The impact of different cell models on the preclinical evaluation of FAP-targeting radiopharmaceuticals

Aim/Introduction: Fibroblast activation protein (FAP) is an attractive cancer-specific biomarker for targeted radionuclide theranostics. Accordingly, FAP-targeting radiopharmaceuticals have shown clinical success for PET/CT imaging. Unfortunately, results with FAP-targeted radionuclide therapy (TRT) have been suboptimal to date. Preclinical research plays a key role in improving FAP-TRT efficacy and selecting the best FAP-targeting radiopharmaceutical for this purpose. For this, selecting an appropriate and clinically relevant tumor model is
crucial. Although FAP is mainly expressed by cancer-associated fibroblasts and not by cancer cells, transduced human (hu) FAP-expressing cancer cells are often used as a model. To what extent these models are appropriate for studying FAP-targeting radionuclide theranostics remains unknown. Therefore, we aimed to compare tracer behavior in two commonly used FAP-transduced cancer cell lines (HT1080 (fibrosarcoma) and HEK293 (human embryonic kidney)), one endogenous FAPexpressing glioblastoma cell line (U87MG), and an endogenous FAP-expressing pancreatic stellate cell line (PS-1).
Materials and Methods: Tracer uptake studies were performed by incubating cells with 1 nM 20MBq/nmol [111In]In-FAPI-46 or 10 nM RTX1370S, a fluorescent FAP-targeting tracer (kindly provided by Ratio Therapeutics). Additionally, membrane-bound versus internalized [111In]In-FAPI-46 was separated using Glycine-NaCl and NaOH, respectively. Radioactivity was measured in a γ-counter and RTX1370S was imaged by confocal microscopy. Moreover, [111In]In-FAPI-46 autoradiography was performed on HT1080huFAP, HEK293ThuFAP, and U87MG xenografts. Lastly, the expression of human (hu)FAP and murine (mu)FAP was analyzed by IHC and RT-qPCR.
Results: HT1080huFAP had the highest [111In]InFAPI-46 uptake in vitro (22.4%), followed by HEK293huFAP (17.3%). Unexpectedly, higher FAP expression did not result in higher uptake, as a more intense FAP staining and a 3.5-fold higher mRNA expression were observed for HEK293huFAP vs HT1080huFAP. U87MG and PS-1 demonstrated much lower [111In]In-FAPI-46 uptake (2.2% and 2.9%, respectively). Autoradiography studies indicated similar [111In]In-FAPI-46 binding to HT1080huFAP and HEK293huFAP (31.6% vs 32.6%, respectively), and contrary to in vitro uptake, a remarkable ex vivo binding to U87MG xenograft
(15.4%). The presence of muFAP mRNA levels was measured in all xenografts, indicating radiotracer binding is partly facilitated by the infiltration of murine fibroblasts in this model. In addition, confocal microscopy demonstrated potent uptake of RTX1370S, and while [111In]In-FAPI-46 was mainly internalized in all cell lines, membrane-bound RTX-1370S was visible in one cell line, illustrating that tracer localization can be impacted by the model.
Conclusion: Our data shows that FAP (radio)tracer behavior is model specific and that, depending on the research question, appropriate model selection is crucial for evaluating and understanding FAP-targeting radiopharmaceuticals.