ et al. 2009). However, due to the typically small size of insulinomas and their intrapancreatic location (Jansen et al. 2019), these techniques show limited sensitivity and may fail to detect the lesion in a substantial proportion of cases.

In this context, nuclear medicine has emerged as a promising complementary approach, providing functional imaging based on biological processes rather than purely anatomical features. Positron emission tomography (PET), in particular, is a tool for oncologic imaging, enabling the visualisation of biological activity and improving tumour detection when conventional imaging methods are inconclusive.

However, despite these technical advances, the detection of insulinomas remains difficult with conventional radiopharmaceuticals such as [¹¹¹In]In-pentetreotide or [⁶⁸Ga]Ga-edotreotide (Christ et al. 2009; Wild et al. 2006). This limitation has progressively shifted research toward the identification of molecular targets that are overexpressed in insulinoma cells, with the objective of improving lesion detectability and achieving higher tumour-to-background contrast. In this context, the glucagon-like peptide-1 receptor (GLP-1R) has emerged as a particularly relevant target, as it is physiologically expressed in pancreatic β-cells and markedly overexpressed in most insulinomas, while showing limited expression in surrounding healthy tissues. This biological profile makes GLP-1R an attractive target for the development of dedicated radiopharmaceuticals (Wild et al. 2006; Bertherat et al. 2003).

The growing interest in GLP-1R targeting originates from advances in incretin biology and peptide therapeutics. Among the peptide families with hypoglycaemic properties, the exendin family was characterised in the early 1990s, with exendin-4 first isolated in 1992 from the venom of the Gila monster by John Eng (Sowa-Staszczak et al. 2016). Exendin-4 is a potent GLP-1R agonist and, unlike endogenous GLP-1, it is resistant to degradation by dipeptidyl-peptidase-4, resulting in a prolonged biological half-life and sustained receptor binding affinity (Selvaraju 2014).

Initially developed as an antidiabetic agent (exenatide), exendin-4 was subsequently explored as a GLP-1R-targeting tracer due to the strong receptor expression observed in insulinoma tissue. The first GLP-1R-targeted radiopharmaceuticals were developed for scintigraphy using indium-111-labelled exendin-4 with different chelators, notably [¹¹¹In]In-DTPA-exendin-4 and later [¹¹¹In]In-DOTA-exendin-4 (Christ et al. 2009; Wild et al. 2006), as well as [^99mTc]Tc-labelled GLP-1 analogues (Sowa-Staszczak et al. 2016). Building on these encouraging results, subsequent developments progressively shifted toward PET, with the aim of benefiting from higher sensitivity, improved spatial resolution, and quantitative capabilities (Brom et al. 2010).

Within this context, several gallium-68-labelled exendin analogues were investigated using different chelators such as NOTA, NODAGA, DO3A, and deferoxamine (Xie et al. 2023). Among them, [⁶⁸Ga]Ga-NODAGA-exendin-4 has demonstrated excellent in vivo stability and high diagnostic performance for insulinoma localisation in both preclinical and clinical studies, outperforming conventional imaging modalities in terms of lesion detection and tumour-to-background contrast (Spang et al. 2016).

Despite its promising clinical performance, the routine implementation of [⁶⁸Ga]Ga-NODAGA-exendin-4 remains challenging. Although several preparation methods for exendin-4-based radiopharmaceuticals have been reported in the literature, no harmonised production process has yet been established for routine clinical use. Published procedures describe a range of technical approaches, including generator elution with or without pre-purification, the use of different buffering systems such as 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) to facilitate gallium-68 complexation, and the addition of excipients such as polysorbate or ethylenediaminetetraacetic acid (EDTA) to limit peptide adsorption or metal-related impurities (Ragot et al. 2026; Menon et al. 2024; Migliari et al. 2022; Boss et al. 2020; Kaeppeli et al. 2019). As a result, existing protocols differ substantially in terms of workflow complexity, formulation strategy, and quality control requirements.

This methodological heterogeneity complicates the implementation of [⁶⁸Ga]Ga-NODAGA-exendin-4 in a clinical radiopharmacy setting, particularly in the absence of a dedicated monograph in the European Pharmacopoeia. To ensure patient safety and regulatory compliance, the development of an optimised preparation method is necessary and must be based on Good Preparation Practices and the relevant general mo nographs of the European Pharmacopoeia.

This work aims to harmonise and optimise the preparation of [⁶⁸Ga]Ga-NODAGA-exendin-4 through systematic variation and evaluation of critical synthesis parameters, with validation across multiple commercially available gallium-68 generators.

By harmonising these parameters through structured optimisation, the ultimate objective is to define a fully automated and reproducible radiosynthesis process characterised by maximised radiochemical purity and optimised synthesis yield, fully compatible with routine clinical radiopharmacy practice, and in compliance with current specifications (EDQM 2026a, b), or, where no European Pharmacopoeia monograph exists, with standards commonly applied in practice.

Methods

General information

To support the development of the radiosynthesis procedure, a review of previously reported methods for galllium-68 labelling of exendin-4 analogues was performed. As summarised in Table 1, the reported protocols show variability in terms of precursor concentration, buffer systems, heating conditions, purification steps, and the use of additives.

Table 1.

Galllium-68 labelling of exendin-4 analogues protocols available in the literature

Paper	Ragot et al. (2026)	Menon et al. (2024)	Migliari et al. (2022)	Boss et al. (2020)	Kirsi et al. (2014)
Radiosynthesis type	Automated	Automated	Automated	Automated	Not reported
Module	miniAllinOne (Trasis)	Modular-Lab standard (E&Z)	GRP synthesiser (Scintomics)	GRP synthesiser (Scintomics)	Not reported
68Ga pre-purification	Yes strong cation exchange (SCX)	Yes (SCX)	Yes (SCX)	Yes (SCX)	No
Exendin-4 amount	10 µg	50 µg	10 µg	10 µg	5 nM (± 23 µg)
Buffer	Sodium acetate 0.7 M	2,5 M HEPES	2,5 M HEPES	2,5 M HEPES	Sodium Acetate 0.2 M
Scavenger	Ascorbic acid	Ascorbic acid	Ascorbic acid	Ascorbic acid	No
Additives	EDTA/ Polysorbate 80 (after radiolabelling)	EDTA/ Polysorbate 80 (after radiolabelling)	EDTA/ Polysorbate 80 (after radiolabelling)	EDTA/ Polysorbate 80 (after radiolabelling)	No
Radiolabelling T°C and duration	95–100 °C 15 min	95 °C 15 min	100 °C 15 min	100 °C 15 min	95 °C 15 min
Final purification	Hydrophilic-lipophilic-balanced (HLB)	HLB	HLB	HLB	No
RCY* (%)	67	78.7	45	Not reported	Not reported
RCP** (%)	97	99.1	> 95	Not reported	Not reported

*RCY = decay-corrected radiochemical yield, ** RCP= radiochemical purity

The optimised method presented in this article is derived from an existing process and consumables developed at Trasis, which have been specifically adapted to exendin-4 (see Fig. 1). The original automated sequence was initially designed for the radiosynthesis of small molecule/peptide-based radiopharmaceuticals such as [⁶⁸Ga]Ga-PSMA-11, [⁶⁸Ga]Ga-DOTATOC, [⁶⁸Ga]Ga-DOTATATE and [⁶⁸Ga]Ga-DOTANOC (Li et al. 2025; Reverchona et al. 2020; Maus et al. 2020; Isal et al. 2018; Moussaron et al. 2021).

Fig. 1.

Chemical Structure of [⁶⁸Ga]Ga-NODAGA-exendin-4 (figure created using ChemDraw)

Optimised radiosynthesis of [⁶⁸Ga]Ga-NODAGA-exendin-4

All optimisations were carried out using the GeGant^® (ITM, Munich, Germany) generator available at the start of the project and were later transferred to the Galli RD^® (IRE-Elit, IRE, Fleurus, Belgium) and GalliaPharm^® (Eckert & Ziegler, E&Z, Berlin, Germany) generators. The initial synthesis was performed using 10 µg of precursor (Ragot et al. 2026; Migliari et al. 2022; Boss et al. 2020), an amount frequently reported in literature protocols due to limitations on the maximum injectable quantity in humans (Velikyan et al. 2017; Kraihammer et al. 2023, Tokgöz et al. 2023). Furthermore, the initial experiments were carried out by incubating the reaction mixture at 95 °C for 15 min (Menon et al. 2024), based on existing data as well as the conditions used in the original process for PSMA-11 and DOTA-X radiolabelling. During sequence refinement, higher precursor concentrations were tested, along with different radiolabelling times and temperatures to identify optimal conditions for maximum conversion while respecting the process constraints. Additional adjustments were also explored, such as (i) reversing the order of generator elution and precursor addition to the reactor, (ii) introducing adjuvants at varying concentrations into the saline pouch, and (iii) reducing the precursor dilution factor during the labelling step (see Table 2). In addition, the radio-thin layer chromatography (TLC) protocol also had to be adapted, as commonly used mobile phases and standard strips have been reported to provide unreliable or ambiguous separation of gallium-68-labelled exendin (Migliari et al. 2022; Kaeppeli et al. 2019; Brom et al. 2016), particularly with respect to free and colloidal gallium species (results non-disclosed).

Table 2.

Parameters evaluated during the optimisation of [⁶⁸Ga]Ga-NODAGA-exendin-4 radiosynthesis with final conditions highlighted in bold

Parameters	Tested/selected conditions
Precursor concentration	10-20-25-30-50 µg
Labelling time	10-12-14–15 min
Labelling temperature	85-95-105 °C
Polysorbate 20	0/0.05/0.5%
TLC stationary phases*	1. iTLC SG paper 11 cm (Agilent) 2. DC-Fertigfolien Polygram SIL G/UV254 (Macherey-Nagel) 3. TLC cellulose F, plastic sheets (Merck) 4. Longer iTLC SG paper: 12 cm (Agilent)
TLC mobile phases*	1. Ammonium 1 M/DMF 50:50 pH 5.5 2. EDTA 0.05 M pH 5 3. Ammonium 1 M/DMF 30:70 pH 8.5 4. Ammonium 1 M/Methanol pH 3.5 5. Ammonium 1 M/DMF 30:70 pH 5.5 6. Ammonium 1 M/DMF 30:70 pH 3.5 7. Sodium Citrate 0.1 M pH 5

* Optimised TLC analysis conditions were established in Liège and were slightly different from those applied in Nancy (see Table 4)

Reagents and chemicals

All radiosyntheses were carried out on the miniAllinOne synthesiser (operated by the Supervision software, Fig. 2) using commercially available consumables from Trasis : generic “68Ga without pre-purification” cassettes and the associated reagent kits specifically designed for each generator type (ITM, IRE, and E&Z, see Table 3). The composition of acetate buffer used to bring the generator eluate to adequate pH, as well as the inclusion of a hydrochloric acid (HCl) syringe to elute the generator, varies depending on the generator employed. Beyond these generator-specific variations, some components remain consistent across all kits. These include (i) ethanol (EtOH), used for eluting the radiolabelled product from the HLB cartridge (Oasis HLB Plus Short Cartridge 225 mg Sorbent, 60 μm Particle Size, Waters), (ii) a saline bag (0.9%) for the final formulation, (iii) a Phosphate buffered saline (PBS) 10x/polysorbate 20 (0.5%) solution, which helps reducing peptide adsorption to the synthesis cassette and tubing, thereby reducing residual activity and improving apparent radiochemical yield (Ragot et al. 2026; Menon et al. 2024; Migliari et al. 2022; Tokgöz et al. 2023; Siebels et al. 2024; Lefebvre et al. 2021).

Fig. 2.

Kit components, cassette and layout involved in the [⁶⁸Ga]Ga‑NODAGA‑exendin‑4 procedure. Set-up visualisation on the miniAllinOne system (left), including the kit components and cassette required for gallium‑68 radiolabelling of exendin‑4. The layout within the Suppervision software allows real-time monitoring of the radiosynthesis steps

Table 3.

Reagent characteristics of kits depending on the generator employed

Generator	GeGant® (ITM)	Galli RD® (IRE-Elit)	GalliaPharm® (Eckert & Ziegler)
Acetate buffer	0.3 M, 1 mL	0.5 M, 1 mL	0.7 M, 1 mL
HCl	0.05 M, 4 mL	0.1 M, 1.1 mL (from the generator itself)	0.1 M, 5 mL

The Lys40-NODAGA-exendin-4 precursor was purchased from piCHEM (Grambach, Austria) and supplied as a Good Manufacturing Practice (GMP) grade lyophilised powder (50 µg). Gallium-68 was obtained from 68Ge/68Ga generators: ITM, IRE-Elit, E&Z.

Validation and quality control analysis

To validate the developments carried out using the ITM generator, three validation runs were performed using the full set of parameters identified as optimal (Table 2, bold), with the aim of assessing the reproducibility and robustness of the automated process. Following successful validation on the ITM, the procedure (see Fig. 3) was subsequently tested on the second in-house IRE generator as well as on the E&Z generator through a collaboration with the radiopharmacy at CHRU of Nancy.

Fig. 3.

Flowchart summarising the automated synthesis of [⁶⁸Ga]Ga-NODAGA-exendin-4, incorporating the optimised steps from precursor handling to final radiolabelled product

All quality control assays usually required on the final product (EDQM 2026a, b) were performed for all validation runs (Table 4) (Tokgöz et al. 2023; Costes et al. 2023). These included visual inspection, pH measurement, radiochemical purity assessment by TLC and high performance liquid chromatography (HPLC) and residual solvents quantification by gas chromatography (GC). Additionally, measurements of residual activity in all compartments of the cassette were performed and labelling yields (non-decay corrected, n.d.c and decay corrected, d.c.) were determined.

Table 4.

Specifications for final product quality control

Parameters	Tests	Specifications
Appearance	Visual inspection	Clear and colourless
pH	pH strip	4.0 < pH < 8.0
Radiochemical purity	HPLC and TLC	[68Ga]Ga-NODAGA-exendin-4 ≥ 95% of total activity Sum of free 68Ga, 68Ga colloids and radio-HPLC impurity ≤ 5%
Residual solvent (EtOH as excipient)	GC	EtOH ≤ 10% V/V
	/	Polysorbate 20 < 3 mg/kg/day (32)

TLC analyses, following methodological optimisation as mentioned above, were performed using iTLC SG plates, with a mobile phase consisting of ammonium acetate (AcONH₄, pH 5.5) and N, N-dimethylformamide (DMF). HPLC analyses were performed using a C18 column with a mobile phase gradient ranging from 95% of 0.1% Trifluoroacetic acid (TFA) in water/5% of 0.1% TFA in acetonitrile to 100% of 0.1% TFA in acetonitrile. Radiochemical yields were calculated based on the initial activity and the activity measured in each key compartment of the cassette (reactor, HLB cartridge, sterile filter, waste vial, and final product), while accounting for the decay of gallium-68 over time. The pH was measured with pH-indicator strips (pH 2.0–9.0). Residual solvent quantification was performed by GC (GC-2010 Plus, Shimadzu). A capillary stationary phase composed of 6% cyanopropyl phenyl and 94% dimethyl polysiloxane was employed (30 m length, 0.25 mm inner diameter, 1.4 μm film thickness), with injections of 1.5 µL per sample. The detector temperature was set to 300 °C and the column temperature to 40 °C, using helium as the carrier gas, for a total running time of 20 min. All site-related specificities of the quality control (QC) analyses are detailed in Table 5.

Table 5.

Site-dependent specificities of QC analyses

Specificities	CHRU Nancy	Trasis Liège
Synthesisers	miniAllinOne 5397 Hardware version 5.2	miniAllinOne 5254 Hardware version 5.1
TLC plates Agilent	iTLC-SG 10 cm	iTLC-SG 12 cm
TLC scanners	MiniGITA, Elysia-Raytest	Bioscan AR-2000
TLC mobile phases	1.3 M AcONH4-DMF (50/50, v/v)	1 M AcONH4-DMF (30/70, v/v)
HPLC systems	Vanquish ThermoFischer	Alliance Waters
HPLC Columns	Acclaim 120 C18, 150 × 3 mm, 3 μm, ThermoFischer	XBridge BEH Shield RP18, 130Å, 5 μm, Waters
HPLC detectors	Uv : VC-D11-A Gamma: Raytest Gaby	Uv : PDA waters Gamma: Raytest nod
HPLC methods	Overall analysis time: 28 min Wavelength (λ): 220 nm Flow: 0.6 mL/min Temperature: 30 °C Injection volume: 40 µL	Overall analysis time: 15 min λ: 220 nm Flow: 0.6 mL/min Temperature: RT Injection volume: 50 µL
Ionisation chambers	Capintec CRC-55R	Capintec CRC-120R
pH-indicator strips	pH-fix 2–9, Macherey Nagel	MQuant pH-indicator, Supelco

The full development was transferred to Nancy for validation on their Eckert & Ziegler generator.

Results

Optimised radiosynthesis

Optimisation of the automated radiosynthesis followed an iterative process where each selected parameter was varied individually (see Table 6), and in some cases in combination, to assess its impact on both the quality of the radiolabelled product and the radiochemical yield. The values presented in the table below summarise the complete set of tests performed during synthesis optimisation, illustrating the impact of specific changes.

Table 6.

Impact of modified process parameters on yields and residual activities on the cassette components

Experiments	Exendin (µg)	Temperature (°C)	Time (min)	Polysorbate 20 (%)	RCY (n.d.c.) %*	Activity in waste vial %*	Activity in radiolabelling reactor %*
1	10	95	15	0	18	41	27
2	10	95	15	0.05	33	36	14
3	50	95	15	0.5	73	2	1
4	20-25-30	95	15	0.5	68-64-64	9-11-15	1-2-1
5	20	95	10-12-14	0.5 (after radiolabelling)	56-52-57	18-22-18	11-8-6
6	20	95	12	0.5 (after radiolabelling) + steps inversion	63	3	2

*Activity % are expressed as % of the total activity. Summary of the main changes between tests (highlighted in bold). Each row is a single test, detailing the precursor amount used, the temperature, the synthesis time, as well as the polysorbate 20 addition to the process and the stage at which it was introduced. All of these modifications are presented here with their impact on the RCY, the residual activity in the reactor, and the activity recovered in the waste (compartments with the most pronounced effects)

Based on these results, the addition of PBS/Polysorbate to the saline pouch clearly improves radiochemical yield and reduces activity losses within the reactor. As expected, increasing the amount of precursor also has a beneficial impact on yields (50 µg vs. 10 µg). However, since the precursor quantity is limited by human injectability requirements (Ragot et al. 2026; Velikyan et al. 2017; Kraihammer et al. 2023; Tokgöz et al. 2023), intermediate amounts (20-25-30 µg) were evaluated and provided acceptable conversion yields. Finally, a precursor quantity of 20 µg was selected, as no significant improvement was observed at higher levels (25–30 µg). The radiolabelling temperature and reaction time were also varied to determine whether conditions previously optimised for precursors such as PSMA-11 and DOTATOC remained suitable for exendin-4 (Li et al. 2025; Reverchona et al. 2020; Maus et al. 2020). While the optimal temperature remained unchanged, the radiolabelling time could be reduced to 12 min (Fig. 4), which is particularly advantageous when working with short-lived isotopes such as gallium-68. Finally, several steps of the process were re-examined to further optimise the synthesis performance : (i) a rinsing step of the precursor vial with saline was removed, leading to a higher precursor concentration during radiolabelling and ensuring that polysorbate was only introduced in the reactor after completion of the synthesis. In addition, the generator elution and precursor addition steps were interchanged (elution first, precursor transfer second) compared to the original process. These two major process modifications improved the yields while maintaining low levels of residual activity.

Fig. 4.

Reaction scheme depicting the radiolabelling of NODAGA-exendin‑4 with gallium‑68, alongside key radiolabelling parameters (figure created using ChemDraw)

Validation

Once all optimised parameters were selected (see Table 7), the automated process was validated by performing three radiosynthesis on all three generators ; ITM, IRE, E&Z. The goal of this validation was to verify that the process systematically delivers QC compliant [⁶⁸Ga]Ga-NODAGA-exendin-4 as well as to demonstrate batch-to-batch reproducibility.

Table 7.

Optimised parameters used for validation runs on all three generators

Parameters	Validation runs
Temperature of reaction	95 °C
Time of reaction	12 min
Sequence	1st : generator elution 2nd : precursor transfer into the reactor Addition of PBS 10x/Polysorbate 20 0.5% in NaCl 0.9% after the radiolabelling step
Precursor amount	20 µg

All validation syntheses using the ITM generator were carried out in our laboratory in Liège. At the end of each synthesis, the activity in the relevant parts of the cassette, as well as in the final product and the waste vial, was measured using an ionisation chamber (see Table 5 for system references and Table 8 for results). These measurements allowed the calculation of decay corrected and non-decay corrected radiochemical yields.

Table 8.

Residual activities on cassette components and radiochemical yields for the ITM generator validation runs

ITM	Sterilising filter (Pall, 0.22 μm)	Reactor	Final product	HLB cartridge	Waste vial	RCY (n.d.c.)	RCY (d.c.)**
Run 1*	6	2	85	4	4	66	85
Run 2*	5	2	87	4	2	66	87
Run 3*	5	1	86	4	4	67	86

*Activity % are expressed as % of the total activity, **decay-corrected to a reference time corresponding to the time of the start of synthesis

The measured activities were consistent across all three runs, indicating excellent process reproducibility. In addition, minimal activity levels were observed in the reactor (from 1.3 to 2%), the waste vial (from 2.1 to 4.2%) and the purification column (from 3.5 to 5.8%), confirming that the synthesis proceeded efficiently and provided high radiochemical yields (66 to 67% n.d.c. ; 85 to 87% d.c.).

The entire process was then validated on the IRE and E&Z generators (see Table 9). Again, the measured residual activities and the calculated yields demonstrated the transferability of the automated radiosynthesis regardless of the generator used (RCY = 60–65% n.d.c. and 76–82% d.c.), with good and consistent performance. The syntheses performed with the IRE generator were carried out in the same facilities as those used for the ITM generator, whereas the syntheses using the E&Z generator were conducted in collaboration with the radiopharmacy of the CHRU of Nancy.

Table 9.

Yields and residual activity measured in different cassette components for the GMP generators

IRE	Sterilising filter (Pall, 0.22 μm)	Reactor	Final product	HLB cartridge	Waste vial	RCY (n.d.c.)	RCY (d.c.)**
Run 1*	9	9	77	4	2	61	76
Run 2*	9	8	78	4	2	62	78
Run 3*	7	6	81	4	2	64	81

E&Z	Sterilising filter (Pall, 0.22 μm)	Reactor	Final product	HLB cartridge	Waste vial	RCY (n.d.c.)	RCY (d.c.)**
Run 1*	17	3	76	4	Not measured	60	76
Run 2*	11	2	78	3	6	61	78
Run 3*	9	2	83	3	4	65	82

*Activity % are expressed as % of the total activity, **decay-corrected to a reference time corresponding to the time of the start of synthesis

Quality control analysis

Measurements of pH, final product volume and radiochemical purity by TLC and HPLC were consistently performed after each validation run (see Table 10). All results were within the expected specifications for a gallium-68 radiopharmaceutical product (Table 3). Radiochemical purity, a key quality parameter, was above the 95% limit in all cases. Besides product purity, excipients were confirmed to remain below their maximum authorised limits for human use.

Table 10.

Quality control data for the nine validations runs performed

ITM	RCP TLC (%) Spec : ≥ 95%	RCP HPLC (%) Spec : ≥ 95%	EtOH (%) Spec : ≤ 10% v/v	pH Spec : 4.0 < pH < 8.0	Final volume (mL) Spec : 10 ± 0.5 mL*
Run 1	98.9	≥ 99	7.7	5.5-6	9.8
Run 2	96.9	≥ 99	7.6	5.5-6	9.7
Run 3	97.7	≥ 99	7.7	5.5-6	9.6

IRE	RCP TLC (%)	RCP HPLC (%)	EtOH (%)	pH	Final volume (mL)
Run 1	99.6	≥ 99	Not reported**	5.5	9.9
Run 2	99.4	≥ 99	Not reported**	5.5	9.8
Run 3	99.3	≥ 99	Not reported**	5.5	9.7

E&Z	RCP TLC (%)	RCP HPLC (%)	EtOH (%)	pH	Final volume (mL)
Run 1	99.9	> 97	Not reported**	7	9.6
Run 2	97.7	> 96	Not reported**	6	9.7
Run 3	97.7	> 96	Not reported**	6	9.5

The specifications (Spec) from Table 4 are mentioned in the first line of the table but are applicable to all data. * Internal specification usually applied on automated processes developed by Trasis. ** EtOH was not quantified for these runs. However, based on the amount of EtOH used to elute the radiolabelled product (900 µL) from the HLB cartridge and the final product volume (10 mL), the EtOH content will always be below 10% v/v

Accordingly, the ethanol used to elute the compound of interest from the purification cartridge was quantified, and the polysorbate 20 used to minimise losses of radiolabelled product in the reactor was estimated, with both remaining in all cases below their maximum authorised limits (Ragot et al. 2026; European Medicines Agency 2019). It is to be noted that EtOH content is expected to comply to the ≤ 10% specification by design, since the total volume of EtOH in the kit does not exceed 0,9 mL for a final formulation volume of 10 ± 1 mL. Representative TLC chromatograms for one validation run on each generator (see Fig. 5: A) ITM, B) IRE and C) E&Z generators) clearly show a well-defined [⁶⁸Ga]Ga-NODGA-exendin-4 peak with limited tailing thanks to the optimised mobile phase.

Fig. 5.

TLC and HPLC chromatograms of [⁶⁸Ga]Ga-NODAGA-exendin-4 obtained with each generator. A-A’) ITM, B-B’) IRE and C-C’) E&Z. Chromatograms A and B were acquired using the BIOSCAN AR-2000 scanner (CRC, ULiège) with a 1 M AcONH₄-DMF (30/70) mobile phase, while chromatogram C was obtained using an ELYSIA Raytest reader (CHRU of Nancy) with a 1.3 M AcONH₄-DMF (50/50) mobile phase. All analyses were performed using iTLC-SG strips, although the strip size differed slightly between Trasis’ lab and the CHRU (see Table 5). HPLC chromatograms A′ and B′ show the radioactive profile at T0 post-radiosynthesis for [⁶⁸Ga]Ga-NODAGA-exendin-4 obtained at the CRC, while C′ shows the corresponding profile obtained at the CHRU of Nancy

QC analyses performed at the CHRU of Nancy revealed the presence of a secondary peak of low intensity at a shorter retention time (R_t ≈ 6 min), in addition to a shoulder on the main peak of the radiolabelled product. These features were further accentuated when the tests were repeated after extended time intervals (T + 2 h and T + 4 h, results not shown). A review of the literature indicates that this behaviour is consistent with the limited stability of the precursor once placed in solution, leading to degradation and the formation of an oxidised species, this phenomenon can be further increased due to radiolysis (Migliari et al. 2022; Velikyan et al. 2017; Lefebvre et al. 2021; Janota et al. 2016; Nelson et al. 2022). Following these observations, this degradation pathway was investigated by assessing the addition of antioxidants during the radiosynthesis process. These assessments were repeated using the optimised procedure with both Galli RD^® and GalliaPharm^® generators and validated by reproducing runs on each system (see Table 11).

Table 11.

Yields and residual activity obtained from validation runs after scavenger addition

IRE	Reactor	Final product	HLB cartridge	Waste vial	RCY (n.d.c.)	RCY (d.c.)**
Run 1*	5	90	4	1	71	90
Run 2*	4	91	4	1	72	91
Run 3*	3	92	4	1	73	92

E&Z	Reactor	Final product	HLB cartridge	Waste vial	RCY (n.d.c.)	RCY (d.c.)**
Run 1*	2	93	3	2	73	93
Run 2*	3	93	4	1	73	93
Run 3*	3	91	4	1	72	91

*Activity % are expressed as % of the total activity, **decay-corrected to a reference time corresponding to the time of the start of synthesis. Yields and residual activity were measured in the different cassette components for the process implemented on GMP generators, following the addition of ascorbic acid with the precursor and sodium ascorbate in the saline used for formulation

Ascorbic acid was incorporated during precursor aliquoting (15 mg ascorbic acid for 20 µg precursor) to stabilise precursor/radiolabelled product during radiolabelling, while sodium ascorbate addition to the saline solution (540 mg sodium ascorbate in 50 mL saline) aimed at increasing the stability of the formulated [68Ga]Ga-NODAGA-exendin-4 radiopharmaceutical. Those steps resulted in a significant reduction of the oxidised form, with the stabilising effect being maintained for QC analyses performed at extended time points (T + 2 h and T + 4 h, Fig. 6). Moreover, they were associated with a significant improvement in radiosynthesis yields, enabling the highest yields reported to date (Table 11).

Fig. 6.

Representative chromatograms from validation runs on IRE generator obtained at T0 (A), T +2 h (B) and T +4 h (C) without scavenger addition, compared with chromatograms obtained after the addition of ascorbic acid and sodium ascorbate at T0 (A’), T +2 h (B’) and T +4 h (C’). The orange arrows highlight the appearance of additional peaks and the shoulder on the main peak, both observed in the absence of scavengers

Discussion

The present work addresses a well-recognised translational bottleneck in the clinical implementation of GLP-1 receptor-targeted PET imaging, namely the absence of a harmonised, robust and readily transferable radiosynthesis protocol for [⁶⁸Ga]Ga-NODAGA-exendin-4 (Menon et al. 2024; Migliari et al. 2022; Kraihammer et al. 2023; Tokgöz et al. 2023; Nelson et al. 2022). Although the diagnostic performance of GLP-1R-targeted tracers for insulinoma localisation has been documented in both preclinical and clinical studies (Menon et al. 2024; Boss et al. 2020; Kraihammer et al. 2023; Boss et al. 2022; Selvaraju et al. 2013; Antwi et al. 2019; Luo et al. 2016; Boss et al. 2024), their routine adoption has been hampered by heterogeneous synthesis strategies, variable precursor handling and limited generator-to-generator transferability (Isal et al. 2018; Chakravarty et al. 2016; Durieux et al. 2023; Paty et al. 2024). In this context, the present study deliberately emphasises practical process optimisation, automation and regulatory compatibility. By systematically evaluating and validating key synthesis parameters, we propose a pragmatic and clinically implementable solution that aligns with current expectations for routine hospital radiopharmacy (Migliari et al. 2022; Tokgöz et al. 2023; Nelson et al. 2022).

A central outcome of this work is the demonstration that formulation-related parameters, which are often underreported in the literature, exert a major influence on radiochemical yields and residual activity distribution (Ragot et al. 2026; Migliari et al. 2022; Velikyan et al. 2017; Nelson et al. 2022). Previous reports on [⁶⁸Ga]Ga-labelled exendin-4 analogues have primarily focused on precursor amount, buffer composition or generator elution profiles (Menon et al. 2023; Kraihammer et al. 2023; Tokgöz et al. 2023; Nelson et al. 2022), while adsorption phenomena within disposable synthesis cassettes and optimisation of analytical methods for QC release have received comparatively little attention (Ragot et al. 2026; Menon et al. 2024; Velikyan et al. 2017; Siebels et al. 2024). The present data notably shows that the addition of polysorbate 20 after completion of the radiolabelling step is a decisive factor in improving apparent yields and reducing activity losses within the reactor and tubing. This observation is fully consistent with earlier descriptions of peptide adsorption to polymeric surfaces at low mass concentrations, particularly for highly lipophilic or surface-active peptides. Importantly, delaying the addition of polysorbate 20 minimises the risk of interfering with gallium complexation chemistry and avoids radiolabelling mixture dilution, thereby having a positive impact on radiochemical conversion. From a routine production perspective, this translates into improved reproducibility and a more predictable activity balance, both of which are critical for clinical scheduling (Ragot et al. 2026; Menon et al. 2024; Migliari et al. 2022; Kraihammer et al. 2023; Tokgöz et al. 2023; Nelson et al. 2022).

Several reports have demonstrated that higher precursor amounts can drive radiochemical conversion towards completion (Menon et al. 2024; Velikyan et al. 2017), but such approaches are frequently incompatible with accepted limits for injectable peptide mass in humans (Migliari et al. 2022; Kraihammer et al. 2023; Tokgöz et al. 2023). By identifying 20 µg of NODAGA-exendin-4 as an optimal compromise between conversion efficiency and clinical acceptability, this study refines earlier protocols and provides a quantitatively justified rationale for precursor minimisation. The observation that yields do not increase further beyond approximately 20–25 µg is in agreement with the notion that, under optimised labelling conditions, factors other than precursor availability become rate-limiting (Xie et al. 2023; Nelson et al. 2022). This finding is particularly relevant for centres seeking to balance performance, cost and regulatory compliance.

Reaction temperature and labelling duration were assessed with reference to conditions commonly employed for other gallium-68-labelled peptides (Li et al. 2025; Reverchona et al. 2020; Maus et al. 2020), including somatostatin and PSMA derivatives. Consistent with the broader literature, an elevated temperature of 95 °C was confirmed as optimal for NODAGA-exendin-4 radiolabelling. However, the present data demonstrate that the reaction time can be safely reduced from 15 to 12 min without compromising radiochemical purity or yield. Given the short physical half-life of gallium-68, even modest reductions in synthesis time are of practical importance and directly increase the available activity for patient administration.

Beyond the optimisation of individual parameters, the refinement of the synthesis workflow itself proved to be a key contributor to overall performance. The inversion of generator elution and precursor addition, together with the removal of an unnecessary precursor vial rinsing step, resulted in a higher effective precursor availability at the moment of radiolabelling and a more efficient use of the eluate activity. The present results reinforce the importance of treating automated synthesis as an integrated sequence, in which seemingly minor procedural adjustments can produce substantial cumulative gains in robustness and yield.

The validation data generated using disposable cassettes and reagent kits on the miniAllinOne synthesiser strongly support the transferability of the optimised protocol across different commercially available gallium-68 generators. Despite known differences in generator design, elution profiles and metallic impurity levels, comparable yields, residual activity distributions and quality control outcomes were obtained using ITM, IRE-ELiT and Eckert & Ziegler generator systems. This robustness across different generator platforms represents a major advantage over several previously reported methods, which are often optimised for a single generator type. From the perspective of routine radiopharmacy practice and regulatory harmonisation, such flexibility substantially lowers the barrier to adoption and facilitates inter-centre standardisation.

Quality control results consistently met predefined acceptance criteria, with radiochemical purity exceeding 95% in all validation runs and residual solvents and pH values remaining well within pharmacopoeia limits (EDQM 2026a, b). These findings confirm that the simplification and streamlining of the synthesis did not compromise product quality. On the contrary, the observed reduction in variability supports the notion that robust process design is a prerequisite for reliable routine production, particularly for peptide-based radiopharmaceuticals.

Stability investigations performed during validation, revealed the emergence of minor secondary peaks and shoulders in HPLC chromatograms at later time points. Such features have been previously reported for exendin-4 derivatives and are generally attributed to oxidative or chemical degradation of the peptide once in solution (Kraihammer et al. 2023; Lefebvre et al. 2021; Janota et al. 2016). The present observations therefore reflect an intrinsic limitation of the molecular stability rather than a deficiency of the radiosynthesis protocol. Although these minor impurities are unlikely to have clinical relevance given the short interval between synthesis and administration, their identification remains important from a quality and documentation standpoint (Ragot et al. 2026; Migliari et al. 2022; Velikyan et al. 2017).

The validation focused on CMC optimization and standardization of [68Ga]Ga-NODAGA-exendin-4 and did not include prospective clinical imaging data. However, since the precursor employed in these radiosynthesis is identical in all respects with the precursor used in routine clinical use, the emergence of in vivo differences would not be anticipated. Stability assessments beyond the typical clinical administration window were exploratory and not intend to formally support extended shelf-life claims. Moreover, beyond improving the stability of the final product, the addition of this protective agent enabled highest ever reported non-decay-corrected synthesis yields of 71–73%. While the protocol was successfully transferred between centres, broader multi-centre experience would further strengthen process robustness data.

Conclusion

This study demonstrates that a simplified, efficient and harmonised automated synthesis of [⁶⁸Ga]Ga-NODAGA-exendin-4 can be achieved while improving radiochemical performance and product quality. By explicitly addressing clinically relevant constraints, such as precursor mass limits, synthesis duration, automation using plug-and-play consumables, and generator variability, the proposed protocol bridges the gap between methodological optimisation and routine clinical application. The work provides a solid and transferable foundation for the routine clinical implementation of GLP-1R PET imaging in patients with suspected insulinoma.

Abbreviations

AcONH₄

Ammonium acetate

CRC

Cyclotron Research Centre

CT

Computed Tomography

d.c.

Decay Corrected

DMF

Dimethylformamide

E&Z

Eckert and Ziegler generator : GalliaPharm

EDTA

Ethylenediaminetetraacetic Acid

EtOH

Ethanol

⁶⁸Ga

Gallium-68

GC

Gas chromatography

GLP-1R

Glucagon-like peptide-1 Receptor

GMP

Good Manufacturing Practice

H₂O

Dihydrogen Monoxide

HCl

Hydrochloric Acid

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

HLB

Hydrophilic-lipophilic balanced

HPLC

High-performance liquid chromatography

¹¹¹In

Indium-111

IRE

IRE-Elit generator : Galli RD^®

iTLC

Instant Thin Layer chromatography

ITM

ITM generator : GeGant^®

λ

Wavelength

MeOH

Methanol

MRI

Magnetic Resonance Imaging

NaCl

Sodium chloride

n.d.c.

Non-decay corrected

PBS

Phosphate Buffered Saline

PET

Positron Emission Tomography

PLC

Programming logic control

QC

Quality control

RCP

Radiochemical Purity

RCY

Radiochemical Yield

RT

Room temperature

SCX

Strong Cation Exchange

Spec

Specifications

T°C

Temperature

TFA

Trifluoroacetic Acid

TLC

Thin Layer chromatography

Uv

Ultraviolet

Authors’ contributions

ED contributed to project development, process optimisation and validation, and manuscript writing and revision. MS contributed to manuscript writing and revision. MH contributed to experimental optimisation and validation. CC contributed to manuscript revision and improvement. CV supervised process development and optimisation and contributed to manuscript revision. NV contributed to process validation, manuscript writing, and revision. CW contributed to manuscript revision and improvement. All authors read and approved the final manuscript.

Funding

This project (study/ work) is co-funded by the French State-Region contract CPER 2015–2020 (Contrat de Plan Etat Région- IT2MP Innovations Technologiques, Modélisation et Médecine Personnalisée), and CPER 2021–2027 (R-IRM, Réseau Grand Est en Imagerie et Robotique Médicales), by the French “Grand Est” Region (IRMGE project) and by the European Union through the European Regional Development Fund.

Data availability

A majority of the data generated or analysed during this study is included in this article. Data not included can be made available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable. This study did not involve human or animal data or tissue.

Consent for publication

Not applicable. This manuscript does not contain data from any individual person.

Competing interests

ED, CV, CW and MH are employees of Trasis, which may have a financial interest in the subject of this work. The authors declare that they have no other competing interests.