Data were collected between December 2022 and May 2023. Inclusion criteria for the present study were as follows: i) Patients with pathologically confirmed primary GI cancer (diagnosed via histopathological examination or narrow-band imaging magnification endoscopy); ii) patients with clinical suspicion of recurrence or metastasis supported either by abnormal follow-up CT/MRI findings or other clinical/radiological evidence; iii) patients with atypical ¹⁸F-FDG PET findings with either low ¹⁸F-FDG uptake in lesions despite radiological/clinical evidence confirming recurrence/metastasis or high ¹⁸F-FDG uptake in lesions despite radiological/clinical evidence confirming benign or inflammatory nature; and iv) patients who provided written informed consent for ⁶⁸Ga-FAPI-04 PET/MRI examination.

All patients were recruited from the First Affiliated Hospital of Anhui Medical University (Hefei, China). Tumor classification was performed according to the World Health Organization Classification of Tumors (23). Exclusion criteria were as follows: i) Failure to provide written informed consent; ii) inability to maintain a supine position for the full duration of the imaging procedure; iii) confirmed claustrophobia (unable to tolerate closed imaging equipment); and iv) concurrent diagnosis of other malignant tumors.

Patients for whom MRI examination was contraindicated (for example, due to implanted pacemakers, metallic intrauterine devices or ferromagnetic hip prostheses) underwent ⁶⁸Ga-FAPI-04 PET/CT instead of PET/MRI. A flowchart for the patient selection process is depicted in Fig. 1. The present study was performed in line with the principles of the Declaration of Helsinki. Ethics approval for the present study was granted by the Ethics Committee of Anhui Medical University (Hefei, China; date, May 9th 2022; approval no. PJ2022-05-09).

FAPI-04 was provided by Jiangsu Huayi Nuclear Medicine Co., Ltd., with a net weight of 872.91 g. The labeling of FAPI with ⁶⁸Ga was performed using methods previously described by Lindner et al (24) and Chen et al (25), with minor modifications to suit the characteristics of the ⁶⁸Germanium (⁶⁸Ge)/⁶⁸Ga generator (New Radiomedicine Technology Co., Ltd.) and the equipment available at the present study institution (Fig. S1). The ⁶⁸Ge/⁶⁸Ga generator was eluted with 0.1 M HCl to obtain 3.0 ml of ⁶⁸GaCl₃ solution. The FAPI-04 precursor (20 µg) was then combined with 375 µl of 1.25 M sodium acetate (pH 4.0), followed by the addition of 3.0 ml of the ⁶⁸GaCl₃ solution previously prepared. The reaction was conducted in a thermostat at 95°C for 10 min. Afterward, the mixture was passed through an activated Sep-Pak^{® 18}C column (Xintuo Trading Co., Ltd.) and washed with 0.8 ml of 80% ethanol. The eluate was then filtered through a 0.2-µm microporous membrane (Pall Corporation) and collected into a sterile, autoclaved vial. The ¹⁸F-FDG synthesis was performed by Yantai Dongcheng Pharmaceutical Group Co., Ltd. All products underwent comprehensive quality control, including radiochemical purity, specific activity and sterility tests, and met the required standards before clinical use.

The radiochemical purity of the two radionuclides was assessed using a high-performance liquid chromatograph (HPLC) [model LC-16; Shimadzu Instruments (Suzhou) Co., Ltd.]. The HPLC system conditions were as follows: Voltage, 220±22 V; frequency, ~50±0.5 Hz; power, 150 VA; injection volume, 25 µl; mobile phase A: deionized water containing 0.1% trifluoroacetic acid (TFA), mobile phase B: acetonitrile containing 0.1% TFA; flow rate, 1.0 ml/min; absorption wavelength, 320 nm. The column used was Shim-pack VP-ODS (C¹⁸) with a size of 4.6×150 mm and a particle size of 5 µm, manufactured by Shimadzu Instruments (Suzhou) Co., Ltd. The radiochemical purity of ⁶⁸Ga and ¹⁸F was calculated using the area normalization method. The detection conditions of ⁶⁸Ga and ¹⁸F radio-HPLC are shown in Table SI.

The radiochemical purity of ⁶⁸Ga-FAPI-04 was assessed by radio-HPLC within 10 min. The chromatogram displays a single, sharp peak at 3.1–3.2 min, with a signal intensity of 104.7–104.8 mV, while the ¹⁸F-FDG peak appears at 2.5–2.6 min, with a signal intensity of 113.4–113.5 mV. This well-defined peak indicates the presence of a highly pure compound with no detectable impurities or by-products, as evidenced by the stable baseline throughout the chromatogram. The absence of additional peaks confirms the effective separation and successful synthesis of the target radiopharmaceutical Fig. S1.

PET-MRI imaging was performed using a GE Healthcare scanner (SIGNA™; GE HealthCare) equipped with a dedicated molecular MR coil. ⁶⁸Ga-FAPI-04 was labeled on-site and used immediately after labeling. The intravenous injection dose was 1.85–3.7 MBq/kg. After 35 min of injection, respiratory gating imaging was initiated. The field of view covered from the skull to the upper part of the thighs. A step-and-shoot acquisition mode was adopted, with five bed positions being acquired, 4–6 min for each bed position and a total scanning duration of ~60 min. Multi-sequence nuclear MRI included T1-weighted imaging (WI), T2-WI and diffusion-WI. Breath-holding was required during the liver acceleration volume acquisition scan. PET/MRI reconstruction was performed using ordered subsets expectation maximization (OSEM) + time-of-flight (TOF) + point spread function (PSF), with two iterations, 28 subsets, a 440×440 matrix and a 5-mm filter.

¹⁸F-FDG PET/CT examination was performed using a Biograph Vision 450 PET-CT scanner (Siemens Healthineers). The dose of ¹⁸F-FDG was 3.7–5.55 MBq/kg. PET/CT acquisition started 60±20 min after intravenous injection. A CT scan was performed with a table speed of 22.5 mm/sec, followed by a PET scan conducted using a continuous bed movement acquisition mode, with 2–4 min of acquisition for each bed position and a range from the head to the upper part of the thighs. The OSEM algorithm was used for image reconstruction and TOF correction, and PSF modeling were performed. A total of five iterations, four subsets, a 440×440 reconstruction matrix and a 5-mm filter were adopted.

Two board-certified nuclear medicine specialists and a nuclear medicine postgraduate student with a medical licensure performed blinded image interpretation, with access to neither pathological findings nor clinical examination data to ensure the objectivity of the analysis. The images were analyzed using the multiple myeloma (MM) oncology-specific software within the Syngo.via VB30 system (2020; Siemens Healthineers). In cases of assessment discrepancies, a third senior nuclear medicine specialist was consulted to reach a final consensus through joint discussion, minimizing subjective bias and ensuring accuracy and consistency in data interpretation.

Anatomical regions evaluated included the primary tumor, anastomotic sites, abdominal incisions, lymph nodes (cervical, thoracic, abdominal and pelvic regions) and distant metastases (for example, brain, bone, liver, spleen, peritoneum, ovary and kidney). For lymph node classification, the cervical region was defined as the area below the line connecting the inferior margin of the mandible, posterior margin of the mandibular process, mastoid process and external occipital protuberance, and the area above the line connecting the suprasternal notch, clavicle, acromion and C7 spinous process. The thoracic region comprised the area within the thoracic cage and diaphragm, with its upper boundary being the thoracic inlet connecting to the cervical region. The abdominal region was defined as the space between the pelvic inlet and diaphragm, while the pelvic region referred to the area within the pelvic girdle.

Lesions with metabolic activity markedly higher compared with adjacent healthy tissues were classified as PET⁺ and potentially malignant lesions. The criteria for final confirmation of malignant lesions included the following: i) Lesions with malignant biopsy pathology results; ii) abnormal GI vascular patterns observed via standardized narrow-band imaging magnification endoscopy, consistent with histopathology if involving endoscopic biopsy; iii) lesion progression observed using CT or MRI during follow-up; and iv) lesion shrinkage after antitumor therapy. Nuclear medicine specialists qualitatively compared PET⁺ lesions detected using ¹⁸F-FDG PET and ⁶⁸Ga-FAPI-04 PET with the aforementioned final confirmed malignant lesions to determine the diagnostic accuracy of the two tracers.

In addition, a semi-quantitative comparison of the two radionuclides was also performed. Regions of interest were delineated on the cross-sectional images around the PET⁺ potentially malignant lesions for semi-quantitative evaluation. The maximum standardized uptake value (SUV_max) was automatically determined using the MM Oncology Syngo software. Due to the differences in the imaging principles, device performance and reconstruction algorithms of PET/CT and PET/MRI, the SUV_max values measured using the two devices could not be directly used for quantitative comparison. Therefore, the tumor-to-blood-pool ratio (TBR_blood_pool) value was used to enhance comparability. The calculation was performed according to a previous study (26) using the following formula: TBR_blood_pool (lesion)=SUV_max (lesion)/mean SUV_max (descending aorta).

For patients with digestive tract tumors detected using ⁶⁸Ga-FAPI-04 PET or ¹⁸F-FDG PET, a board-certified oncologist staged them according to the eighth edition of the American Joint Committee on Cancer guidelines (27). Clinical management and treatment recommendations were put forward based on the staging results. The differences in the staging of ⁶⁸Ga-FAPI-04 PET/MRI targeted molecular imaging and their impacts on tumor treatment were recorded. The changes in treatment methods were rated based on how much the imaging results influenced clinical decisions. If the modification was made to the treatment plan that the clinician had already anticipated, it was recorded as a minor change. If the FAPI PET results led to a change in the type of treatment or the treatment intent, it was classified as a major change.

All statistical analyses were performed using SPSS (version 26.0; IBM Corp.). To determine the diagnostic accuracy of ⁶⁸Ga-FAPI-04 PET and ¹⁸F-FDG PET, the present analysis evaluated the sensitivity, specificity, accuracy, positive predictive value (PPV) and negative predictive value (NPV) for both imaging modalities, with statistical comparisons conducted using McNemar's test. True-positive (TP) was defined as lesions with significantly elevated SUV_max on FDG or FAPI imaging, confirmed as malignant by pathology or follow-up. True-negative (TN) refers to lesions with no significant SUV_max elevation and verified as disease-free using the gold standard diagnostic method (pathological biopsy) or long-term follow-up. False-positive (FP) denotes focal SUV_max elevation without malignant findings on histopathology or imaging follow-up. False-negative (FN) indicates no significant SUV_max abnormality but evidence of malignancy at corresponding sites per gold standard diagnostic method or follow-up. PPV and NPV were calculated as PPV=TP/(TP + FP) and NPV=TN/(TN + FN), following the method used by Hill et al (28). Compared with ¹⁸F-FDG, a significant elevation in TP and TN for ⁶⁸Ga-FAPI-04 was defined as a statistically significant increase, with P<0.05 considered to indicate a statistically significant difference.

For the semi-quantitative analysis of ⁶⁸Ga-FAPI-04 and ¹⁸F-FDG PET imaging (including TBR_blood_pool), the Shapiro-Wilk test was performed for all subgroups with a sample size of >10. The test results confirmed that the tracer uptake data of target lesions conformed to a normal distribution, and the data were expressed as the mean ± standard deviation. Since all data in this study were obtained from the same subjects who underwent both examinations, the measurements were paired. Therefore, the paired t-test was used to compare the differences in tracer uptake of target lesions between the two imaging modalities and assess the statistical significance of such differences.

For subgroups with a sample size of ≤10, including primary lesions, anastomotic recurrence, abdominal wall recurrence, preoperative metastatic lymph nodes, cervical, thoracic and pelvic metastatic lymph nodes, as well as liver, kidney, peritoneal, bone and splenic metastases, the normality of the data could not be validly verified. The semi-quantitative analysis results of ⁶⁸Ga-FAPI-04 and ¹⁸F-FDG PET imaging for these subgroups were expressed as median (lower quartile-upper quartile). The Wilcoxon signed-rank test was adopted for non-parametric analysis to compare the differences in tracer uptake of target lesions between the two imaging modalities and further evaluate the statistical significance of the differences.

Fisher's exact test was used to verify the associations between different pathological subtypes and the changes in tumor staging and treatment decisions indicated by ⁶⁸Ga-FAPI-04 PET imaging. Simple ratio calculations were employed to determine the number of patients with stage changes and the impact rate on subsequent treatment due to ⁶⁸Ga-FAPI-04 PET, demonstrating its effect on staging and therapy planning. Two-tailed P<0.05 was considered to indicate a statistically significant difference.

The use of ⁶⁸Ga-FAPI-04 PET/MRI is associated with high equipment and drug costs and whether such high costs can bring tangible diagnostic value to patients is a key concern in clinical applications. Therefore, based on data from 22 patients, the present study conducted a preliminary comparison of the hospital operating costs and patient-derived medical benefits between ⁶⁸Ga-FAPI-04 PET/MRI and ¹⁸F-FDG PET/CT.

The hospital operating costs were defined as experiment-related hospital expenditures (rounded to the nearest whole number), which included only medication, equipment and labor costs. The direct costs of ⁶⁸Ga-FAPI-04 PET/MRI covered the consumption of ⁶⁸Ge/⁶⁸Ga generators, the cost of FAPI-04 precursor drugs, the depreciation of PET/MRI equipment and the salary of the technicians. For ¹⁸F-FDG PET/CT, the costs included the direct purchase cost of ¹⁸F-FDG, the depreciation of PET/CT equipment and the salary of the technicians.

Since ⁶⁸Ga-FAPI-04 PET/MRI is part of a clinical trial with no requirements for report writing, miscellaneous clinical expenses, such as fees paid to physicians for report preparation, hospital utility consumption, equipment maintenance, consumables for report printing and monitor repair, were not included in the calculation.

The present study included 22 participants, 12 men and 10 women, with a median age of 62 years (range, 51–77 years). The demographic and clinical characteristics of the patients are summarized in Table I. Specifically, there were 7 (31.8%) patients with esophageal cancer, 7 (31.8%) with gastric cancer, 5 (22.7%) with colon cancer and 3 (13.6%) with rectal cancer. Furthermore, 20 (90.9%) patients received treatments such as surgery, chemotherapy, radiotherapy or immunotherapy.

The present study aimed to explore the diagnostic value of ⁶⁸Ga-FAPI-04 PET/MRI and ¹⁸F-FDG PET/CT in the detection of recurrent and metastatic diseases in patients with GI cancer. A comparison of partial maximum intensity projection images obtained from representative ¹⁸F-FDG and ⁶⁸Ga-FAPI-04 scans is presented in Fig. 2.

The lesion-based analysis demonstrated that the sensitivity for diagnosing primary tumors was 50.0% (3/6) for ¹⁸F-FDG PET/CT and 83.3% (5/6) for ⁶⁸Ga-FAPI-04 PET/MRI (Table II). For cancer types with low ¹⁸F-FDG uptake, such as mucinous cancer (MC) and GI stromal tumors (GIST), ¹⁸F-FDG PET/CT generated more FN results.

By contrast, the lesions on ⁶⁸Ga-FAPI-04 PET/MRI were more clearly defined their semi-quantitative assessment was also significantly improved compared with that of ¹⁸F-FDG PET/CT imaging [TBR_blood_pool, 6.13 (5.71–7.57) vs. 2.62 (1.82–3.03); P=0.004; Fig. 3F]. This difference was even more pronounced (TBR_blood_pool, 5.82 vs. 1.12) in patients with gastric GIST (Fig. 3A).

Anastomotic stroma and abdominal wall incisions were evaluated using two radionuclides, ⁶⁸Ga-FAPI-04 and ¹⁸F-FDG. Anastomotic sites and abdominal wall incisions from 17 patients were included in the evaluation, with 3 patients presenting with recurrence and metastasis. While ¹⁸F-FDG PET/CT failed to detect these recurrences, ⁶⁸Ga-FAPI-04 PET/MRI identified two of them (Fig. 3B). ⁶⁸Ga-FAPI-04 improved the diagnostic accuracy to 66.67% for anastomotic recurrences that remained undetectable using ¹⁸F-FDG (Table II).

Although it is well known that ⁶⁸Ga-FAPI-04 is likely to be taken up due to the fibrous remodeling at the anastomotic healing site, thus affecting the examination results (29), ⁶⁸Ga-FAPI-04 still exhibits certain advantages compared with ¹⁸F-FDG in detecting anastomotic recurrence of pathological types with low ¹⁸F-FDG metabolism, such as mucinous adenocarcinoma and poorly differentiated adenocarcinoma (PDA).

The tracer uptake [TBR_blood_pool, 7.79 (5.97–8.58)] at the recurrent anastomotic sites of two patients and the metastatic abdominal wall incision of 1 patient, as shown by ⁶⁸Ga-FAPI-04 PET, was markedly higher compared with that of ¹⁸F-FDG PET [TBR_blood_pool, 2.00 (1.83–1.92)]. ¹⁸F-FDG had almost no uptake at the recurrent anastomotic sites of the 2 patients and had a little uptake at the metastatic abdominal wall incision (Table III).

The imaging effect of ⁶⁸Ga-FAPI-04 PET was also enhanced, allowing for a clearer scope delineation of the lesions (Fig. 3G). However, there was no comparable difference in the accuracy and semi-quantitative parameters of the two radionuclides in detecting anastomotic recurrence and abdominal wall incisions. This may be due to the low anastomotic recurrence rate of the patients and the small sample size in the present study (P>0.05).

Notably, a patient with colon cancer (patient 21) exhibited mild ¹⁸F-FDG uptake (TBR_blood_pool, 2.28) at the abdominal wall incision 9 months after surgery. Ordinarily, this would not be considered a metastasis at the abdominal wall incision due to the potential for physiological uptake in the surgical area. Nevertheless, the uptake of ⁶⁸Ga-FAPI-04 was higher (TBR_blood_pool, 6.48), and a low signal on T1WI and a high signal on T2WI were observed on PET/MRI (Fig. 3C). Considering factors such as fibrosis, this specific case was not defined as metastasis due to the high uptake of ⁶⁸Ga-FAPI-04 at that time. The pathological results obtained after a 2-month follow-up confirmed that the abnormal uptake and signals in the abdominal wall indicated the presence of metastasis.

Among the included patients, 45.45% (10/22) were diagnosed with lymph node involvement. Using pathological sections or subsequent imaging follow-up, 91 TP lymph nodes were identified and used as the control standard. The TBR_blood_pool of ⁶⁸Ga-FAPI-04 used for imaging of metastatic lymph nodes was notably increased compared with that of ¹⁸F-FDG and the imaging effect was notable (Fig. 4E-G). There was no significant difference in the uptake of ¹⁸F-FDG and ⁶⁸Ga-FAPI-04 in inflammatory lymph nodes (Fig. 5E-G).

When evaluating imaging modalities for the detection of untreated lymph node metastases, ¹⁸F-FDG demonstrated limitations. Among patients followed up for 1 year after ¹⁸F-FDG imaging, non-metastatic lymph nodes were misdiagnosed as metastatic, leading to overestimation of lymph node involvement compared with histopathological findings (the gold standard). Accordingly, the diagnostic accuracy of ¹⁸F-FDG was lower than that of ⁶⁸Ga-FAPI-04 (71.79 vs 88.46%; P<0.001) (Table II). By contrast, ⁶⁸Ga-FAPI-04 exhibited significantly improved diagnostic performance, with higher specificity (66.67 vs. 36.11%; P=0.009) and accuracy (73.33 vs. 48.89%; P=0.02) compared with ¹⁸F-FDG (Table II). Semi-quantitatively, ⁶⁸Ga-FAPI-04 exhibited a significantly higher TBR_blood_pool [6.07 (5.77–6.37) vs. 1.94 (1.92–1.97); P<0.001; Fig. 4E].

Among treated patients, ¹⁸F-FDG failed to detect 9 cases of lymph node metastases. ⁶⁸Ga-FAPI-04 exhibited significantly higher diagnostic sensitivity (100 vs. 89.02%; P=0.002) accuracy (93.69 vs. 80.18%; P=0.003) and NPV (100.00 vs. 64.00%; P=0.002) compared with ¹⁸F-FDG (Table II). The TBR_blood_pool of ⁶⁸Ga-FAPI-04 (9.13±2.93 vs. 4.01±1.72; P<0.001) (Table III) was also significantly higher, indicating stronger imaging and diagnostic capabilities (Fig. 4F) (17,27).

The present study evaluated the detection of metastatic (Fig. 4) and non-metastatic lymph nodes in four regions (Fig. 5). The qualitative comparison results of the diagnosis of lymph nodes in these four regions are shown in Table IV.

When evaluating abdominal metastatic lymph nodes, ⁶⁸Ga-FAPI-04 demonstrated significantly higher diagnostic sensitivity (100 vs. 87.5%; P=0.002), accuracy (97.56 vs. 86.59%; P=0.02) and NPV (100 vs. 47.06%; P=0.02) compared with ¹⁸F-FDG.

In semi-quantitative parameter comparisons, ⁶⁸Ga-FAPI-04 demonstrated significantly higher uptake compared with ¹⁸F-FDG in metastatic lymph nodes across all anatomical regions: Cervical [7 nodes, TBR_blood_pool, 16.23 (14.47–16.95) vs. 7.25 (5.911–8.66); P<0.001], thoracic [4 nodes, 10.00 (8.06–12.00) vs. 4.87 (3.47–6.81); P<0.001], abdominal (72 nodes, 8.52±2.29 vs. 3.67±1.33; P<0.001) and pelvic [6 nodes, 4.11 (3.86–4.49) vs. 1.55 (1.41–1.67); P<0.001] (Fig. 4A-D and H-K).

For non-metastatic (inflammatory/benign) lymph nodes, ¹⁸F-FDG frequently exhibited FP uptake in thoracic lymph nodes due to inflammation, whereas ⁶⁸Ga-FAPI-04 achieved significantly higher diagnostic accuracy in thoracic lymph nodes (72.55 vs. 7.84%; P<0.001). In semi-quantitative analysis of benign lymph nodes, ⁶⁸Ga-FAPI-04 exhibited significantly lower TBR_blood_pool in inflammatory thoracic lymph nodes compared with ¹⁸F-FDG (1.88±0.87 vs. 2.71±1.14; P<0.001) (Fig. 5B and I), with no significant differences in uptake between the two tracers observed in cervical, abdominal or pelvic benign lymph nodes.

In total, 12 patients had distant organ metastases in various sites, including the liver, kidney, pancreatic head, lung, peritoneum, bone, brain, ovary and spleen. Notably, 1 patient had simultaneous liver and spleen metastases. ¹⁸F-FDG missed one peritoneal metastasis, one bone metastasis, one pancreatic head with surrounding peritoneal involvement, three splenic metastases, three liver metastases and one renal metastasis, whereas ⁶⁸Ga-FAPI-04 clearly detected all these metastatic lesions except for one renal metastasis and one small liver metastasis. However, for lung metastases, ¹⁸F-FDG PET/CT exhibited improved imaging performance compared with ⁶⁸Ga-FAPI-04 PET/MRI.

In detecting distant organ metastases in GI tumors with atypical metabolism, ⁶⁸Ga-FAPI-04 demonstrated significantly higher sensitivity (92.31 vs. 61.54%; P=0.008) and accuracy (88.89 vs. 62.96%; P=0.03) compared with ¹⁸F-FDG (Table II). ⁶⁸Ga-FAPI-04 also exhibited significantly higher uptake in distant metastases (TBR_blood_pool, 6.87±4.16 vs. 3.31±2.19; P<0.001) as well as in the liver, peritoneum, bone and spleen. A comparison of all TBR_blood_pool values is listed in Table III.

Compared with ¹⁸F-FDG PET, ⁶⁸Ga-FAPI-04 PET induced staging changes in 36.4% of cases (8/22). Among these, 31.8% (7/22) had an upstaging. Specifically, 57.1% (4/7) were whole or partial PDA, 28.6% (2/7) were whole or partial MC and 14.3% (1/7) were adenocarcinoma with unclear differentiation. In 4.5% (1/22) of cases, a downstaging was noted, specifically in a case of metastatic duodenal adenocarcinoma.

⁶⁸Ga-FAPI-04 PET detected M stage upgrading in 22.7% of cases (5/22) and N stage upgrading in 9.1% (2/22). The comparison between ⁶⁸Ga-FAPI-04 and ¹⁸F-FDG imaging revealed statistically significant differences in the disease stage (P=0.004). However, there was no significant difference in N stage staging between the two imaging modalities. These data were derived from Table V.

The present study also conducted an analysis between biopsy results and staging changes indicated by ⁶⁸Ga-FAPI-04 PET. Fisher's exact test was performed to evaluate the associations. The results demonstrated that mucinous (P=0.018), poorly differentiated (P=0.005), and adenocarcinoma (P=0.004) pathological subtypes were all significantly associated with staging changes indicated by ⁶⁸Ga-FAPI-04 PET (Table VI). The results demonstrated that the coincidence rate between pathological findings and the diagnostic results of lesion staging changes detected by ⁶⁸Ga-FAPI-04 PET was 62.5% (5/8), indicating a high degree of consistency between its staging judgments and the pathological gold standard. The nature of the remaining lesions was confirmed using imaging follow-up.

According to the subgroup analysis, among these lesions whose staging was changed based on ⁶⁸Ga-FAPI-04 PET detection, 62.5% (5/8) were poorly differentiated adenocarcinoma or MC. This suggested that ¹⁸F-FDG PET uptake is atypical in poorly differentiated adenocarcinoma or MC and these subtypes are more likely to achieve medical benefits in staging via ⁶⁸Ga-FAPI-04 PET detection.

The influence of ⁶⁸Ga-FAPI-04 PET on staging is summarized in Table V. The analysis presented in the current study focuses on the differences in N and M staging, as well as overall staging.

⁶⁸Ga-FAPI-04 influenced the management of subsequent oncological treatments in 27.3% of cases (6/22). Among them, 66.7% (4/6) experienced major treatment changes and 33.3% (2/6) had minor changes (these data were calculated from Table V). Overall, patients diagnosed with PDC and MC had the highest proportion of treatment changes, accounting for 83.3% (5/6).

The following patient-level changes were documented in detail: ⁶⁸Ga-FAPI-04 successfully detected early local recurrence of PDA and MC (Fig. 3B) and pancreaticoduodenal metastases (Fig. 6C). These findings prompted notable modifications to the therapeutic strategy at the initiation of systemic chemotherapy. Notably, the recurrent metastases were confirmed via gastroscopic follow-up 1 month after initial assessment and the patient was followed up for a total of 1 year.

In patients with metastatic moderately differentiated esophageal and rectal cancer (patients 19 and 22), ⁶⁸Ga-FAPI-04 PET/MRI demonstrated increased involvement of the bones, liver and spleen, while ¹⁸F-FDG PET/CT failed to detect evident tumors. This discrepancy led to the recommendation of additional systemic therapy, which was regarded as a minor adjustment to the existing treatment plan.

The present study also conducted an analysis between biopsy results and changes in treatment decisions induced by ⁶⁸Ga-FAPI-04 PET. Fisher's exact test was subsequently performed for statistical analysis, and the results revealed that pathological features, including mucinous subtype (P=0.01), poorly differentiated histology (P<0.001) and adenocarcinoma (P<0.001), were all associated with changes in treatment decisions indicated by ⁶⁸Ga-FAPI-04 PET (Table VI). The results demonstrated that the coincidence rate between pathological findings and the diagnostic results of ⁶⁸Ga-FAPI-04 PET for lesions leading to treatment decision changes was 83.3% (5/6), indicating a high degree of consistency between its treatment-related judgments and the pathological gold standard. The nature of the remaining lesion was determined based on follow-up.

According to the subgroup analysis, among these lesions for which treatment plans were adjusted based on ⁶⁸Ga-FAPI-04 PET detection, 83.3% (5/6) were poorly differentiated adenocarcinoma or MC. This suggested that ¹⁸FDG PET uptake is atypical in poorly differentiated adenocarcinoma or MC and these subtypes are more likely to achieve medical benefits in terms of treatment plan adjustment via ⁶⁸Ga-FAPI-04 PET detection.

Table V summarizes the subsequent impact of FAP-targeted molecular imaging on treatment decisions.

In PET/MRI, the multiple signal sequences of MRI can provide additional imaging parameters and offer high soft-tissue resolution, which aids in accurate diagnosis and confers added value to the detection of primary and metastatic lesions in abdominal and pelvic tumors.

In the present study, MRI identified a renal metastasis in patient 4, who was diagnosed with differentiated adenocarcinoma of the ascending colon. However, this metastatic lesion was not detected by either of the two radiotracers.

Regarding the potential cost benefit of ⁶⁸Ga-FAPI-04 PET/MRI compared with ¹⁸F-FDG PET/CT in diagnosing atypical GI tumors, calculations show that the hospital cost of ⁶⁸Ga-FAPI-04 PET/MRI was higher (RMB 98,583 for 22 patients; RMB 4,481 per person) than that of ¹⁸F-FDG PET/CT (RMB 24,347 for 22 patients; RMB 1,107 per person), resulting in a total cost difference of RMB 98,583. However, for 6 patients whose treatment decisions were altered based on PET/MRI findings, the indirect benefits largely offset this cost difference, generating greater overall value.

In patients 5 and 20, ⁶⁸Ga-FAPI-04 PET/MRI detected anastomotic recurrences missed by both ¹⁸F-FDG and initial pathology, eliminating the need for subsequent gastroscopy follow-ups and pathological examinations. This saved ~RMB 2,200 per patient (for gastroscopy, anesthesia, pathology and staining), totaling RMB 4,400 for both. It also avoided one subsequent hospitalization for each patient (~RMB 8,000 per hospitalization) and one follow-up PET/CT scan (RMB 7,000 per scan), resulting in total savings of RMB 34,400.

In patients 16, 19, 21 and 22, ⁶⁸Ga-FAPI-04 PET/MRI identified additional metastases (pancreatic head, multiple bone lesions, abdominal wall nodules and liver/spleen metastases, respectively). This reduced the need for three subsequent CT follow-ups per patient (~RMB 500 per CT), saving RMB 1,500 per patient. It also avoided one hospitalization (RMB 8,000 each) and one PET/CT follow-up scan (RMB 7,000 for 1 patient), generating total savings of RMB 66,000 for these 4 patients.

For the 22 patients, the additional hospital cost of ⁶⁸Ga-FAPI-04 PET/MRI over ¹⁸F-FDG PET/CT was ~RMB 98,583, while the total benefits from altered management amounted to RMB 100,400, thus covering the cost difference. Additionally, patients saved considerable time and reduced the risk of disease progression.

These findings suggest that although ⁶⁸Ga-FAPI-04 PET/MRI has higher initial costs, its long-term economic benefits through optimized treatment decisions support its clinical adoption in cases where ¹⁸F-FDG results are inconclusive for GI tumors.