Radiolabeling was performed by an automated procedure using a Modular-Lab module (Eckert & Ziegler, Berlin, Germany) with a ⁶⁸Ge/⁶⁸Ga-Generator (GalliaPharm® by Eckert & Ziegler AG, Berlin, Germany) in a laminar flow hood. In summary, a StrataXC cartridge combined radioactivity of two generators. The StrataXC cartridge was preconditioned with 5 ml of 0.2 M HCl (Merck suprapur®). After trapping the ⁶⁸Ga activity, the StrataXC cartridge was rinsed with air, and the remaining activity was eluted with 800 µl of 0.06% HCl in acetone into the reaction vial. The vented reaction vial contained 45 µg of precursor, 150 µl 0.2 M sodium acetate (Merck suprapur®) buffer (pH 3.8–4.0) and 150 ml of absolute ethanol (Ph. Eur.) to reach a reaction pH of 2.0–2.5. After a reaction time of 5 min at 95°C, the reaction was quenched by adding water and the product was trapped onto a C18-cartridge (WAT023501, Waters) preconditioned with 1 ml EtOH and 3 ml H₂O. Purification involved eluting the product with 2 ml EtOH:H₂O directly through a sterile filter into the product vial and rinsing the cartridge with 8 ml NaCl 0.9% (B.Braun) again through the sterile filter for product dilution. The quality control release criteria (endotoxin level <5.00 EU/ml; radio-TLC > 95%; radio-HPLC > 95%; pH 4.0–8.0) were met for all radiosyntheses. Sterility was tested retrospectively after complete decay (3 days post-synthesis). The syntheses typically yielded 450 ± 45 MBq of the final product, starting from about 600 MBq. ⁶⁸Ga-Trivehexin was administered in compliance with the German Medicinal Products Act (Arzneimittelgesetz §13 Abs. 2b) and with the approval of the responsible local regulatory authority.

Between July 2021 and October 2022, a retrospective analysis was conducted on 44 patients (23 female, 21 male; mean age 62.6 years, range 37–82 years) who had undergone ⁶⁸Ga-Trivehexin PET/CT (Table 1). One patient was examined three times. All patients provide written informed consent in advance. The data analysis received approval from the responsible local ethics committees (EK-242052023).

All PET measurements were conducted using a digital Siemens Biograph Vision 600 scanner (Siemens Medical Solutions, Erlangen, Germany). A low-dose CT (120 kVp, 78 mAs, spiral pitch factor of 1.5, 512 × 512 matrix with a pixel distance of 0.98 mm) without contrast was performed for attenuation correction before the subsequent PET scan. The 3D PET data was obtained in list mode using continuous bed motion. Patients received intravenous administration of 139 MBq (range: 84–160 MBq) of ⁶⁸Ga-Trivehexin, molar activity 31 ± 8 MBq/nmol, mass dose 4.7 ± 1.6 nmol, with no adverse effects associated with the radiotracer application. For dynamic PET imaging, ⁶⁸Ga-Trivehexin was injected intravenously over 20 s. List mode acquisition commenced at the start of ⁶⁸Ga-Trivehexin infusion, covering 45 min in one bed position. Subsequent sequential PET frames (6 × 30 s, 7 × 60 s, 7 × 300 s) were reconstructed. After dynamic PET and 55–85 min after ⁶⁸Ga-Trivehexin injection, static ⁶⁸Ga-Trivehexin-PET/CT scans were acquired with five bed positions assessing the whole body, aligning with the exposure time of ¹⁸F-FDG.

PET images were reconstructed following our standard routine using an ordered subset expectation maximization (OSEM) 3D iterative reconstruction algorithm with 4 iterations and 5 subsets (4i5s), applying point spread function, and time of flight with an image matrix size of 440 × 440, resulting in a voxel distance of (1.65 × 1.65 × 1.65) mm³. Reconstructions were performed with attenuation and relative scatter correction and no post-filtering.

Data were analyzed by a certified nuclear medicine physician and radiologist using dedicated workstation and software (Syngo MMWP and Syngo TrueD, Siemens Medical Solutions, Erlangen, Germany). For quantitative analysis of tracer uptake, a SUV-based analysis in the respective lesions was performed. For static PET scans, SUVmean and SUVmax of pancreatic cancer and metastases were measured using a volume of interest (VOI) technique. VOIs were defined by an automatic isocontour with a cutoff at 50% of SUVmax. The tumor-to-liver ratio (TLR) was calculated. To obtain the SUVmax for the liver, three VOIs (>1.5 cm³) were drawn from the right hepatic lobe and averaged (SUVavg), respectively. TLR was calculated as TLR = SUVmax/SUVavg. For dynamic PET imaging analysis, VOIs of pancreatic cancer (SUV pancreatic cancer), pancreas (SUV pancreas background), stomach wall (SUV stomach wall), blood pool (SUV blood pool), liver (SUV liver), as well as liver metastasis (SUV liver metastasis) and lymph node metastasis (SUV lymph node metastasis) were drawn and applied to the entire dynamic dataset.

Formalin-fixed paraffin-embedded (FFPE) tissue blocks of pancreatic ductal adenocarcinomas (PDAC) from the archive of the Institute of Pathology of the University hospital Dresden were cut into 2 µm sections using a rotating microtome. Human tissue was immunohistochemically examined using an anti-human Integrin β6 (ITGB6) mouse monoclonal antibody [clone 442.5C4] (#407317, dilution 1:100, Merck Millipore, Burlington, Massachusetts, USA). Immunohistochemistry (IHC) was performed on an autostainer (Bond RX^m, Leica Biosystems, Wetzlar, Germany). Antigen retrieval (AR) was achieved by using enzyme pretreatment (Bond^TM Enzyme Pretreatment E1, #AR9551, Leica Biosystems, Wetzlar, Germany) for 5 min. Finally, antibody binding was visualized by using a brown chromogen [3,3’-diaminobenzidine (DAB)] (#DS9800, Bond Polymer Refine Detection, Leica Biosystems, Wetzlar, Germany). Slides were scanned in 40× magnification using a bright field slide scanner (Leica AT2) and subsequently digitally evaluated (Aperio ImageScope ×64).

Membranous staining intensity of the majority of tumor cells on each slide (score 0–3) and the frequency of β6-integrin positive tumor cell membranes (in%) were assessed per whole tissue section. A combined final score was then calculated by multiplying the factors score × frequency [histological score modified according to Sipos et al. (21)].

Data were retrospectively collected from patients undergoing clinical examinations. The dataset included patient demographics, examination details, and various medical findings. All statistical analyses were performed using Python (version 3.8.10) in combination with Pandas (version 1.2.4) and Statsmodels (version 0.12.2). Basic statistical measures, including mean, median, standard deviation, minimum, and maximum were computed for age, dose, and time until examination.

For static PET parameters, such as SUVmax of pancreatic cancer, metastases, and TLR, comprehensive data were collected. Box plots were generated for a visual assessment of data distribution, outliers, and spread. A one-way ANOVA test was conducted to ascertain statistically significance differences among group means. Upon identifying significant differences (p < 0.05) in the ANOVA test, a Tukey HSD post-hoc test was applied to identify which specific groups’ means (compared pairwise) were different. To analyze tumor metastases intensity metrics across subgroups with and without previous chemotherapy, we conducted a comprehensive analysis on a dataset comprising SUVmax and TLR measurements. Subsequently, t-tests for independent samples were performed to assess the statistical significance of observed differences. The threshold for statistical significance was set at p < 0.05. Cases with insufficient data for robust analysis were flagged and excluded from further evaluation.

For dynamic PET data, measurements (SUVmax and SUVmean) were taken at specific time intervals post-injection, varying from short intervals at the beginning to longer intervals towards the end. Each interval comprised multiple measurements for every organ and tissue (n = 11, n (liver metastases) = 13, n (lymph node metastases) = 13). Averaging over time within each interval yielded interval-specific SUVs. Data were presented for SUV measurements over time (t). The primary objective was to identify the time point where the SUV of the tumor (either mean or max) peaked while maximizing the difference between the tumor's SUV and that of other regions. The central aim was to derive ratios of SUVmax of pancreatic cancer to background tissue (tumor-to-background ratio, TBR), specifically, tumor-to-pancreas, tumor-to-liver, tumor-to-blood pool, tumor-to-stomach wall, and to determine the time point for each ratio to peak. Therefore, we employed the following mathematical formulation:tmax(R)=argtmaxR(t)Where: •

tmax(R) indicates the time when the ratio R achieves its maximum value.

In our dataset of 44 patients, the mean age was approximately 62.6 years, with a range from 37 to 82 years (Table 1). The gender distribution was nearly equal, with 52% being female and 48% male.

The average administered activity was 139 MBq, ranging from 84 to 160 MBq. Static acquisitions were started between 55 and 85 min p.i. (avg. 62 min).

At the commencement of the investigation, a collective of 23 patients were identified with pre-existing distant metastases, of which 10 presented simultaneous distant metastatic and lymph node metastases. One patient exhibited only lymph node metastases. The predominant pattern revealed solitary pancreatic cancer in 17 patients. A collective of 13 patients exhibited liver metastases (Figure 1), while six had peritoneal carcinomatosis (Figure 2). Less frequent observations included secondary neoplasms, osseous metastases, pulmonary metastases, splenic metastases, pleural carcinomatosis and metastases in soft tissues, among other observations. Furthermore, there have been patients with liver metastases occurring concurrently with lymph node metastases and pancreatic cancer, but less frequently.

Uptake of ⁶⁸Ga-Trivehexin in non-metastatic organs was most frequently observed in the uterus, tissues recently tangentially affected by surgery or intervention, and less frequently in non-enlarged lymph nodes (Table 2). Notably, intense uptake was observed in a supratentorial brain metastasis (left cerebellopontine angle) of a squamous cell carcinoma of the tonsil (SUVmax 18.5), in a simultaneous extrahepatic cholangiocarcinoma of pancreatobiliary subtype (SUVmax 21.5) with intense uptake of the entire pancreatic parenchyma, and a histologically confirmed squamous cell carcinoma of the left upper lobe (SUVmax 8.8) as seen in Figure 3. It is important to note that these are cases of pancreatic cancer with concomitant metachronous tonsillar carcinoma and its metastases. Specifically, it was a tonsillar carcinoma with cervical metastases on the right with subsequent surgical resection and RCTx.

In our dataset, the frequency for each category was as follows: pancreatic tumor (n = 40), liver metastases (n = 39), lymph node metastases (n = 21), peritoneal carcinomatosis (n = 17), pulmonary metastases (n = 2), bone metastases (n = 4), splenic metastasis (n = 1), pleural carcinomatosis (n = 3) and soft tissue metastases (n = 4) (Figure 4). Preliminary analysis indicated variability in SUVmax and TLR measurements across each category as seen in Table 3. A one-way ANOVA test was conducted (F-value = 4.55, p = 7.1193 × 10⁻⁵), demonstrating significant differences in the means across groups (p < 0.05). This warranted the execution of further post-hoc tests. Tukey post-hoc analysis revealed a significant difference (p < .001) between the groups SUVmax (liver metastasis) and SUVmax (pancreatic cancer) [4.74, 95%-CI (1.74, 7.75)], SUVmax (lymph nodes metastasis) and SUVmax (pancreatic cancer) [4.07, 95%-CI (0.47, 7.67)], TLR (liver metastasis) and TLR (pancreatic cancer) [1.82, 95%-CI (0.83, 2.80)], TLR (pancreatic cancer) and TLR (peritoneal carcinomatosis) [−1.88, 95%-CI (−3.15, −0.61)], and TLR (pancreatic cancer) and TLR (pleural carcinomatosis) [−2.79, 95%-CI (−5.42, −0.18)].

In the examination of tumor metastases intensity across two subgroups, delineated as with or without CTx prior scan, various metrics were assessed for their significance. Notably, the TLR (pancreatic cancer) and TLR (peritoneal carcinomatosis) metric revealed a significant difference. Other metrics, such as SUVmax (pancreatic cancer) and SUVmax (peritoneal carcinomatosis), approached significance but did not surpass the commonly accepted threshold of p < 0.05 as seen in Table 4. Several metrics could not be adequately assessed due to insufficient data.

During dynamic PET imaging, both liver and pancreatic tissues showed a temporary increase in activity, with the SUVmean of the pancreas surpassing that of the liver (Figure 5). After reaching a maximum, the SUVmean of both organs steadily decreased. Pancreatic cancers exhibited delayed uptake compared to the surrounding normal pancreatic tissue. Over the dynamic scanning process, there was an accumulation of ⁶⁸Ga-Trivehexin within the tumor, ultimately leading to the formation of a plateau, following only a slight reduction in the SUVmean. Liver metastases and lymph node metastases had a comparable pattern to pancreatic malignancies.

Interestingly, the rate and magnitude of activity uptake in the stomach wall surpasses that observed in pancreatic cancer during the initial phase of the dynamic scan. After approx. 10 min, the SUVmean of the stomach wall reached a plateau and subsequently decreased. After 17.5 min, the SUVmean of pancreatic cancer exceeded that of the stomach wall. The difference in SUVmean values between pancreatic cancer and other organs, including the stomach wall, pancreas background, and liver, gradually increased throughout the dynamic measurement until completion. The analysis pinpointed the optimal imaging time for maximum contrast between pancreatic cancer and surrounding tissues (TBR) to be 45 min following radiotracer injection (Figure 6).

23 primary pancreatic ductal adenocarcinoma (PDAC) resection specimens and two distant metastases (liver, lung) contained a sufficient amount of tumor cells for immunohistochemical evaluation of membranous ITGB6 expression. All PDACs showed clear positivity for membranous ITGB6. The examined samples demonstrated varying degrees of ITGB6 positivity: 3/25 (12%) scored 1, 9/25 (36%) scored 2, and 13/25 (52%) scored 3 (for examples see Figure 7).