In vitro and ex vivo evaluation of preclinical models for FAP-targeted theranostics: differences and relevance for...

Fibroblast activation protein (FAP) is an attractive target for cancer theranostics. Although FAP-targeted nuclear imaging demonstrated promising clinical results, only sub-optimal results are reported for targeted radionuclide therapy (TRT). Preclinical research is crucial in selecting promising FAP-targeted radiopharmaceuticals and for obtaining an increased understanding of factors essential for FAP-TRT improvement. FAP is mainly expressed by cancer-associated fibroblasts in the tumor stroma and less on cancer cells themselves. Therefore, other (complex) factors impact FAP-TRT efficacy compared to currently clinically applied TRT strategies. For accurate evaluation of these aspects, selection of a representative preclinical model is important. Currently mainly human cancer cell lines transduced to (over)express FAP are applied, lacking clinical representation. It is unclear how these and more physiological FAP-expressing models compare to each other, and whether/how the model influences the study outcome. We aimed to address this by comparing FAP tracer behavior in FAP-transduced HT1080-huFAP and HEK293-huFAP cells, and endogenous FAP-expressing U-87 MG cancer cells and PS-1 pancreatic stellate cells. [¹¹¹In]In-FAPI-46 and a fluorescent FAP-targeted tracer (RTX-1370S) were used to compare tracer binding/uptake and localization in vitro and ex vivo. Additionally, FAP expression was determined with RT-qPCR and anti-FAP IHC.

Although FAP expression was highest in HEK293-huFAP cells and cell line derived xenografts, this did not result in the highest tracer uptake. [¹¹¹In]In-FAPI-46 uptake was highest in HT1080-huFAP, closely followed by HEK293-huFAP, and a 6-10-fold lower uptake for U-87 MG and PS-1 cells. However, ex vivo U-87 MG xenografts only showed a 2-fold lower binding compared to HT1080-huFAP and HEK293-huFAP xenografts, mainly because the cell line attracts murine fibroblasts as demonstrated in our RT-qPCR and IHC studies.

The interaction between FAP and FAP-targeted tracers differs between models, indicating the need for appropriate model selection and that comparing results across studies using different models is difficult.