A Theranostic Approach for CAR-T Cell Therapy

In this issue of Clinical Cancer Research, Lee and colleagues (1) explore how molecular imaging may be used to guide chimeric antigen receptor (CAR)-T cell therapies directed against fibroblast activation protein α (FAP). FAP has been identified as a promising therapeutic target by IHC studies performed in the 1980s (2, 3). These studies showed that FAP is highly expressed by most sarcomas and by cancer-associated fibroblasts (CAF) in the stroma of many carcinomas. In contrast, FAP expression by normal adult tissues is highly restricted. Resting fibroblasts in normal tissues lack FAP expression but express FAP when activated during tissue remodeling, for example, during wound healing, inflammatory conditions, or after myocardial infarction (4).

FAP is a membrane-bound peptidase of the dipeptidyl peptidase 4 family and can exist in a heterodimeric complex with dipeptidylpeptidase 4 which is also known as FAP β (4). FAP exhibits a post-proline dipeptidyl peptidase as well as endopeptidase activity and can degrade matrix proteins such as collagen I and III but also peptide hormones and growth factors, such as neuropeptide Y and FGF-21. FAP is involved in extracellular matrix remodeling and can promote the migration of stromal and tumor cells. In addition, the number of FAP-expressing stromal cells has been shown to correlate with higher levels of angiogenic factors and an immunosuppressive microenvironment (4, 5). Depletion of CAFs has caused tumor regression in animal models (6), but an increase in tumor invasiveness and tumor progression has also been observed (7). Detailed analyses of the tumor stroma have defined several subtypes of CAFs, such as iCAFs and myoCAFs (8), which show different gene expression profiles and opposing effects on tumor growth. Clinical studies have shown that the degree of FAP expression is generally correlated with a worse prognosis, although with notable exceptions in some studies (9). In summary, FAP is a highly interesting but complex therapeutic target.

As early as 1994, a clinical trial reported that the radiolabeled FAP antibody F19 specifically accumulates in colorectal cancer metastases (10), highlighting the potential of FAP-targeting imaging agents for cancer detection and staging. These encouraging results were not followed up for several years because the slow pharmacokinetics of antibodies make it necessary to image patients several days after injection. However, a humanized version of F19, Sibrotuzumab, did undergo clinical testing for treatment of patients with FAP-positive malignancies. Sibrotuzumab as a naked antibody was found to be safe but demonstrated only limited efficacy (11). Several FAP antibody–drug conjugates and bispecific antibodies have also shown promising results in preclinical studies. Clinical trials are ongoing for some of these agents (RO7122290 and MP0317).

These therapeutic approaches do not specifically target the enzymatic function of FAP but use the FAP protein to deliver cytotoxic substances to the tumor tissue or to elicit an immune response. In addition, inhibitors of FAP peptidase activity have been developed to inhibit the invasiveness of malignant tumors (4). One of these inhibitors, the small molecule Talabostat (Val-boroPro) which inhibits FAP and other peptidases has been tested clinically but showed limited activity (12).

More recently, FAP-specific, high-affinity inhibitors, based on a quinanoline-cyanopyrrolidine scaffold have been developed (13). Data on the biological effects of these new inhibitors are limited so far, but radiolabeled derivatives of the inhibitor UAMC1110 (14) have been successfully translated to the clinic for imaging of FAP expression with PET (15). The most extensively studied compound is ⁶⁸Ga-FAPi-04 which consists of the UAMC1110 FAP binding motif, a piperazine linker and the chelator DOTA for labeling with the positron emitting radioisotope Gallium-68 (14). ⁶⁸Ga-FAPi-04 is rapidly internalized by FAP-expressing cells and demonstrates excellent pharmacokinetics in humans. Tumor-to-organ ratios on human ⁶⁸Ga-FAPi-04 PET/CT scans have been shown to be similar or superior to ¹⁸F-FDG-PET scans in several malignancies (16). This has stimulated the development of a variety of high-affinity FAP ligands for nuclear imaging and targeted radionuclide therapy (16–18). Multiple studies on the diagnostic accuracy of PET with ⁶⁸Ga-FAPi-04 and other FAP ligands and their impact on patient management have been published or are ongoing (16).

The ¹⁸F-FAPi-74 (16) used by Lee and colleagues (1) is a FAP ligand with optimized pharmacokinetics that can be labeled with the commonly available PET isotope Fluorine-18, which makes ¹⁸F-FAPi-74 more broadly available than ⁶⁸Ga-FAPi-04. Lee and colleagues show that small animal PET with ¹⁸F-FAPi-74 allows for noninvasive imaging of FAP expression within the mouse stroma of A549 non–small cell lung cancer xenografts. The PET signal was robust enough to monitor FAP-targeted CAR-T cell therapy. At 2 weeks after CAR-T cell injection, the PET signal decreased by approximately two-thirds while the tumor volume had not significantly changed when compared with baseline. Loss of FAP expression in the tumor tissue after CAR-T cell injection was confirmed by ex vivo immunofluorescence microscopy, which also showed a close correlation between the PET signal and the level of FAP expression.

Because of its highly restricted expression in normal organs, FAP is an attractive target for CAR-T cell therapy in cancer and other diseases (4), although FAP expression has been described in the bone marrow of mice and some experimental studies reported hematotoxicity following CAR-T cell therapy (4).

Currently, CAR-T cell therapy is monitored by assessing tumor volume and glucose metabolic activity with CT or ¹⁸F-FDG PET/CT. These imaging modalities are nonspecific and assess relatively late effects of CAR-T cell therapy (loss of viable tumor cells). Furthermore, response assessment by these imaging modalities may be misleading due to “pseudoprogression,” that is, an increase in tumor size and metabolic activity due to infiltration of the tumor tissue by inflammatory cells. CT and ¹⁸F-FDG PET/CT also cannot provide insights why CAR-T cell therapies are effective or not.

PET imaging of CAR-T cell targets can facilitate the preclinical and clinical development of CAR-T cells in oncology and other diseases (19, 20) by quantifying the degree of FAP expression before therapy, assessing the heterogeneity of FAP expression in patients with multiple metastases, and provide unique insights in the effectiveness of the CAR-T cells. Baseline PET imaging can help in patient stratification by quantifying target expression in tumor and normal tissues. During follow-up, common issues in CAR-T cell therapy such as antigen escape can be assessed. For example, if no tumor shrinkage is observed after CAR-T cell therapy, PET can assess whether the target is still expressed within the tumor tissue. Persistent target expression would demonstrate that the CAR-T cells were ineffective, whereas loss of target expression would indicate that the tumor tissue has lost the target but remains viable. Furthermore, reappearance of the CAR-T cell target signal can potentially be used as an early marker of tumor recurrence.

The preferred targets for molecular imaging agents in oncology share several requirements with CAR-T cell targets, that is, the target should be present on the surface of cancer cells, and there should be limited expression by normal tissues. Therefore, it is no coincidence that an optimized molecular imaging agent for FAP expression was available for the study by Lee and colleagues (1). Similar experiments could be performed with prostate-specific membrane antigen (PSMA)-targeting CAR-T cells and PSMA ligands. Radiation exposure caused by PET imaging with small molecular imaging agents is usually lower than of CT imaging, making it feasible to perform multiple serial imaging studies in a patient before and after CAR-T cell therapy. Thus, targeted CAR-T cells and PET could well become a powerful theranostic pair that facilitates the clinical translation of CAR-T cell therapies (Fig. 1).

W.A. Weber reports grants from NC3Rs during the conduct of the study; in addition, W.A. Weber, V. Morath, and K. Fritschle have a patent for DTPA-R CAR-T cell imaging issued. No disclosures were reported by the other author.

NC3Rs CRACK IT Challenge 32: Transgene Track supported W. Weber, K. Fritschle and V. Morath. DFG project number 495342067 supported Z. Varasteh.