A head-to-head comparison of 68Ga-DOTA-FAPI-04 and 18F-FDG PET/MR in patients with nasopharyngeal carcinoma: a prospective study

To conduct a head-to-head comparison of the diagnostic ability of 68Ga-DOTA-FAPI-04 (68Ga-FAPI) and 18F-FDG PET/MR in nasopharyngeal carcinoma (NPC) patients. Patients diagnosed with NPC were prospectively enrolled. All patients underwent head-and-neck 68Ga-FAPI PET/MR and 18F-FDG PET/MR within 1 week. Primary tumor, lymph node numbers, and tracer uptake were compared by SUVmax and visual evaluation. The primary tumor volumes derived from 68Ga-FAPI, 18F-FDG PET, and MRI were also compared. Fifteen patients were enrolled from June to August 2020. Both 68Ga-FAPI and 18F-FDG PET had 100% detection rate of the primary tumor. The 68Ga-FAPI SUVmax of primary tumors (13.87 ± 5.13) was lower than that of 18F-FDG (17.73 ± 6.84), but the difference was not significant (p = 0.078). Compared with 18F-FDG, 68Ga-FAPI PET improved the delineation of skull-base invasion in eight out of eight patients and intracranial invasion in four out of four patients. When 25%SUVmax of 68Ga-FAPI or 20%SUVmax of 18F-FDG was utilized as a threshold for determining tumor volume, it was highly consistent with MRI. 18F-FDG PET detected much more positive lymph nodes than 68Ga-FAPI (100 vs 48). The SUVmax of 48 paired lymph nodes was significantly lower on 68Ga-FAPI than 18F-FDG (8.67 ± 3.88 vs 11.79 ± 6.17, p < 0.001). Additionally, 68Ga-FAPI further detected four highly suspected small, distant metastases in three patients. Compared with 18F-FDG, 68Ga-FAPI changed overall staging in six of fifteen patients, with three patients being up-staged, and three down-staged. 68Ga-FAPI outperforms 18F-FDG in delineating the primary tumor and detecting suspected distant metastases, particularly in the evaluation of skull-base and intracranial invasion, suggesting 68Ga-FAPI hybrid PET/MR has the potential to serve as a single-step staging modality for patients with NPC. However, its value regarding lymph node and distant metastases evaluation needs further study. NCT04554719. Registered September 8, 2020 - retrospectively registered, http://clinicaltrails.gov/show/NCT04554719


Introduction
Nasopharyngeal carcinoma (NPC) is a common malignancy in East and Southeast Asia, and particularly in southern China [1]. In patients with localized, advanced-stage disease, it is characterized by a high rate of skull-base and intracranial invasion. Determining the extent of tumor invasion and staging by noninvasive imaging techniques are critical for treatment choice and subsequent planning [2,3].
Both magnetic resonance imaging (MRI) and fluorine-18 fludeoxyglucose ( 18 F-FDG) PET/CT are established imaging modalities for NPC management which are currently recommended for use in NPC staging by the National Comprehensive Cancer Network (NCCN) [3]. 18 F-FDG is the most frequently utilized tracer in PET imaging [4]. However, due to high physiologic glucose uptake in the brain and some healthy tissues [5], and the fact that CT has limited soft-tissue resolution compared with MRI, the sensitivity of 18 F-FDG PET/CT for the T staging of NPC is inadequate, particularly when characterizing skull-base and intracranial invasion [6]. Hybrid PET/MR is a new modality that acquires MR and PET images simultaneously, providing high softtissue contrast and multi-parametric information regarding the disease. Several studies have investigated the feasibility and clinical potential of 18 F-FDG PET/MR in NPC and have found it superior to PET/CT and head-and-neck MRI for assessing the extension of the primary tumor into the parapharyngeal spaces, skull-base, intracranial area, and nodal metastases. Therefore, PET/MR could serve as a reliable, onestep imaging examination for NPC assessment [7][8][9].
Gallium-68 labeled fibroblast activation protein inhibitor ( 68 Ga-FAPI) has recently been introduced as a promising tumor imaging agent targeting cancer-associated fibroblasts [10,11]. A high uptake of radioactive FAPI has been confirmed in many types of tumors, including head-and-neck cancers [10,12,13]. However, there are currently no studies that evaluate the use of 68 Ga-FAPI PET for NPC. Additionally, the background level of 68 Ga-FAPI in the normal brain is reported to be very low [12]. We hypothesize, therefore, that 68 Ga-FAPI has good diagnostic value in NPC. In this study, we conducted a prospective study to investigate whether simultaneous 68 Ga-FAPI head-neck PET/ MR is better than 18 F-FDG PET/MR for detecting primary tumor invasion, lymph node involvement, and distant metastases in NPC patients, and to explore the clinical impact (TNM staging) of 68 Ga-FAPI PET/MR compared with 18 F-FDG PET/MR.

Patients
This prospective study was approved by our hospital's Institutional Review Board (IRB 20200290). All patients signed an informed consent form prior to participation in the study. The inclusion criteria were patients (1) with histologically proven NPC, and (2) who had no anti-tumor therapy 4 weeks prior to the examination. Exclusion criteria included patients (1) who had other primary malignancies at the time of examination, (2) with a history of neck surgery, (3) who suffered from hyperglycemia or severe hepatic and renal insufficiency, or (4) had contraindications for MRI. 18 F-FDG PET/MR and 68 Ga-FAPI PET/MR acquisition All patients had fasted for at least 6 h before 18 F-FDG administration. Blood glucose was tested to ensure a normal blood glucose level. Whole-body (from top of head to mid-thigh) PET/CT was performed approximately 60 min after the intravenous injection of 18 3 ) with TOF and point spread function (PSF)-ordered subset expectation maximization (OSEM) algorithms with 28 subsets and three iterations, followed by a 3-mm Gaussian filter. A 32-channel headneck united coil was used and the MRI sequences included axial and sagittal T1-weighted imaging (fast spin echo, TE/ TR = 14/500 ms, section thickness 4 mm), coronal T1weighted imaging (fat-suppression fast spin echo, TE/ TR = 10/500 ms, section thickness 4 mm), axial T2weighted imaging (fast spin echo, TE/TR = 90/5000 ms, section thickness 4 mm), and coronal T2-weighted imaging (fat-suppression fast spin echo, TE/TR = 90/5000 ms, section thickness 4 mm). The matrix of the axial plane was 320 × 224, and the matrix of the sagittal and coronal plane was 288 × 192. Contrast-enhanced MR was performed during either 18

Imaging interpretation
The acquired PET and MR images were sent to the AW workstation (AW4.6, GE Healthcare, USA) for registration, fusion, and measurement. Both 18 F-FDG and 68 Ga-FAPI PET/MR images were visually evaluated by two experienced nuclear medicine physicians (C. Q. with 12 years of experience in nuclear oncology, and F. L. with 20 years of experience in MR and 5 years of experience in nuclear oncology) who are aware of clinical information of patients. The two reviewers conducted a frame-by-frame comparative analysis of the fused images of 18 F-FDG PET/MR and 68 Ga-FAPI PET/MR, and reached a final TNM diagnosis based mainly on the fusion images. Any initial differences of opinion were resolved by consensus.
T staging assessment According to the NCCN guidelines [3], contrast-enhanced MR images are the gold standard for evaluating primary lesions and their invasion of adjacent tissues. Compared with contrast-enhanced MR images, a lesion showing 18 F-FDG or 68 Ga-FAPI uptake that exceeded uptake in the adjacent tissue was considered to have tumor infiltration. The region of interest of the primary tumor was manually drawn along the margin of the lesion on the axial contrast-enhanced MR images and T2-weighted images because of their clear tumor borders. The maximum standardized uptake value (SUVmax) of the region of interest was measured on the PET images. The tumor volume shown on the PET images was calculated based on SUV thresholds of 20%, 25%, and 30% of SUVmax, respectively.
N staging assessment On MR images, the diagnosis of metastatic lymph nodes should satisfy at least one of the following criteria: (a) necrosis or extracapsular spread was observed, (b) the shortest axial diameter was ≥5 mm in the retropharyngeal region or ≥ 10 mm in other regions, (c) there was a cluster of three or more lymph nodes of borderline size [14]. On PET images, if 18 F-FDG or 68 Ga-FAPI uptake exceeded the surrounding tissue, it was regarded as a positive lymph node uptake. The site and number of positive lymph nodes were recorded and compared between the modalities.
M staging assessment The presence of distant metastases was determined based on the tracer uptake and MR signal intensity abnormalities. The site of the metastases was also recorded.

Statistics
All analyses were performed using the SPSS statistical package (version 22.0; SPSS, Chicago, IL, USA). Differences in SUVmax between groups were determined using a paired t test. Two-tailed p values of less than 0.05 were considered statistically significant.

Patient cohort
Fifteen patients-eight men and seven women, aged from 36 to 69 years (mean 51.2 ± 9.4 years)-were included in this study from June to August 2020. The median time interval between the two scans was 1 day (range 1-3 days). All patients tolerated 68 Ga-FAPI and 18 F-FDG PET/MR well, and no 68 Ga-FAPI-related adverse effects were observed. Table 1 shows the clinical information and the MR, 18 F-FDG, and 68 Ga-FAPI PET-based TNM staging of all the patients. Of the 15 patients, 14 were newly diagnosed with NPC, in which three cases were well-differentiated keratinizing squamous cell carcinoma (WHO Type I), two cases were nonkeratinizing differentiated carcinoma (WHO Type II), eight cases were nonkeratinizing undifferentiated carcinoma (WHO Type III), and one case was keratinizing carcinoma with unknown differentiation status. The other case was recurrent NPC with an unknown pathological type; the patient was diagnosed in 2016 and underwent radiotherapy which ended in the same year. No distant metastases were revealed on whole-body 18 F-FDG PET/CT in any of the patients.

Primary tumor detection and visual evaluation of primary tumor invasion
18 F-FDG and 68 Ga-FAPI had equivalent detection ability for primary tumor in all NPC patients with 100% positive detection rate. Figure 1 shows the images of four representative patients. Primary tumor uptake was higher in 18 F-FDG (SUVmax 17.73 ± 6.84) than 68 Ga-FAPI (SUVmax 13.87 ± 5.13), but no significant differences were found (P = 0.078, Table 2). We further compared the correlation between the uptake of two tracers and found that the SUVmax of 68 Ga-FAPI had no relationship with 18 F-FDG (r = 0.227, p = 0.416, Fig. 2a). The T staging of 68 Ga-FAPI was consistent with the MRI findings in all the patients, but the T staging in one of the patients was underestimated by 18 F-FDG (Table 1).
A visual evaluation of primary tumor invasion with two tracers was conducted ( Table 3). All 15 patients had nasopharyngeal invasion with significant uptake on both 68 Ga-FAPI and 18 F-FDG. The invasion extent and the borders of the lesions were clearly delineated by both modalities, comparable to MR. However, when comparing the 68 Ga-FAPI and 18 F-FDG PET images, a visual evaluation was discordant in ten patients. 68 Ga-FAPI was found to be superior to 18 F-FDG in six of the 15 patients (40%) and inferior to 18 F-FDG in four patients (26.7%).
Twelve patients had a parapharyngeal space invasion. The tumor extent revealed by 68 Ga-FAPI PET in all patients was comparable to MR, whereas the 18 F-FDG images of two patients were inferior to the MR images. The number of cases of 68 Ga-FAPI that were superior, equivalent, and inferior to 18 F-FDG was five, two, and five, respectively. Figure 3 shows an example of parapharyngeal space invasion, with a greater extent on 68 Ga-FAPI PET and MR than 18 F-FDG PET.
Eight patients had skull-base invasion. 68 Ga-FAPI PET showed comparable accuracy to that of head-and-neck MRI for detecting the involvement of the skull base in seven out of eight patients. In all eight patients (100%), both 68 Ga-FAPI PET and MR showed a larger extent or a clearer border of skull-base invasion than 18 F-FDG PET (Fig. 4).
Four patients had intracranial invasion, which was clearly visualized by 68 Ga-FAPI PET and MR. However, 18 F-FDG PET was either negative or indeterminate because of the physiologic uptake of 18 F-FDG in the brain (Fig. 5).

Comparison of primary tumor volume between the modalities
As shown in Tables 4 and 5, tumor volumes measured by contrast-enhanced MRI and T2-weighted images were highly consistent (intraclass correlation coefficient, ICC 0.93), which indicates that the volumes measured by enhanced MR and T2-weighted images had a high level of credibility. The delineation volume of 68 Ga-FAPI 25%SUVmax and 18 F-FDG 20%SUVmax as thresholds had the highest credibility and consistency level with the volume measured by MR (Table 5, ICC 0.85-0.91).

Distant metastases
68 Ga-FAPI was superior to 18 F-FDG in 100% (3/3) of the patients with suspected distant metastases. 68 Ga-FAPI aided the detection of three small skull lesions in two patients, which were initially overlooked during MRI interpretation and false-negative on the 18 F-FDG PET images (Fig. 6). In another patient, a small suspicious metastasis on the pons was detected by 68 Ga-FAPI and enhanced MRI, which was missed by 18 F-FDG PET because of high physiologic uptake.

Changes in the overall staging
Overall, compared with 18

Discussion
In this study, we prospectively compared the diagnostic and staging effectiveness of 68 Ga-FAPI and 18 F-FDG head-and-neck PET/MR in NPC patients. Our results  showed that 68 Ga-FAPI was superior to 18 F-FDG in assessing skull-base and intracranial invasion and the small metastases on the skull and brain, which changed the overall staging in six patients. This suggested that 68 Ga-FAPI PET/MR may be another viable imaging candidate for diagnosing NPC. However, its value for N and M staging needs to be further explored. The primary tumor in NPC can be visualized by both 18 F-FDG and 68 Ga-FAPI. Although 18 F-FDG showed higher uptake than 68 Ga-FAPI, the difference was not significant. Interestingly, no correlation was found in the primary tumor uptake values of the two tracers, which may be explained by different principles. 18 F-FDG reflects the glucose utilization of the lesion [4], but 68 Ga-FAPI reflects the activity of the cancer-associated fibroblast [11]. 68 Ga-FAPI showed an equivalent primary tumor detection to 18 F-FDG, but it has the additional advantage of low brain background level making it ideal for evaluating skull-base and intracranial invasion, as was shown in this study where a larger extent or a clearer border was shown on 68 Ga-FAPI PET than 18 F-FDG PET in all 8 patients with skull-base invasion, and 68 Ga-FAPI PET clearly visualized 4 patients with intracranial invasion. Therefore, 68 Ga-FAPI is superior to 18 F-FDG PET for exact T staging. Radiotherapy plays an important role in NPC, so the delineation of the target area is very important [15]. The current gold standard is MRI, but it can give false-positive results because of the detection of edema and inflammation [16], and it only provides anatomic information. 18 F-FDG PET has limited value in locally advanced NPC patients because of physiologic uptake in a normal brain. Therefore, an alternative method is needed. In this study, when 25%SUVmax of 68 Ga-FAPI was utilized as a threshold for tumor volume determination, it was highly consistent with the contrastenhanced MRI-the current gold standard. Recently, one study demonstrated that FAPI-based gross tumor volumes (GTVs) were significantly larger than conventional CT-GTVs in 14 patients with head-and-neck squamous cell carcinoma [13]. This indicates that 68 Ga-FAPI PET/MR may play an important role in GTV delineation in NPC patients, particularly for those with locally advanced disease. This needs to be studied in more detail in the future.
Conventional imaging techniques such as ultrasound, CT, and MR cannot accurately assess small metastatic cervical lymph nodes because their assessment based mainly on the shape and size of the lymph nodes, and the presence of necrosis [17]. 18 F-FDG PET/CT was reported to be more effective than conventional imaging in identifying small metastatic cervical lymph nodes [18,19]. However, another study reported inconsistent results; for only PET/CT-positive lymph nodes, the true-positive and false-positive rates were 27% (13/48) and 73% (35/48), respectively, and no significant difference was observed for SUVmean and lymph node sizes between the true-positive and false-positive readings [20]. False positives are common for cervical lymph nodes, resulting from a high incidence of inflammation and reactive hyperplasia [20,21]. In our study, significantly more 18 F-FDG-avid lymph nodes were detected than those of 68 Ga-FAPI (100 vs 48). Further analysis showed good correlations between 18 F-FDG and 68 Ga-FAPI uptake in paired double-positive lymph nodes, which was different from that of primary tumors. Unfortunately, the imaging findings were not verified by histopathological results. These results proposed interesting questions: should the double-positive lymph nodes be considered as true metastases, and how to diagnose the only FDGavid lymph nodes? Therefore, the difference and mechanism of 18 F-FDG and 68 Ga-FAPI uptake in primary and lymph node lesions needs to be further verified by histopathological results.
Although 18 F-FDG showed good diagnostic accuracy for distant malignant tumor metastases [22], 68 Ga-FAPI PET surprisingly detected three small skull lesions highly suspected of metastases in two patients which were missed by 18 F-FDG in this study. The reasons may be related to the component of activated fibroblasts and/or myofibroblasts in the osseous lesions [23]. Although none of the bone metastases could be diagnosed by histopathology, we highly suspected these lesions were metastases, because the primary NPC tumors have strong positive uptake of 68 Ga-FAPI, and MR signals of the skull lesions were abnormal through careful observation, and the lesions became slightly larger in the follow-up MR imaging after chemotherapy and nasopharyngeal radiotherapy (Fig.  6). One pons metastasis was also detected by 68 Ga-FAPI rather than 18 F-FDG due to the low brain background of 68 Ga-FAPI. Interestingly, in three patients up-staged by 68 Ga-FAPI PET, the 18 F-FDG SUVmax of primary tumor is much higher than those of other patients, indicating high aggressiveness of disease. Pang reported 68 Ga-FAPI outperforms 18 F-FDG PET/CT in identifying bone metastasis in a patient with metastatic breast cancer [24]. A recent study also reported more bone metastases with higher uptake were observed on 68 Ga-FAPI PET/CT than 18 F-FDG in gastrointestinal cancers [25]. Therefore, 68 Ga-FAPI may be superior to 18 F-FDG in detecting small bone metastases. As a new, nonspecific tumor imaging agent, 68 Ga-FAPI showed good performance in the diagnosing and staging of NPC. However, the number of cases in this study was relatively small, and none of the lymph nodes or bone metastases were diagnosed by histopathology. Therefore, the diagnostic efficacy of 68 Ga-FAPI PET/MR regarding lymph nodes and bone metastases requires further investigation. The uptake mechanism of 68 Ga-FAPI in the primary tumor and metastases also needs further study. Despite these limitations, our preliminary study showed that 68 Ga-FAPI PET/MR may be a candidate for becoming a single-step staging modality for NPC.

Conclusion
In this preliminary study, 68 Ga-FAPI may outperform 18 F-FDG in the delineation of primary tumors and the detection of suspicious distant metastases because of its low background level in the brain, which suggests that simultaneous 68 Ga-FAPI PET/MR may serve as a single-step staging modality for patients with NPC. The N staging value of 68 Ga-FAPI PET is indefinite. Multi-center randomized, controlled studies using a larger patient population will be needed to further strengthen our findings in the future.