Comparison of [68Ga]Ga-DOTA-FAPI-04 PET/CT with [18F]FDG PET/CT for Nodule Characterization and Staging in NSCLC Patients

Xin Zhou Peking University Cancer Hospital: Beijing Cancer Hospital Shuailiang Wang Peking University Cancer Hospital: Beijing Cancer Hospital Xiaoxia Xu Peking University Cancer Hospital: Beijing Cancer Hospital Xiangxi Meng Peking University Cancer Hospital: Beijing Cancer Hospital Huiyuan Zhang Peking University Cancer Hospital: Beijing Cancer Hospital Annan Zhang Peking University Cancer Hospital: Beijing Cancer Hospital Yufei Song Peking University Cancer Hospital: Beijing Cancer Hospital Hua Zhu Peking University Cancer Hospital: Beijing Cancer Hospital Zhi Yang Peking University Cancer Hospital: Beijing Cancer Hospital Nan Li (  rainbow6283@sina.com ) Peking University Cancer Hospital: Beijing Cancer Hospital https://orcid.org/0000-0001-8619-7550


Introduction
Cancer is associated with broblasts at all stages of disease progression including oncogenesis, proliferation and metastasis. The broblasts playing the subtle role in the regularization of tumor microenvironment have been recognized as cancer-associated broblasts (CAFs) [1][2][3] . The presence of CAFs is associated with various types of tumor entities, including the non-small cell lung cancer (NSCLC) [4] , in which the CAFs are commonly identi ed by their marker, broblast activation protein (FAP) [4][5][6][7][8] . FAP is a type II membrane-bound glycoprotein with dipeptidyl peptidase and endopeptidase activity, and is considered as a promising target for cancer theranostics [6,8] .
FAP-targeting molecules were recently developed as imaging and therapeutic agents based on the FAPspeci c inhibitor (FAPI) [9] . In previous studies, FAPI-04 showed the most favorable pharmacokinetics in vivo, and was selected for further characterization in different tumor entities [10,11] . A recent study demonstrated that among 28 types of tumor including NSCLC, high-quality images have been obtained with [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT [12] , con rming the high tumor uptake. Several systemic researches of prevalent cancers including glioblastomas, malignancies within the lower gastrointestinal tract, hepatic malignancies and head and neck cancers revealed the value of [ 68 Ga]Ga-DOTA-FAPI-04 in diagnosis, differential diagnosis and imaging-guided interventions [13][14][15] . According to previous studies, for NSCLC patients, [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT might obtain good lesion detection in brain with low cerebral uptake, which is regarded as limitation for [ 18 F]FDG PET/CT [12,16] . However, the impact of Pathological results and follow-up imaging results are adopted as the gold standard for lung nodules and lymph nodes (LNs) diagnosis. Pathological results are acquired from resection or aspiration biopsy. Follow-up imaging diagnosis criteria are set as follows: in ammatory nodules diagnosis must have completely resolution in the next six-month follow-up CT scans [17] ; for metastatic LNs diagnosis, signi cant responses (at least 30% decrease in long diameters) to anti-tumor therapy should be in followup CT scans [18] ; non-metastatic LNs diagnosis is de ned as LNs with no morphological changes while obvious responses of primary tumor (≥30% decrease of long diameter) exist to anti-tumor therapy.

Examination Procedures
Patients were intravenously injected with [ 68 Ga]Ga-DOTA-FAPI-04 (1.9-3.7 MBq/kg) and underwent a whole-body scan 40-60 min p.i.. Imaging was performed on a Biograph mCT Flow 64 PET/CT scanner (Siemens Healthcare). PET images acquired in 3-dimensional mode using FlowMotion (Siemens Healthcare) at a speed of 1.5 mm/s from the vertex to the upper thigh for each subject. PET images, attenuation corrected with low-dose CT images, were reconstructed using an ordered-subsets expectation maximization (OSEM) method with 200 × 200 matrix, 2 iterations, and 11 subsets. The size of the postreconstruction Gaussian lter was 5 mm. A high-resolution breath-holding CT scan was performed for each patient (120 kV, 146 mAs). CT images reconstructed using SAFIRE algorithm with 512 × 512 matrix and the slice thickness was 3-5 mm.
Patients' general information such as heart rate, respiratory rate, body temperature and blood pressure were measured before injection, during the scanning period, and every 6 hours after PET/CT scans for two days. All patients received the [ 18 F]FDG PET/CT scan rst, and followed by a [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT scan. Acquisition and reconstruction parameters of both scans remained consistent.

Imaging Analysis
Siemens workstation (Syngo.via VB20, MM Oncology) was used for image postprocessing. Two experienced nuclear medicine physicians blinded to the prior ndings reviewed all images independently and any discordant results were resolved by consensus.
Longest diameter, location and size of the solid component from high-resolution CT were recorded.
Nodules morphology, including lobulated, spicules of margin, pleural indentation, vacuole sign and calci cation were depicted as well. Classi cation of nodules are de ned as follows. Ground glass opacity (GGN): homogeneous opacities viewed using the lung window which cannot be viewed in mediastinal window; part-solid nodules (PSN), consisted of GGNs with a solid component both in the lung and mediastinal window; solid nodules (SN), solid component without GGN [19] . According to former researches [20,21] , SUV max ≥ 2.5 was set as the criterion for NSCLC diagnosis in [ 18 F]FDG PET, and the cutoff value for [ 68 Ga]Ga-DOTA-FAPI-04 PET SUV max was also set as 2.5 based on the consensus of the participating physicians.
LNs of diagnosed NSCLC patients characterized by short diameter ≥ 0.5 cm and SUV max of [ 18 F]FDG higher than the blood pool were considered as suspected metastatic LNs. Short and long axis dimension, density and calci cation status of LNs were also depicted based on CT images. SUV max and SUV mean of lung nodules, metastatic lesions and blood pool from both [ 18 F]FDG and [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT were obtained by software (Syngo.via). All LNs were classi ed into high, intermediate and low risks according to the risk categories based on the relationship of LNs location and lung nodules, which was speci ed in supplemental information [22] . The density of LNs was identi ed as high, iso and low density, compared with mediastinal soft tissue. SUV max ≥2.5 was also set as the criteria for metastatic LNs diagnosis in [ 18 F]FDG [23] . For multiple metastasis patients, the highest metabolic lesion was identi ed for each metastatic organ.

Statistics
One-way ANOVA was conducted, as well as statistical tests including Wilcoxon signed-rank test, Mann-Whitney U test, and Chi-square test, with the IBM SPSS Statistics (version 24; IBM Corp.) software.
Logistic regression (Forward LR) was used to analyze the in uence of parameters on metastatic or nonmetastatic LNs. P-values of less than 0.05 were considered statistically signi cant.

Patient Characteristics
Sixty-ve patients were enrolled in this study in total, including 34 males and 31 females. PSN. Imaging characteristics and one-way ANOVA among groups are shown in Table 1. The density of SCC and Inf & G nodules were basically recognized as SN, while the density of AC nodules was different.
Meanwhile, adenocarcinoma tended to have vacuoles, pleural indentation signs and lower uptake. Only lobular sign incidence exhibited differences between NSCLC and Inf & G, but the other imaging characteristics did not.   (Fig. 4). This indicated the potential of 68 Ga-DOTA-FAPI-04 for differential diagnosis of metastatic and non-metastatic LNs. One case (Fig. 5)     and FAPI SUV max /FDG SUV max could be used in differential diagnosis of metastatic and non-metastatic LNs while density and short diameters could not (Table.4). The area under receiver operating characteristic (ROC) curve of FAPI SUV max and FAPI SUV max /FDG SUV max was 0.927 and 0.954, respectively (Fig. 6), which indicated a prominent diagnostic ability of these two parameters.
FAPI SUV max ≥6 and the ratio of FAPI SUV max /FDG SUV max ≥1.1 were set as the cut-off value to analyze the diagnostic e cacy by the perspective of lesions and cases. The combination of these two conditions and [ 18 F]FDG criteria were also analyzed and shown in  (Fig. 7).

Discussion
[ 68 Ga]Ga-DOTA-FAPI-04 is becoming a promising tracer for general tumor diagnosis with its low radiation dose, fast tracer kinetics, favorable tumor-to-background ratios and invulnerability towards blood glucose uctuations [11] . Some clinical researches have revealed its applications in differential diagnosis and staging of several different kinds of tumors including glioblastomas, gastrointestinal tumors, hepatic nodules, etc. [12][13][14][15] . At present, pulmonary nodules stratifying and management mainly relies on the National Lung Screening Trial and Lung-RADS with CT imaging [24,25] , which indicated the probability of malignancy rather than diagnosis of NSCLC. Thus, diagnosis of NSCLC by any single examination remains a di culty. The underlying advantages of [ 68 Ga]Ga-DOTA-FAPI-04 application in NSCLC have been discussed [12,16] , and studies indicated NSCLC with highly uptake of [ 68 Ga]Ga-DOTA-FAPI-04. has been demonstrated as a useful, yet incomplete, tool for lung nodules diagnosis. One of the main limitations is that many benign lesions also present high FDG uptake which can easily lead to falsepositive results [27,28] . SCC and solid nodules of AC showed high uptake of both In ammatory nodules presenting high uptake of [ 68 Ga]Ga-DOTA-FAPI-04 might be due to the abundant activated broblast in in ammatory tissues [29] . Therefore, the false-positive cases caused by It was reported that among incidentally-detected small pulmonary nodules, PSNs had a high malignant potential (62.5-89.6%) [30,31] . In the case of adenocarcinoma appearing as SN, the SUV max of [ 18 F]FDG has been reported to be relatively low, ranging from 0.4 to 2.6 (mean SUV max 1.3) [32] . This phenomenon was also found in the present study, the mean [ 18 F]FDG SUV max of PSNs of AC in our study was 2.1 ± 1.1.
And the uptake of [ 68 Ga]Ga-DOTA-FAPI-04 (SUV max : 4.8 ± 2.8) was higher than [ 18 F]FDG with signi cant differences, which might be attributed to that partial solid nodules of AC were relatively slow-growing cancers which do not actively use glucose but the interstitial component proliferate rapidly as IHC implied in Fig. 3 [27] . N staging is crucial to the management and prognosis of NSCLC patients. However, previous studies have shown that metastatic LNs concerning NSCLC staging diagnosis remains di cult in [ 18 F]FDG PET/CT examination because high uptake of [ 18 F]FDG can be seen in reactive LNs in mediastinum and bilateral hilum, especially in hyperplasia, in ammation, infection and granulomatous disease [33,34] .The recognized consensus indicated that positive mediastinal nodal on PET/CT should be veri ed histologically by endobronchial ultrasound mediastinoscopy [35] . In our study, [ 18 F]FDG showed a satisfactory sensitivity albeit a poor speci city which was similar to the former studies [34] . Still, [ 68 Ga]Ga-DOTA-FAPI-04 showed excellent capacity to differentiate metastatic and non-metastatic LNs in multivariate regression analysis while other characteristics including density, calci cation and short diameter all failed, as non-metastatic LNs exhibited much lower uptake than metastatic LNs. Therefore, before [36] . Hence, the low FAPI uptake and low ratio of FAPI to FDG could attribute to the differential diagnosis of metastatic and non-metastatic LNs, these criteria help to identify more true negative patients who may avoid the invasive biopsy in clinical practice. This still requires further veri cation with more cases.
For M staging in this study, the average [ 18 F]FDG SUV max of primary lesions was 12.4 ± 7.6. In 6 of the 9 patients, the primary lesion's 2-18 F-FDG SUV max was 7.8 ± 2.2, which corresponded to a low to moderate Especially for those patients with metastasis in brain, skull and liver, [ 68 Ga]Ga-DOTA-FAPI-04 images showed lower background activity and higher image contrast than [ 18 F]FDG [37] . As in the case depicted ( Figure. 7), small metastasis of skull is easily missed on [ 18 F]FDG images, due to the high uptake level of brain, but it was exhibited apparently in [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT, which is consistent with former studies [16] .

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. supplementaryinformation.docx