Fibroblast activation protein-based theranostics in cancer research: A state-of-the-art review

In recent years, quinoline-based fibroblast activation protein (FAP) inhibitors (FAPI) have shown promising results in the diagnosis of cancer and several other diseases, making them the hotspot of much productive research. This review summarizes the literature for the state-of-the-art FAPI-PET imaging for cancer diagnosis compared with fluorodeoxyglucose (FDG)-PET. We also summarize the use of FAPI-PET for therapeutic regimen improvement and fibroblast activation protein (FAP)-targeted molecule modification strategies, as well as preliminary clinical studies regarding FAP-targeted radionuclide therapy. Our qualitative summary of the literature to date can inform future research directions, medical guidelines, and optimal clinical decision-making.


Introduction
Approximately 19.3 million new cancer diagnoses and approximately 10.0 million cancer deaths occurred worldwide in 2020, according to estimates from Global Cancer Statistics [1]. Cancers develop in complex environments composed of tumor cells and the surrounding stroma. The "seed and soil" theory emphasized the interactive importance of both components as early as 1889 [2]. However, diagnostic and therapeutic paradigms have predominantly targeted only tumor cells.
The widespread application of immunotherapy in clinical trials has led to increased research attention being paid to the tumor microenvironment (TME). The tumor stroma or TME comprises all the noncancer components in the tumor tissue, including cancer-associated fibroblasts (CAFs), the extracellular matrix (ECM), various types of immune cells, and intertwined blood vessels. The TME can develop an immunosuppressive niche in which tumor cells are protected from conventional therapies, resulting in treatment failure [3].
CAFs are among the most abundant components of the TME in solid tumors [4]. However, CAFs are heterogeneous cells with both tumor-promoting and tumor-suppressive effects observed in different situations [5]. CAFs model and remodel the ECM structure, which can become a physical barrier against the infiltration of immunocytes, which have a killer function, or a structural scaffold for intercellular interaction between tumor cells and non-tumor cells in the TME, thereby regulating tumor initiation, neovascularization, and metastasis [6]. On the other hand, CAFs can secrete multiple chemokines and cytokines, such as transforming growth factor-β (TGFβ), CC-chemokine ligand 2 (CCL2), and interleukin-6 (IL-6), in order to recruit immunocytes Ivyspring International Publisher with inhibitory functions in the tumor stroma, thereby facilitating immune evasion [7].
CAFs have several biological markers, including α-smooth muscle actin, fibroblast activation protein (FAP), and platelet-derived growth factor receptor-β (PDGFRβ). FAP is a type II integral membrane glycoprotein belonging to the serine protease family involved in ECM remodeling and fibrogenesis [8]. In colorectal cancers, FAP expression is higher at the invasive front than in the tumor center [9].
Although the prognostic value of FAP in cancers has been inconsistent throughout the literature, high expression of FAP has been shown to be an independent poor prognostic marker for outcomes in lung cancer, hepatocellular carcinoma, and colon cancer in studies with large sample sizes (n = 138-449 patients) [10][11][12]. In a murine model, tumor growth could be potentiated by the constitutive expression of FAP, which could, in turn, be meaningfully attenuated by anti-FAP antibodies [13]. However, sibrotuzumab (a humanized version of the murine anti-FAP antibody) failed as a treatment regimen in an early phase II trial for metastatic colorectal cancer [14]. Although FAP antibodies have shown limited response in tumor therapy, small molecules targeting FAP have attracted increasing attention in the area of tumor theranostics.

Targeting FAP for tumor imaging
FAP, expressed at low levels in healthy tissues, is involved in various pathological conditions and is detected in over 90% of malignant epithelial tumors [8,15]. Thus, molecular imaging (including positron emission tomography [PET] and single-photon emission computed tomography [SPECT]) targeting FAP is a promising diagnostic imaging modality.
A series of quinoline-based FAP inhibitors (FAPIs) was developed by the University Hospital Heidelberg group based on clinical and preclinical research [16][17][18]. The first FAPI variants (FAPI-01 and FAPI-02) were reported in 2018, demonstrating that FAPI-02 has an improved binding affinity to human FAP compared with FAPI-01 [16]. Impressively, 68 Ga-FAPI-02 PET images showed favorable tracer uptake in tumor tissues and low background uptake in normal healthy organs, resulting in high-contrast images in tumor-bearing murine models and three cancer patients [16]. Subsequently, FAPI-04 was identified as the best tracer from among 15 modified FAPIs for PET imaging applications, showing a higher tumor uptake in murine xenograft models than FAPI-02 [17]. Notably, 68 Ga-FAPI-04 was evaluated in a cohort of 80 patients presenting with 28 types of tumors (54 primary tumors and 229 metastases). It showed overall intense tracer uptake (with a maximum standard unit value [SUVmax] > 6) and high-contrast images in various highly prevalent cancers, including sarcomas, cholangiocarcinoma, and esophageal, breast, lung, hepatocellular, colorectal, head-neck, ovarian, pancreatic, and prostate cancers [19]. To improve the pharmacokinetics of this PET tracer, FAPI-46 was discovered from 15 other FAPI derivatives, and it demonstrated an enhanced tumor-to-background ratio (TBR) compared with FAPI-04 [18]. Interestingly, 68 Ga-FAPI-46 PET/CT imaging acquisition at an early time point (10 min p.i.) had an equivalent lesion detection rate compared with a late time point (60 min p.i.) in a previous study [20].
It should be noted that increased FAPI uptake has been reported in many non-oncological conditions, including inflammatory lesions, fibrotic disease, trauma, arthritis, degenerative bone disease, immunoglobulin 4 [IgG-4] related diseases, connective tissue disease, and atherosclerosis [23,24]. Thus, imaging interpretation with 68 Ga/ 18 F-FAPI PET/CT should be interpreted with caution to avoid misdiagnosis. Interestingly, increased FAPI uptake in non-oncological conditions could open possibilities for broader use of FAPI for corresponding diseases. For example, FAP-specific PET/CT could be used in the discrimination between inflammatory and fibrotic activity in IgG-4 related diseases [25], evaluation of the progression in atherosclerotic plaques [26], and assessment of the disease activity in fibrotic interstitial lung diseases [27,28].

Comparing 68 Ga-FAPI and 18 F-FDG uptake in various types of cancer
As FAPI is a novel PET tracer in cancer imaging, it is critical to evaluate lesion detection rates and diagnostic efficacy for FAPI compared with 18 F-fluorodeoxyglucose ( 18 F-FDG), the dominant tracer in oncology. To the best of our knowledge, Chen et al. conducted the first head-to-head study comparing 68 Ga-FAPI-04 and 18 F-FDG PET/CT in a cohort of 75 patients (54 patients identified at initial assessment and 21 patients with recurrence detection) with 12 different tumor entities. This prospective study demonstrated that 68 Ga-FAPI-04 had a higher sensitivity as compared with 18 F-FDG in identifying primary tumors (98.2% vs. 82.1%, P = 0.021), lymph node metastases (86.4% vs. 45.5%, P = 0.004), and bone and visceral metastases (83.8% vs. 59.5%, P = 0.004) [29]. However, the limited number of patients harboring each cancer type enrolled in this study did not allow for a subgroup comparison in terms of diagnostic efficacy for the same tumor type. Representative MIP images of both 18 -FDG PET/CT and 68 Ga-FAPI PET/CT in 8 patients with different types of cancer are shown in Figure 1. Some studies have recently compared the diagnostic efficacy of 68 Ga-FAPI and 18F-FDG in various types of tumors. 68 Ga-FAPI-04 PET/CT demonstrated improved sensitivity when compared with 18 F-FDG PET/CT for cancers of unknown primary origin (CUP), while the sensitivity for detecting primary tumors was comparable in nasopharyngeal carcinoma (NPC), oral squamous cell carcinoma, and Waldeyer's tonsillar ring cancer [30][31][32][33][34]. Impressively, 68 Ga-FAPI-04 PET/CT pinpointed 39% (7/18) of primary head and neck CUP tumors among primary patients with negative 18 F-FDG findings [30]. In another cohort of 45 patients with NPC, 68 Ga-FAPI-04 PET/CT showed higher radiotracer uptake than 18 F-FDG for primary tumors, regional lymph nodes, and distant metastases, resulting in higher sensitivity for the detection of lymph nodes and distant metastases [31]. Interestingly, Qin et al. reported that 68 Ga-FAPI-04 PET/CT detected a smaller number of positive lymph nodes compared with 18 F-FDG PET/CT (the detected number of positive lymph nodes was 48 vs. 100) [32].

Head and neck cancer
However, no suspicious FDG-avid lymph nodes were confirmed histologically. This study indicated that 68 Ga-FAPI PET/CT might be more specific than 18 F-FDG for differentiating reactive lymph nodes from tumor metastatic lymph nodes [35], and it is very likely that the FDG-positive/FAPI-negative lymph nodes are reactive lymph nodes. However, this finding requires validation in future research.

Digestive system cancer
In gastric cancer, Pang et al. and Qin et al. reported that the FAPI-derived SUVmax in primary and metastatic lesions was significantly higher than the FDG-derived SUVmax [39,40], while Jiang et al. and Kuten et al. reported that FAPI-derived TBR was higher than FDG-derived TBR without accompanying differences in the SUVmax [41,42]. In colorectal cancer, 68 Ga-FAPI-04 PET/CT has been demonstrated to have equal sensitivity in primary tumor detection compared to 18 F-FDG PET/CT (6/6, 100%), although the FAPI-derived SUVmax was statistically significantly higher than the FDG-derived SUVmax in a previous study [39].
Although 68 Ga-FAPI-04 has significantly higher uptake compared with 18 F-FDG in esophageal cancer, their sensitivity in detecting primary tumors was shown to be comparable [46]. Pancreatic tumors are characterized by intense stromal desmoplastic reactions surrounding cancer cells, and CAFs are the main effector cells involved in this desmoplastic reaction. As expected, 68 Ga-FAPI-04 PET/CT shows higher sensitivity in detecting primary tumors, including lymph nodes, and metastases than 18 F-FDG PET/CT in pancreatic cancer, mainly due to the intense 68 Ga-FAPI uptake in pancreatic tumor lesions [43]. However, non-specific 68 Ga-FAPI uptake in tumor-induced pancreatitis should be noted, as intense 68 Ga-FAPI uptake is normally observed throughout the whole pancreas in this disease. This phenomenon is frequently observed in tumors located in the head of the pancreas. As a solution, dual-time point 68 Ga-FAPI PET/CT (1 h early-point and 3 h late-point scans) may help differentiate pancreatitis from malignancy [43,44]. 68 Ga-FAPI-04 PET/CT was demonstrated to detect a greater number of cancerous lesions with a higher SUVmax for primary tumors, including lymph nodes, and distant metastases compared with 18 F-FDG PET/CT in a cohort of 48 breast cancer patients [47].

Breast cancer
Kömek et al. reported a similar conclusion in a cohort of 20 patients [48].

Hematological neoplasms
Head-to-head comparison studies of hematological neoplasms are relatively rare compared to studies evaluating solid tumors. Lan et al. reported that 68 Ga-FAPI-04 PET/CT demonstrated inferior sensitivity (50.65 vs. 96.75%) and accuracy (51.28 vs. 95.51%, P < 0.001) than 18 F-FDG in a subgroup of eight patients with hematological neoplasms, including multiple myeloma and lymphoma [52].
The primary lesion detection rates for 68 Ga-FAPI and 18 F-FDG PET/CT (based on published data) in various types of cancer are summarized in Figure 2. 68 Ga-FAPI PET/CT outperformed 18 F-FDG PET/CT in CUP, breast cancer, hepatocellular carcinoma, intrahepatic cholangiocarcinoma, gastric cancer, and pancreatic cancer. In these diseases, 68 Ga-FAPI PET/CT may have the potential to replace 18 F-FDG in the future. Conversely, 68 Ga-FAPI PET/CT was inferior to 18 F-FDG PET/CT for diagnosing hematological neoplasms. However, it should be noted that Figure 2 is not based on a systematic literature search of PubMed/MEDLINE and Cochrane Library database analysis, and the number of patients enrolled in previous studies was very small. It should also be noted that the results of Figure  2 only aimed at comparing the detection rate of primary tumors between 18 F-FDG and FAPI PET/CT based on existing investigations, including retrospective or smaller prospective studies. For more information, a systematic review was conducted by Treglia et al. to compare radiolabeled FAPI and 18 F-FDG PET/CT in oncologic imaging [53]. Clear comparability, especially superiority or inferiority, between FAPI vs. FDG should be demonstrated in future studies.
Overall, we conclude that 68 Ga-FAPI PET/CT has comparable or improved diagnostic performance in imaging various cancers compared to 18F-FDG PET/CT, particularly in cancer types that normally show a low-to-moderate 18 F-FDG uptake [54,55]. Lower background uptake in normal organs leads to higher TBRs with 68 Ga-FAPI when compared with 18 F-FDG [56]. As stroma components can comprise up to 50-90% of the tumor environment, stroma-targeted PET/CT imaging could be more sensitive than glucose metabolism-based PET imaging in detecting small lesions or lesions with low glucometabolic activity [57]. Moreover, 68 Ga-FAPI PET/CT provides advantages over 18 F-FDG PET/CT because it does not require fasting and imaging acquisition at early time points following injection (i.e., 10-60 min after tracer administration). However, it is too arbitrary to say FAPI PET/CT will replace FDG PET/CT based on the current clinical studies. The Role of FAPI PET in cancer imaging and management should be further explored in larger prospective trials before any conclusion should be reached.

Changes in cancer management according to 68 Ga/ 18 F-FAPI PET/CT
In addition to studies comparing 68 Ga-FAPI and 18 F-FDG PET/CT for the diagnosis of various types of cancer, several studies have explored the added value of 68 Ga-FAPI PET/CT in cancer management compared with standard care of imaging (SCI). In oral squamous cell carcinoma, 68 Ga-FAPI-04 PET/CT showed superior sensitivity (81.3% vs. 50.0%) and specificity (93.3% vs. 61.5%) in detecting lymph node metastases compared with CE-MRI [33]. Specifically, compared with CE-MRI, 68 68 Ga-labeled fibroblast activation protein inhibitor (FAPI) positron emission tomography/computed tomography (PET/CT) (lower row) reveal abnormal activity in the nasopharynx. However, intense 68 Ga-FAPI uptake is observed in the skull base (white arrow) along with normal FDG uptake, confirmed by magnetic resonance imaging. The TNM stage was upgraded from T2N2 (FDG-based) to T3N2 (FAPI-based). B. Imaging findings in a 56-year-old treatment-naive male patient with esophageal squamous cell carcinoma. 18 F-FDG PET/CT for tumor staging to decide the most proper treatment strategy. Maximum intensity projection (MIP) image 18 F-FDG PET/CT reveals an intense FDG-avid mass in the mid-esophagus, while the MIP image of 68 Ga-FAPI-04 shows intense uptake of FAPI in the primary tumor and paraesophageal lymph node. This FAPI-positive lymph node, suggestive of nodal metastasis, was later confirmed by histopathology. Tumor staging was upgraded to stage IIIB based on FAPI. Adapted with permission from [31], copyright 2021 Springer, and [58], copyright 2020 Springer.
Optimization of TNM staging via FAPI PET/CT results in improving oncologic management. For example, the course of clinical management was changed in 13 (30%) patients with sarcomas following 68 Ga-FAPI-PET in a previous investigation, including major changes (e.g., changes in therapeutic strategy) in seven (16%) patients [51]. However, 68 Ga-FAPI PET/CT imaging is not always superior to conventional SCI. For example, 68 Ga-FAPI-04 PET/CT detected fewer lesions as compared with MRI of the liver (85% [41/ 48] vs. 100% [48/48], P = 0.34) in a cohort of 32 patients with liver cancer [36]. Therefore, it is recommended that 68 Ga-FAPI PET/CT be used as a tool complementary to 18 68 Ga-FAPI PET/CT images in tumor radiotherapy for evaluating gross tumor volume (GTV) contouring [60]. Four thresholds (three-, five-, seven-and ten times that of the 68 Ga-FAPI uptake in normal tissue surrounding the tumor) were used to generate FAPI-GTV values (FAPI × 3, 5, 7 and 10) in 14 patients with head and neck cancers, which were subsequently compared with GTV values measured conventionally via CE-CT and MRI (CT-GTV). The area covered by FAPI-GTV (FAPI × 3 and 5) was significantly different compared to the area covered by CT-GTV, and the FAPI × 3 threshold was recommended as the best among the imaging modalities [60]. Similarly, 18 F-FAPI-74 × 3 and 68 Ga-FAPI-04 × 2 thresholds were considered optimal with respect to radiotherapy planning for lung cancer and locally recurrent pancreatic cancer [21,61]. Windisch et al. compared FAPI-GTV and MRI-GTV values in glioblastoma and found that the FAPI-GTV values for all thresholds were greater than the MRI-GTV values [62]. In addition to using background FAPI uptake, Zhao et al. used demarcations at 20%, 30%, and 40% of the SUVmax as thresholds for FAPI-GTV; their results demonstrated that 68 Ga-FAPI-04 × 40% of the SUVmax was ideal for reflecting the actual tumor volumes in esophageal cancer (Figure 4) [46]. Interestingly, PET/CT with different FAPI variants ( 68 Ga-FAPI-02, 68 Ga-FAPI-46, and 68 Ga-FAPI-74) acquired at three time points (10 min, 1 h, and 3 h) with a threshold of 25-35% of the SUVmax was used to delineate three FAPI-GTVs in adenoid cystic carcinomas, all of which were more accurate than CT-GTVs (based on CE-CT and CE-MRI); the 68 Ga-FAPI PET/CT images acquired 1 h post-injection were presumed to reflect the ideal time point for contouring [63]. Without implementing the FAPI threshold mentioned above, Koerber et al. and Ristau et al. reported that 68 Ga-FAPI-04/46 PET/CT improved target volume delineation in 6/6 anal canal carcinoma patients and 6/7 esophageal cancer patients [59,64].
The published data on the impact of FAPI PET/CT on radiotherapy are shown in Table 1. However, differing FAPI variants and thresholds for various types of tumors may be obstacles to the widespread use of FAPI PET/CT in GTV contouring. A well-designed prospective study with a large patient population is warranted to evaluate the overall survival benefit from FAPI-derived GTVs compared with GTVs derived from SCI in a heterogeneous grouping of cancers and imaging modalities.
For example, in a study of 22 cycles of 177 Lu-FAP-2286 administered to 11 patients with diverse adenocarcinomas (mean injected activity, 5.8 GBq), grade 3 adverse events occurred in three patients, and no grade 4/5 adverse events occurred [79]. Ferdinandus et al. reported 13 cycles of 90 Y-FAPI-46 administered to nine patients (with a mean injected activity of 3.8 GBq for the first cycle and a mean injected activity of 7.4 GBq for any subsequent cycle), with new grade 3/4 adverse events occurring in four patients [80]. In contrast, only one patient had a new grade 3 adverse event in a study of 36 cycles of 177 Lu-FAPI-46 administered to 18 patients (median injected activity, 3.7 GBq) [81]. The measured mean absorbed dose of 177 Lu-FAPI-04 was 0.37 Gy/GBq in tumor lesions [82], much lower than that of 90 Y-FAPI- 46 (median, 1.28 Gy/GBq) and 177 Lu-FAP-2286 (3.00 Gy/GBq in bone metastases) [80,79]. Recently, another FAPI dimer, DOTA(SA.FAPi) 2 , was developed and synthesized by Ballal et al. ( Figure  5D) [83]. Radionuclide therapy with 177 Lu-DOTA(SA.FAPi) 2 was administered to 15 patients with radioiodine-refractory differentiated thyroid cancer (DTC; RR-DTC). The results of that study revealed that the numbers of patients with complete response, partial response, and stable disease were 0, 4, and 3, respectively. None of the patients experienced grade 3/4 hematological, renal, or hepato-toxicity.
However, most of the studies mentioned above showed mixed responses to FAP-targeted radionuclide therapy because of the different tumors and patient conditions evaluated ( Table 2). It should be noted that the latest studies were mainly aimed at evaluating the feasibility and safety of FAP-targeted radionuclide therapy; the number of patients enrolled in these studies was very limited, and the patient cohorts were heterogeneous. In addition, most patients received FAP-targeted radionuclide therapy as the last line of treatment with poor performance status. Although the tumor half-life of FAP-2286 (average of 44 h for bone and 32 h for single liver metastases) is prolonged compared to FAPI-02/04, it is still shorter than the tumor half-life of PSMA [79,84].  PSMA-targeted radionuclide therapy is reportedly very effective with beta-radionuclides such as 177 Lu. Moreover, 177 Lu-PSMA directly targets cancer cells in radiotherapy, while the ionizing radiation of radiolabeled FAPIs mainly kills CAFs and indirectly kills cancer cells adjacent to CAFs via crossfire effects. These reasons may partially explain the difference in treatment response to targeted radionuclide therapy between PSMA and FAPI. Therefore, further research to enhance the therapeutic efficacy of FAP-targeted radionuclide is of great importance, including optimizing the chemical structure of the FAPI vector (e.g., multimerization and chemical conjugation with albumin binder), shortening the time interval between treatments, increasing the administered dose of therapeutic radionuclide, and combination treatments with other types of treatment (e.g., immunotherapy, external-beam radiotherapy, and molecular targeted therapy).
It has been reported that a tumor size of 1-2 mm requires the formation of stroma to support the tumor [57]. Thus, radionuclide therapy targeting FAP may be highly effective for treating advanced cancer patients with widespread metastases. Moreover, in an autochthonous model of pancreatic ductal adenocarcinoma, depleting FAP-positive CAFs induced T-cell accumulation in cancer cells and synergistically enhanced anti-tumor effects within PD-L1 immunotherapy [85]. Therefore, exploring optimal combination therapies with radionuclide therapy targeting FAP, especially with respect to immunotherapy, is warranted in future research. It must be noted that FAP is overexpressed in various epithelial cancers and is also expressed in many non-oncological diseases [23]. In a cohort of 91 patients, 81.3% of the presenting patients had non-tumor FAPI uptake, including degenerative lesions and physiological uptake in normal salivary glands, mammary glands, and the uterus [86]. In order to select patients who are most likely to benefit from this therapeutic regimen, careful pre-therapeutic evaluation with 68 Ga-FAPI PET/CT is recommended prior to FAP-targeted radionuclide therapy.

Conclusion
FAPI variants labeled with 68 Ga or 18 F have shown impressive results in a broad spectrum of cancers. Well-designed clinical trials with large patient populations are needed to define the role of this diagnostic agent, as 18 F-FDG is the dominant tracer in clinical oncology at present. In addition to CAFs, intense FAP expression is also related to fibrosis, arthritis, atherosclerosis, and autoimmune diseases. Thus, FAPI uptake in non-malignant diseases must be carefully identified. Regarding FAP-targeted radionuclide therapy, one direction for future research is improving the pharmacokinetic properties of tracers via chemical modification. The other potential direction is to explore optimal combination therapies (e.g., combining with external-beam radiotherapy, chemotherapy, and immunotherapy) to synergistically enhance antitumor efficacy. Overall, FAPI-based imaging and therapy of cancer have been a highly vibrant research field over the past few years. We look forward to future studies and rapid translation of the most promising FAPI ligands into the clinical arena to benefit patients with various types of cancer.