Patients characteristics

Between March 2021 and June 2023, 162 patients with suspected primary hepatobiliary malignancies underwent paired [18F]FDG and [18F]FAPI-04 PET/CT scans. Among them, 132 patients were excluded due to their final diagnoses confirmed by pathological examination: hepatocellular carcinoma (94), extrahepatic cholangiocarcinoma (10), hepatic metastases (12), and other benign liver lesions (16). Additionally, 7 patients with pathology-confirmed ICC were excluded due to anti-tumor treatment (2), the presence of another primary cancer (3), or recurrent ICC (2). As a result, a total of 23 patients with histological proven ICC (17 men and 6 women; median age, 61 [range, 39–71] years old) were finally enrolled. The flow chart was presented in Fig. 1. Among them, 26.1% (6 of 23) of patients were accompanied with obstructive cholangitis, and 56.5% (13 of 25) of patients were identified to had extrahepatic lesions (13 patients with lymph node metastasis and 6 patients with distant metastasis) (Table 1). According the eighth edition of the American Joint Committee on Cancer staging system, 8 patients were classified as clinical stage II, 9 patients as clinical stage III, and 6 patients as clinical stage IV.

Fig. 1

Study flowchart shows inclusion and exclusion criteria ICC = intrahepatic cholangiocarcinoma; 18F = fluorine 18; FAPI = fibroblast activation protein inhibitor; FDG = fluorodeoxyglucose; PET/CT = positron emission tomography/computed tomography

Table 1 Baseline patient characteristics of the included patientsInter-reader agreement

According to Cohen’s kappa (κ), there was almost perfect inter-reader agreement between the two independent readers for both [18F]FDG PET/CT (κ = 0.805, p < 0.001) and [18F]FAPI-04 PET/CT (κ = 0.863, p < 0.001).

[18F]FDG and [18F]FAPI-04 PET/CT in detection of intrahepatic lesions

In the final cohort of 23 patients, [18F]FAPI-04 PET/CT demonstrated greater sensitivity (100% vs. 87.0%) in identifying primary ICC tumors compared to [18F]FDG PET/CT, based on histological evaluations of biopsy or surgical specimens (Table 2). To assess the diagnostic accuracy of intrahepatic metastasis, a total of 111 suspected intrahepatic metastases (101 metastases, 8 hemangiomas and 2 inflammatory lesions with the final diagnosis) were analyzed, [18F]FDG PET/CT showed no or slight uptake in all 10 lesions, while the 2 inflammatory lesions exhibited increased [18F]FAPI-04 uptake causing 2 false-positive findings. The lesion-based sensitivity, specificity, and accuracy of [18F]FDG PET/CT were 76.2% (77/101), 100% (10/ 10), and 78.4% (87/111), respectively, and 85.1% (86/101), 80.0% (8/10), and 84.7% (94/111) for [18F]FAPI-04 PET/CT (Table 2).

Table 2 Diagnostic performances of [18F]FDG and [18F]FAPI-04 PET/CT for evaluation of ICC primary and metastatic lesions

Among the 23 patients, 6 (26.1%) presented with obstructive cholangitis, intense [18F]FAPI-04 uptake was observed throughout the liver. The hepatic activity of [18F]FAPI-04 in the liver parenchyma background of patients with obstructive cholangitis (SUVmean=2.7 ± 0.87) was significantly higher than that of patients without obstructive cholangitis (SUVmean=0.95 ± 0.33;P < 0.001). There was no significant difference in the hepatic activity of [18F]FDG in the liver parenchyma background of patients with and without obstructive cholangitis (SUVmean: 2.11 ± 0.40 and 2.14 ± 0.45; P = 0.815). Contrastingly, the TBR of [18F]FAPI-04 in patients without obstructive cholangitis was significantly higher than that in patients with obstructive cholangitis (P = 0.004), and the representative images are presented in Fig. 2 and Supplemental Fig. 1.

Fig. 2

A 70-year-old male patient (Patient No. 19) with intrahepatic cholangiocarcinoma (ICC) was confirmed by biopsy. Compared with [18F]FDG PET/CT (A), [18F]FAPI-04 PET/CT (B) showed significantly increased radiotracer uptake in the primary tumor (white arrow in axial images), yet displayed elevated hepatic [18F]FAPI-04 activity in the liver parenchyma background. There were several lymph nodes (dotted arrows), follow-up evaluations confirmed reactive. Pathological findings (C) indicate a heightened expression for FAP

Regarding the 17 ICC patients without obstructive cholangitis (17 primary tumors, and 70 intrahepatic metastases), [18F]FAPI-04 PET/CT visualized more primary tumors (100%, 17/17) than that of [18F]FDG PET/CT (82.4%, 14/17), as confirmed histopathologically. [18F]FAPI-04 PET/CT showed more intense uptake of the tracer by the primary tumors and had a higher TBR and a greater tumor burden than [18F]FDG PET/CT (SUVmax, 14.0 vs. 9.9, P = 0.004; median TBR, 17.9 vs. 4.6, P = 0.001; FTV vs. MTV, 103.0 vs. 61.7, P < 0.001; TLF vs. TLG, 922.3 vs. 395.5, P < 0.001. Table 3; Fig. 3). In the semiquantitative analysis of 70 intrahepatic metastases, [18F]FAPI-04 PET/CT showed higher activity (median SUVmax, 8.25 vs. 5.9, P < 0.001), clearer tumor delineation (median TBR, 6.0 vs. 3.1, P < 0.001) and greater tumor burden (median FTV vs. median MTV, 7.3 vs. 4.4, P = 0.005; median TLF vs. median TLG, 42.9 vs. 20.7, P = 0.004. Table 3) than [18F]FDG. For the 6 ICC patients with obstructive cholangitis (6 primary tumors, and 31 intrahepatic metastases), [18F]FAPI-04 and [18F]FDG PET/CT both visualized all the primary tumor (100%, 6/6), as confirmed histopathologically. [18F]FDG PET/CT showed more intense uptake of the tracer by the primary tumors, had a higher TBR than [18F]FAPI-04 PET/CT (median SUVmax, 19.8 vs. 13.7; median TBR, 11.9 vs. 8.1, but both without statistical significance; Table 3). In the semiquantitative analysis of 31 intrahepatic metastases, [18F]FDG PET/CT also showed higher activity (median SUVmax, 14.1 vs. 5.8, P < 0.001) and clearer tumor delineation (median TBR, 6.6 vs. 2.5, P < 0.001) than [18F]FAPI-04 PET/CT.

Table 3 Comparison of [18F]FDG PET/CT and [18F]FAPI-04 PET/CT for the intrahepatic and extrahepatic lesions of 23 patientsFig. 3

A 57-year-old male patient (Patient No. 4) with intrahepatic cholangiocarcinoma (ICC) was confirmed by postoperative pathology. [18F]FDG PET/CT (A) displayed no uptake in this lesion, conversely, [18F]FAPI-04 PET/CT (B) revealed intense uptake (SUVmax 13.7; TBR 11.4) in the hepatic segment IV/V lesion, as shown on both MIP images (large arrow) and axial images (small arrow). Pathological findings (C) indicate a heightened expression for FAP

[18F]FDG and [18F]FAPI-04 PET/CT for assessment of lymph node metastasis

According to the visual analysis for lymph node metastasis, a total of 121 suspected lymph nodes (88 metastatic lymph node lesions and 33 reactive lymph nodes with the final diagnosis) in 23 ICC patients were evaluated. [18F]FDG PET/CT imaging demonstrated increased uptake in 19 reactive lymph nodes with false-positive judgment for lymph node assessment. Conversely, the majority of reactive lymph nodes demonstrated low or absent radiotracer uptake on [18F]FAPI-04 PET/CT imaging, with only 6 false-positive findings. The sensitivity, specificity and accuracy of [18F]FDG PET/CT were slightly lower than [18F]FAPI-04 PET/CT for lymph node evaluation. (68.2% [60/88] vs. 85.2% [75/88]), P = 0.007; 42.4% [14/33] vs. 81.8% [27/33], P < 0.001; 61.2% [74/121] vs. 84.3% [102/121], P < 0.001. Table 2). The SUVmax (6.5 vs. 5.5, P = 0.042), TBR (5.4 vs. 3.9, P < 0.001), and tumor burden (median FTV vs. median MTV: 2.0 vs. 1.5, P = 0.026; median TLF vs. median TLG: 9.0 vs. 7.8, P = 0.024.) of [18F]FAPI-04 PET in lymph node metastasis were significantly higher and greater than that in [18F]FDG PET (Table 3).

[18F]FDG and [18F]FAPI-04 PET/CT in evaluation of distant metastasis

To assess the diagnostic accuracy of distant metastases, 134 suspicious metastatic lesions from 23 ICC patients were evaluated. Of these, 117 lesions were identified as metastatic lesions in 6 patients based on the reference standards, including 44 bone metastases, 64 peritoneal metastases, 4 pleura metastases, 4 lung metastases and 1 adrenal gland metastasis. [18F]FAPI-04 PET had greater sensitivity (96.6% [113/117] vs. 85.5% [100/117], P = 0.006) and accuracy (91.8% [123/134] vs. 82.8% [111/134], P = 0.028 ) than [18F]FDG PET. The specificity of PET using [18F]FAPI-04 was lower than that of [18F]FDG (58.8% [10/17] vs. 64.7% [11/17]) because more false-positive lesions were observed on [18F]FAPI-04 PET (including fracture [n = 2], osteofibrous dysplasia [n = 2], degenerative osteophyte [n = 3]), but it was not significantly different (P = 0.724).

Regarding the 44 bone metastases, [18F]FAPI-04 PET had a significant higher sensitivity and accuracy than [18F]FDG PET (100% [44/44] vs. 70.5% [31/44], P < 0.001; 87.7% [50/57] vs. 70.2% [40/57], P = 0.022) (Table 2) and [18F]FAPI-04 PET showed higher SUVmax, TBR and larger tumor burdens than [18F]FDG PET in bone metastases evaluation (median SUVmax: 9.7 vs. 5.25, P < 0.001; median TBR: 10.8 vs. 3.0, P < 0.001; median TLF vs. median TLG: 9.8 vs. 4.2, P < 0.001) (Table 3; Fig. 4). [18F]FDG PET/CT showed 4 false positive uptakes (1 fracture, 1 osteofibrous dysplasia, 1 degenerative osteophyte and 1 schmorl node). Thus, the specificity (46.2% [6/13] vs. 69.2% [9/13]) of [18F]FAPI-04 PET/CT was lower than that [18F]FDG PET/CT (Table 2), but it was not significantly different (P = 0.234).

Fig. 4

A 61-year-old male patient (Patient No. 16) with intrahepatic cholangiocarcinoma (ICC) was confirmed by biopsy. Compared with [18F]FDG PET/CT) (A), [18F]FAPI-04 PET/CT (B) demonstrated higher radiotracer uptake in the primary tumor ( SUVmax of 13.7, solid arrow), as well as in peritoneal and bone metastatic lesions (dotted arrows). Pathological findings (C) indicate a heightened expression for FAP

For detecting peritoneal metastases, the [18F]FDG PET/CT showed higher SUVmax than [18F]FAPI-04 PET/CT (median SUVmax: 5.45 vs. 4.45, P = 0.003), while the [18F]FAPI-04 PET/CT demonstrated a larger tumor burden (median FTV vs. median MTV: 1.6 vs. 1.3, P = 0.025) (Table 3; Fig. 5).

Fig. 5

A 61-year-old female patient (Patient No. 22) with ICC was confirmed by biopsy. [18F]FDG PET/CT (A) demonstrated higher radiotracer uptake in the primary tumor (solid arrow), intrahepatic subfoci, lymph node and peritoneal metastatic lesions (dotted arrows), compared with [18F]FAPI-04 PET/CT (B). The pleural metastases exhibited a comparable radiotracer uptake to FAPI PET/CT (white arrows). Pathological findings (C) indicate a heightened expression for FAP

Changes in staging and therapeutic management

In the initial assessment of 23 patients, [18F]FAPI-04 imaging detected primary ICC tumors in 3 patients with [18F]FDG-negative, confirmed by pathology. The patient received resection as early as possible since [18F]FAPI-04 detected the primary lesion. With more lymph node metastases revealed by [18F]FAPI-04 PET/CT than [18F]FDG PET/CT, the TNM staging was upgraded in 3 patients (all from II to III) (Table 4). According to the visual comparative system (Fig. 6), [18F]FAPI-04 PET imaging also demonstrated superior detection of primary tumors, lymph node and bone metastases compared to [18F]FDG PET imaging. As a result, instead of the previously planned surgical treatment for all, 3 patient received systemic chemotherapy, while one patient received palliative systemic treatment. However, [18F]FAPI-04 PET/CT also underestimated the TNM staging (from IV to III) in one patient due to adrenal gland metastasis detected by [18F]FDG, but the planning therapy was not change. Finally, TNM staging was upgraded in 6 patients (6/23, 26.1%) and downgraded in one patient, leading to changes in the planned therapy for 6 patients.

Table 4 Comparison of [18F]FDG PET/CT and [18F]FAPI-04 PET/CT-based TNM staging of 23 treatment-naive patients with ICCFig. 6

A visual comparative system is used to compare the performance of [18F]FDG and [18F]FAPI-04 PET in detecting primary tumors, intrahepatic, lymph nodes, lung, peritoneal, pleura, adrenal and bone metastases. LNM: lymph node metastases; Mets: metastases