Introduction
The fibroblast activation protein (FAP) is a membrane serine protease expressed by fibroblasts. FAP has recently emerged as one of the most promising target structures for molecular imaging and therapy in cancer [
1,
2]. In contrast to the ubiquitous presence of non-FAP-expressing quiescent fibroblasts, FAP expression by activated fibroblasts in the adult is (with few exceptions) linked to pathologic states such as wound healing, organ fibrosis, and cancer, where it is abundantly expressed by cancer-associated fibroblasts (CAFs) in the tumor microenvironment [
3].
Radiolabeled FAP-targeting small ligands based on FAP inhibitors (FAPI) have recently been introduced and demonstrated very promising characteristics for PET imaging such as high and fast uptake in a variety of cancers and rapid clearance from the majority of healthy organs [
4‐
6]. As of now, a growing body of clinical studies has established a high versatility of FAPI-PET to detect the spread of a wide range of cancers. Importantly, FAPI-PET seems to be able to fill clinically urging shortcomings of
18F-FDG, for example in pancreatic cancer [
7], primary liver tumors [
8], gastric and bowel cancer [
9], and breast cancer [
10].
In addition to its role in diagnostic imaging, reliable expression in cancer and wide absence in other tissues potentially qualify FAP as a theranostic molecular target. Radioligand therapy using α- or β-emitting isotopes appears tempting, in light of recent break through clinical studies of theranostic agents targeting somatostatin receptors in neuroendocrine tumors [
11] and prostate-specific membrane antigen (PSMA) in prostate cancer [
12]. However, first published retrospective studies on clinical radioligand therapy with FAPI agents leave open questions regarding their therapeutic efficacy [
13,
14]. In addition, targeted delivery of non-radioactive drugs and chimeric antigen receptor (CAR) T-cells has been discussed to employ FAP for targeted molecular therapy [
15]. For all of such therapeutic approaches, FAP-targeted molecular imaging will play an indispensable role to stratify patients.
Different specific FAP radioligands have been introduced, of which the ligands initially developed at the University of Heidelberg, i.e., FAPI-04 [
6] and FAPI-46 [
16], are most abundantly and very successfully used in published PET imaging studies. Recently, the new very high-affinity FAP ligand OncoFAP was shown to possess selective FAP binding in vitro and striking tumor uptake in animal models after
177Lu or fluorophore labeling [
15]. In this translational study, we describe the radiosynthesis of [
68Ga]Ga-OncoFAP-DOTAGA (
68Ga-OncoFAP), its preclinical evaluation in murine tumor models, and results from first clinical application of
68Ga-OncoFAP in different cancers.
Material and methods
Radiosynthesis, in vitro tests, and assays
68Ga-FAPI-46 was synthesized as described previously [
10]. FAPI-46 precursor for preclinical and clinical imaging was kindly provided under a MTA by U. Haberkorn (Heidelberg, Germany). OncoFAP-DOTAGA was synthesized as previously reported [
15]. Radiolabeling was performed on the basis of the German Pharmaceuticals Act (AMG §13 (2b)), i.e., magistral preparation. Briefly, radiogallium (
T½ = 68 min,
β+ = 89%, and
EC = 11%) was automatically eluted with 0.1 M HCl (0.36%) from a 50 mCi (1.85 GBq)
68Ge/
68Ga radionuclide generator (EZAG, Berlin, Germany) without pre-purification of the eluate and transferred into the reaction vessel of the
68Ga-radiosynthesis module. Two types of single-use disposable cassette-based modules, either a manual iQS Ga-68 Fluidic Labeling Module (itG, Garching, Germany) or a fully automated labeling module (miniAllinOne, Trasis, Ans, Belgium), containing a pre-heated, buffered OncoFAP-DOTAGA solution, were used for radiolabeling. After incubation for a few minutes at ca. 100 °C, the reaction mixture was loaded onto a SPE cartridge, eluted with EtOH in the product vial, and formulated with additional 0.9% NaCl. For clinical imaging, a full QC was performed for each preparation of
68Ga-OncoFAP. All QC parameters (see Suppl. Table
1) were in accordance with the Ph. Eur. standards for
68Ga-DOTA-TOC (monograph 2482). For preclinical imaging, a restricted QC was performed and
68Ga-OncoFAP or
68Ga-FAPI-46 product solutions were subsequently concentrated by rotary evaporation under reduced pressure to remove the ethanol and redissolved in a small volume of physiological saline, suitable for injection into mice.
The metabolic stability of 68Ga-OncoFAP was assessed after incubation with human and mouse blood serum at 37 °C by analytical radio-HPLC performed at different time points after incubation (i.e., 30, 60, 90, and 120 min).
The lipophilicity of
68Ga-OncoFAP-DOTAGA was determined according to previously described procedures [
17].
Enzymatic inhibition activity and selectivity of OncoFAP derivatives was assessed by employing commercially available in vitro fluorescence assays for FAP, dipeptidyl peptidase 8 (DDP8), and prolyl oligo peptidase (POP) (BPS Bioscience, San Diego, CA, US). The fluorogenic substrate (Fluorogenic DPP substrate 1, Ala-Pro-AMC dipeptide, AMC: 7-amino-4-methylcumaryl) was incubated together with recombinant FAP, DDP, or POP/PREP in the presence or absence of test compounds. The enzymatic activity was correlated with the amount of cleaved fluorescent product measured by fluorescence spectroscopy. IC50 of each test compound was determined from the curves obtained by plotting fluorescence intensities at different concentrations of the inhibitor.
Animal studies
All experiments were conducted in accordance with the German Law on the Care and Use of Laboratory Animals and approved by the Landesamt für Natur, Umwelt und Verbraucherschutz of North Rhine-Westphalia, Germany.
For tumor-xenograft models, female NMRI nu/nu mice (Janvier, France), 8–9 weeks old, were housed at a constant temperature (22 °C) and relative humidity (40–55%) under a regular light/dark schedule. Food and water were available ad libitum.
Tumor cells (10
6 cells in 50 μL saline) were implanted subcutaneously in the shoulders of NMRI mice with HT1080 wildtype (FAP
−) in one shoulder and with stably transfected human FAP-expressing HT1080 cells (FAP
+) in the contralateral shoulder. Cells were kindly provided by U. Haberkorn, University of Heidelberg [
5]. At day 11 after implantation, mice received whole-body PET/MRI (1 T Mediso nanoScan) imaging with list mode PET for 60 min and additional non-contrast-enhanced anatomic, coronal T1-weighted MRI, followed by late static PET/MRI imaging 180 min p.i. Either
68Ga-OncoFAP or
68Ga-FAPI-46 (
n = 6 each) was injected into the tail vein with a syringe pump at 1000 μL/min in a volume of 100 μL 30 s after the start of PET. At the next day (day 12 after implantation), the same mice were injected with the respective other FAP-tracer and received identical dynamic PET/MRI for 60 min. Sixty minutes p.i., mice were sacrificed under deep narcosis and organs and body fluids were harvested, weighed, and γ-counted. All involved measuring hardware (dose calibrator ISOMED 2010, automatic γ-counter PERKIN ELMER WALAC 2480, Twilite coincidence detector Swisstrace, PET/MRI) were cross-calibrated for
68Ga. Overall, 3 scans were rejected, because of extravasation (day 11 OncoFAP), inability of i.v. access (day 12 FAPI-46), and insufficient radioactivity (day 12 FAPI-46), respectively. The resulting injected amounts of radioactivity for the remaining 11
68Ga-OncoFAP scans and 10
68Ga-FAPI-46 scans were 19.2 ± 4.4 MBq and 18.2 ± 4.8 MBq, respectively.
Dynamic PET scans were reconstructed in time frames of 1 × 30 s, 5 × 12 s, 5 × 60 s, 4 × 300 s, 2 × 600 s, and 1 × 810 s. Frames 12 and 18 and the late scan are referred to as (rounded) 10 min, 1 h, and 3 h throughout the manuscript. Volume of interests (VOI) of tumors and representative organs were defined on anatomic MRI. Resulting tumor volumes for 68Ga-OncoFAP scans for FAP+ and FAP− tumors were 0.4 ± 0.36 and 0.22 ± 0.15, and for FAPI-46 scans were 0.34 ± 0.20 and 0.2 ± 0.17 (n.s. different between the two tracers).
Preclinical pharmacokinetic modeling
Before PET/MRI at day 12, p.i. mice received surgery to establish an extracorporeal circulation shunting of the blood from the femoral artery to the tail vein or contralateral femoral vein as described previously [
18]. At 40 cm from the femoral artery, the shunt (0.3 mm inner diameter silicon; total extracorporeal volume 56.6 μL) was led through a Twilite coincidence detector. Acquired blood activity curves were calibrated with a separately measured calibration factor for
68Ga according to vendor’s instructions. The curves were corrected for delay and dispersion based on a numerical deconvolution, using dispersion kernels derived from measuring step functions at defined pumping speeds. Standard 2-tissue compartmental (2TCM) pharmacokinetic (PK) modeling and Patlak analysis was performed for the triceps muscle, and FAP
− and FAP
+ tumors using PMOD version 3.703 (PMOD Technologies LLC) and was based on mean PET VOI time activity curves and Twilite-based extracorporeal arterial input functions (AIF). AIF plasma fraction was calculated based on direct hematocrit measurements after imaging (StatStrip® Hb/Hct, Nova Biomedical).
Patients
We retrospectively analyzed 68Ga-OncoFAP-PET/CT and PET/MRI scans of 12 patients. Primary tumors were breast cancer in 8 patients. The other 4 patients had fibrosarcoma, colon cancer, hepatocellular carcinoma, and an unclear cystic pancreatic tumor, respectively. Patients were referred by their treating oncologist on an individual clinical basis to support initial staging or specific diagnostic challenges in relapsed cancer. All patients gave written informed consent for 68Ga-OncoFAP-PET/MRI and/or PET/CT imaging and retrospective scientific analysis. Analysis has been approved by the Ethics Committee of the Medical Association of Westphalia-Lippe and the Medical Faculty of the University of Münster (Az. 2021-408-f-S). This study includes all 68Ga-OncoFAP-PET/CT and PET/MRI scans conducted at the University Hospital Münster 01-06/2021. No exclusion criteria were applied.
Clinical PET/CT and PET/MRI
Patients were injected with 163.3 ± 50 MBq
68Ga-OncoFAP. Patient were scanned in supine whole-body PET/CT (mCT, Siemens Healthineers) or PET/MRI (mMR, Siemens Healthineers) ~ 1 h p.i. as described previously [
10]. Breast cancer patients additionally received prone breast PET/MRI before the whole-body scan ~ 30 min p.i., and subsequent diagnostic contrast-enhanced CT or MRI was added according to the specific clinical demand; see Table
3 for details. One patient underwent dynamic imaging initiated with tracer injection in PET/CT, consisting of list mode PET of a mediastinal field of view for 2 min (reconstructed frames 6 × 10 s, 3 × 20 s) and subsequent whole-body dynamic scanning consisting of 6 bed positions that were scanned 7 × 45 s from the skull base to mid-thighs. Eventually, 40 min p.i., a whole-body scan with 3-min per bed positions was acquired. Reading of PET/MRI and PET/CT was performed according to a standard clinical workflow. Standard uptake value (SUV) measurements were acquired in syngo.via (Siemens Healthineers) with circular or spherical volumes of interests.
Statistics
Statistical analysis was performed using MATLAB (R2020a, MathWorks). Mann-Whitney U tests were performed for pairwise comparisons. p values < .05 were considered statistically significant. If needed, Bonferroni correction was applied to account for multiple testing. All values are displayed as mean ± std if not stated otherwise.
Discussion
OncoFAP has been recently introduced as a new very high-affinity ligand with striking FAP targeting in small animals upon fluorophore- and
177Lu-labeling [
15]. Here, we demonstrate the feasibility of
68Ga-radiolabeling and highly favorable targeting properties in small animal and clinical PET imaging in patients with cancer, validating
68Ga-OncoFAP as a new powerful alternative to clinically established PET tracers.
68Ga-labeled clinical PET tracers require a reliable radiosynthesis, high radiochemical yields, and high radiochemical purities based on an efficient, rapid, and simple labeling procedure using state-of-the-art equipment. 68Ga-OncoFAP-DOTAGA labeling fulfilled these requirements without reservation on a manual and a fully automated synthesis module. The determined low lipophilicity was advantageous for imaging application since the renal excretion route is favored and low background can be obtained. The high FAP binding affinity and metabolic stability were ideally suited for whole-body PET imaging justifying its exploration in small animals and clinical translation.
In vivo blocking experiments are traditionally employed to proof specific binding of PET tracers. We used a different approach of simultaneously implanting individual animals with a FAP-negative and a genetically altered FAP-expressing and otherwise identical tumor line. Accordingly, the observed manyfold difference of uptake between FAP+/FAP− tumor lines can be solely attributed to specific tracer binding. We further substantiated this reasoning by applying pharmacokinetic modeling based on an extracorporeally derived arterial input function. Here, FAP+ and FAP− tumors had similar results for modeling parameters associated with perfusion, permeability, and passive retention, whereas parameters reflecting specific target binding were significantly different.
The radiotracers FAPI-02, FAPI-04, and FAPI-46 developed at the University of Heidelberg, Germany, were the first available radiotracers with excellent imaging properties [
4,
5]. Other FAP radiotracers appear to be so far used only in few centers as, e.g., [68Ga]Ga-DOTA.SA.FAPi [
21,
22], or FAP-2286 [
13]. In lack of direct comparison, it is currently unclear which of these radiotracers possess the optimal properties for PET imaging. To evaluate the competitiveness of OncoFAP, we compared it to the well-established tracer FAPI-46. OncoFAP was recently described as the small organic FAP ligand with the highest affinity [
15]. Our observed higher inhibitory activity in a fluorescence-based FAP enzymatic assay for
natGa-OncoFAP compared to
natGa-FAPI-46 well-substantiates this assessment. We further benchmarked OncoFAP in head-to-head in vivo biodistribution studies against FAPI-46. Again, we observed higher
68Ga-OncoFAP uptake in murine FAP-positive tumors after 1 h in PET imaging and γ-counting. Also, the significant difference between the tracers in pharmacokinetic modeling for parameters k3/k4 and Vs confined to FAP
+ tumors likely reflects a difference in FAP affinity. Washout characteristics of the two tracers, which are relevant for future theranostic applications, as assessed by the 3 h imaging time point, were not found to be statistically different.
Clinical imaging performance of
68Ga-OncoFAP is well in line with prior experience in our center using
68Ga-FAPI-46 [
10]. In the group of 12 patients, the tracer reliably bound to primary cancers, lymph nodes, and distant metastases and is rapidly cleared from unaffected organs. Notably, contrasting the higher-than-expected hepatic uptake in preclinical
68Ga-OncoFAP biodistribution, the only significant difference in tracer uptake was a lower
68Ga-OncoFAP uptake in the liver compared to
68Ga-FAPI-46, rendering relevant hepatic tracer metabolism unlikely at least in humans. Mirroring our recently published results with
68Ga-FAPI-46 in breast cancer [
10],
68Ga-OncoFAP-PET imaging led to establish or substantiate novel sites of disease and facilitated workup in a variety of clinical scenarios. Uptake in cancer was not significantly different between the tracers in our small study sample. Thus, future studies have to investigate whether the experimentally observed better affinity and higher uptake in tumor models finally translates to a superior contrast of
68Ga-OncoFAP in clinical cancer imaging.
Our study features limitations. The preclinical head-to-head comparison of 68Ga-OncoFAP with 68Ga-FAPI-46 is based on a small number of tumor bearing mice that in part redundantly contributed to data for gamma counting, PET quantification, and PK modeling, leading to remaining statistical uncertainties. This explains why we do not overemphasize this comparison. Analysis of clinical translation is retrospective and the number of patients is rather small. The clinical comparison of 68Ga-OncoFAP and 68Ga-FAPI-46 is underpowered and possibly features selection bias.
In conclusion, excellent preclinical and clinical PET imaging characteristics validate 68Ga-OncoFAP as a powerful alternative to currently available FAP tracers. Prospective studies are needed to define its accuracy in relevant clinical scenarios. Moreover, the potential of OncoFAP to deliver therapeutic payloads to cancer requires further preclinical and clinical investigation.
Acknowledgements
We are grateful to the technologists who essentially contributed to this work: Sandra Hoeppner, Monika Trub, Jaqueline Hildgartner, Sven Fatum, Lukas Töns, Roman Priebe, Nina Kreienkamp, Dirk Reinhardt, Stefanie Bouma, and Christine Bätza.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.