The Added Value of [68Ga]Ga-DOTA-FAPI-04 PET/CT in Patients with Head and Neck Cancer of Unknown Primary with [18F]FDG Negative Findings

Purpose [ 18 F]uorodeoxyglucose ([ 18 F]FDG) positron emission tomography/computed tomography (PET/CT) plays an important role in location of primary tumor for patients with head and neck cancer of unknown primary (HNCUP). But sometimes, [ 18 F]FDG PET/CT still cannot nd the primary malignancy. As 68 Ga-radiolabeled broblast activation protein inhibitor (FAPI) PET/CT has promising results in detecting different tumor entities, our study aimed to evaluate the performance of [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT for detecting the primary tumor in HNCUP patients with negative [ 18 F]FDG ndings. Methods A total of eighteen patients (16 and 2 age, 55 24-72 years) with negative [ 18 F]FDG ndings were eligible in this study. All patients underwent [ 18 F]FDG and [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT within one week. Biopsy and histopathological examinations were done in the sites with positive [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT ndings. Results [ 68 Ga]Ga-DOTA-FAPI-04 detected the primary tumor in 7 of 18 patients The primary tumors sites were in nasopharynx palatine tonsil submandibular gland and hypopharynx (2/7). The primary tumors showed mild to intensive uptake of FAPI (mean SUV max , 8.79; range, 2.60-16.50) and excellent tumor-to-contralateral normal tissue ratio (mean SUV max ratio, 4.50; range, 2.17-8.21). In lesion-based analysis, a total of 65 lymph nodes and 17 bone metastatic lesions were identied. The mean SUV max of lymph node metastases were 9.05 ± 5.29 for FDG and 9.08 ± 4.69 for FAPI (p = 0.975); as for bone metastases, the mean SUV max were 8.11 ± 3.00 for FDG and 6.96 ± 5.87 for FAPI, respectively (p = 0.478). The mean tumor-to-background ratio (TBR) values of lymph node and bone metastases were 10.65 ± 6.59 vs. 12.80 ± 8.11 (p = 0.100) and 9.08 ± 3.35 vs. 9.14 ± 8.40 (p = 0.976), respectively. Conclusion We present rst evidence of diagnostic role of [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT in HNCUP, and our study demonstrated that [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT had the potential to improve the detection rate of primary tumor in HNCUP patients with negative FDG ndings. Moreover, [ 68 Ga]Ga-DOTA-FAPI-04 had similar performance in assessing metastases with [ 18 F]FDG.


Introduction
Head and neck cancer of unknown primary (HNCUP) is de ned as a metastatic disease in the cervical lymph nodes with an unidenti able primary tumor [1], even after a thorough diagnostic workup according to the National Comprehensive Cancer Network (NCCN) [2] and American Society of Clinical Oncology (ASCO) guidelines [3]. HNCUP constitutes 1-5% of all head and neck cancers [4][5][6]. Squamous cell carcinoma (SCC) is the most common pathological type of HNCUP, and approximately 90% of these cases are associated with human papillomavirus (HPV) [1]. The most frequent primary site of HNCUP is oropharynx, accounting for 80-90% [7]. However, some factors, like small tumor volume, hidden location, slow growth rate, and tumor involution, hinder primary site identi cation. The absence of primary tumor may result in uncertain treatment decisions and increasing psychological burden for patients with HNCUP.
Medical imaging plays an important role in oncology, particularly in tumor location [8]. Conventional imaging modalities, Computed Tomography (CT) and Magnetic Resonance Imaging (MRI), can provide plentiful anatomical information about primary and metastatic malignancies, including anatomic situation, extracapsular extension, and contralateral lesions. However, the detection rates of primary site for these two imaging modalities range from 9 to 23% in HNCUP [9][10][11]. Positron emission tomography/computed tomography (PET/CT), a typical molecular imaging modality, outperforms CT and MRI in identifying primary tumor with a detection rate of 25-69% by using [ 18 F] uorodeoxyglucose ([ 18 F]FDG) [12][13][14][15]. Nevertheless, some limitations hamper the application of [ 18 F]FDG PET/CT in primary tumor identi cation for HNCUP [16,17]. Firstly, physiological FDG uptake can be seen in any lymphatic structure (especially Waldeyer's ring), salivary glands, and brown fat. Secondly, symmetrical vocal cords and neck muscles uptake are commonly seen if the patient talks or coughs during the uptake period. Thirdly, infection and chronic in ammation (i.e., nasopharyngitis, amygdalitis, and gingivitis) can also result in high FDG uptake. These three limitations may lead to false positive ndings with a rate of 16-25% [4,12,15]. Last but not least, false negative FDG uptake can be seen in small, mucinous, well-differentiated, and necrotic lesions [17]. Therefore, novel speci c radiopharmaceuticals with low background uptake in head and neck, which may better improve the detection rate of primary tumor in HNCUP, are in urgent need.
Cancer associated broblasts (CAFs), accounting for high proportion of most solid tumor mass, plays a vital role in tumor growth, migration, and progression [18]. The major feature to discriminate CAFs from normal broblast is the overexpression of broblast activation protein (FAP) [19]. FAP is a type II membrane-bound glycoprotein belonging to the dipeptidyl peptidase 4 family [20]. The presence of FAP was observed on a variety of epithelial and mesenchymal malignancies [21,22]. Recently, 68 Ga-radiolabeled broblast activation protein inhibitor (FAPI), a novel FAP-targeted PET tracer, has shown great value in diagnosis of diverse carcinomas, including welldifferentiated cancers [23][24][25]. Furthermore, some studies [26,27]  A total of thirty-two patients were enrolled from our center between June 2020 to Feb 2021. However, among these patients, three were diffuse large B-cell lymphoma (DLBCL), three were lung origin, one was digestive system origin, one was female reproductive system origin, and six were with both positive [ 18 F]FDG and [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT ndings. All patients had biopsies of the sites with positive PET/CT ndings for primary tumors. For these patients without clear primary tumor site after PET/CT scan, a three-month follow-up was performed. The metastatic lesions were de ned by pathology, imaging ndings, and follow-up. The data of demographics were collected. This study was approved by Fudan University Shanghai Cancer Center Institutional Review Board (ID 2004216-25) conducted in accordance with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards, and all subjects signed an informed consent form.
Radiopharmaceuticals and PET/CT scanning procedure CT scanning was rst performed for location: scanning ranging from the proximal thighs to head, with 120 kV, 100 mAs, CARE Dose4D, slice thickness 3 mm, increment 2 mm, pitch 1.0, rotation time 0.5s, soft-tissue reconstruction kernel. Immediately after CT scanning, a PET emission scan that covered the corresponding eld of CT was acquired in 3-dimensional mode using FlowMotion with a speed of 2. The emission data were corrected for random, scatter and decay. PET image data sets were reconstructed iteratively using an orderedsubset expectation maximization iterative reconstruction (OSEM) by applying CT data for attenuation correction.
Fusion images were reviewed and manipulated on a multimodality computer platform (Syngo, Siemens, Knoxville, Tennessee, USA). Two experienced nuclear medicine physicians analyzed and interpreted the images independently, and they reached a consensus in case of inconsistency.
For quantitative analysis, maximum or mean of standardized uptake value (SUV) normalized to body weight were manually computed for tumor lesions and healthy tissues by drawing a 3-dimensional volume of interest (VOI). Meanwhile, SUV max ratio for primary tumor was de ned as the quotient of the SUV max of primary tumor and the contralateral normal tissue, and tumor-to-background ratio (TBR) for tumor lesions was calculated according to the formula: TBR = tSUV max /bSUV mean , where tSUV max is the maximum SUV of tumor lesion, and bSUV mean is the mean SUV of muscle.

Statistical analyses
All statistical analyses were performed using SPSS 25.0 (IBM, Armonk, NY, USA). Means with standard deviation (SD) or medians with ranges were used to describe continuous characteristics. To compare the uptake of [ 18 F]FDG and [ 68 Ga]Ga-DOTA-FAPI-04 in metastatic lesions, two-sample t tests were used. Two-tailed p < 0.05 were considered statistically signi cant.

Patients
From June 2020 to April 2021, there were eighteen patients with HNCUP were eligible for this study according to the inclusion and exclusion criterion (Fig. 1). The basic clinical characteristics were presented in Table 1. Among the included eighteen patients (16 males and 2 females; median age, 55 years; range, 24-72 years), two (11.11%) were infected with Epstein-Barr virus (EBV); six (33.33%) were infected with HPV; sixteen (88.89%) were pathologically diagnosed with cervical lymph node SCC and two (11.11%) were adenocarcinoma (AC).  Table 2). Both tracers showed intensive uptake in lymph node and bone metastases. The mean SUV max value of lymph node metastases was 9.05 ± 5.29 for FDG and 9.08 ± 4.69 for FAPI (p = 0.975). In case of TBR, FAPI was a litter higher than FDG (12.80 ± 8.11 and 10.65 ± 6.59, respectively). But the difference was not signi cant (p = 0.100). For bone metastases, the mean SUV max value was 8.11 ± 3.00 for FDG and 6.96 ± 5.87 for FAPI (p = 0.478), and the mean TBR value was 9.08 ± 3.35 and 9.14 ± 8.40 (p = 0.976), respectively.
Generally, no signi cant uptake difference was observed between FDG and FAPI in lymph node and bone For the location of primary tumor, one was in nasopharynx, two were in palatine tonsil (Fig. 3), two were in submandibular gland (Fig. 4), and two were in hypopharynx ( Fig.5 and Table 3). In patient 2 (Fig. 3), right and left palatine tonsil showed similar uptake of FDG (SUV max = 6.4 and 6.0, respectively), with a SUV max ratio of 1.

Discussion
Identifying the primary tumor remains a concern for patients with HNCUP, though the development in imaging, endoscope, and pathology techniques. If without any positive ndings by non-invasive procedures, invasive diagnostic operations, like tonsillectomy, are performed with a risk of bleeding or infection [5]. Thus, novel noninvasive methods may be needed for improving detection rate of primary tumor in HNCUP patients. This study took advantage of trials to investigate the performance of Another previous research demonstrated that an SUV max ratio of FDG uptake between tonsils of ³ 1.6 could be regarded as malignancy and used to guide biopsy [29]. In this study, two patients were diagnosed with palatine tonsil carcinoma by tonsillectomy. Puzzlingly, [ 18 F]FDG PET/CT revealed no visual difference between right and left palatine tonsils in both two cases. Furthermore, the SUV max ratios of FDG uptake were all approximate equivalent to 1.00 (1.07 and 1.04 for patient 2 and 3, respectively), which was mistaken as physiologic uptake. By In addition to high grade physiologic uptake of head and neck, small size of the lesions was the major reason for the false negative FDG ndings due to the partial volume effect and low tumor glucose metabolic activity [30,31].
In this study, [ 18 F]FDG PET/CT missed 3 of 7 primary tumors by reason of the small size (diameter < 10 mm).
Encouragingly, [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT revealed mild uptake (SUV max = 2.60, 3.20, and 3.60 for patient 1, 6, and 7, respectively) and clearly visual difference (SUV max ratio = 2.17, 2.91, and 3.27, respectively) in these primary tumors with small size, which was consistent with previous research [24]. The uptake of FAPI is mainly based on the expression of FAP on CAFs among solid tumor microenvironment. And even small primary tumors of T1 stage could show a moderate FAP expression [27]. Most of the researches focus on SCC, as it is the most frequent pathological type of HNCUP [3][4][5]. However, some other pathological types, like adenocarcinoma and neuroendocrine carcinoma, may cause diagnostic di culties in clinical practice due to inexperience. Moreover, when it comes to cervical metastatic adenocarcinoma, diagnostic resection of salivary gland is not recommended even after thorough non-invasive investigations. Furthermore, salivary gland cancers show paucity of FDG avidity [32], which was proofed again in our study (patient 4 and 5 [33]. However, these tracers are too speci c to identify all types of head and neck cancers. Promisingly, recent studies have demonstrated [ 68 Ga]Ga-DOTA-FAPI-04 can evaluate a broad spectrum of malignancies, including adenocarcinoma, neuroendocrine carcinoma, and well-differentiated carcinoma and so on [23,24]. In this study, [ 68 Ga]Ga-DOTA-FAPI-04 showed intensive uptake in submandibular gland (SUV max = 16.50 and 15.80, respectively), providing su cient information for following surgery. Notably, [ 68 Ga]Ga-DOTA-FAPI-04 had a higher detection rate in adenocarcinoma (2/2, 100%) than SCC (5/16, 31.25%) of HNCUP, which indicated [ 68 Ga]Ga-DOTA-FAPI-04 was more sensitive to adenocarcinoma. However, further research with larger sample size is needed to verify this result.
EBV and HPV infection strongly suggest the location (nasopharynx and oropharynx, respectively) of primary tumor of SCC HNCUP [34]. Gi Cheol Park et al [35] found that the sensitivity and accuracy of HPV were higher than those of [ 18 F]FDG PET/CT (71.4% vs. 49.2% and 85.2% vs. 68.5%, respectively). In the current cohort, primary tumors were detected in 3 of 6 HPV-positive patients, and 2 were palatine tonsil SCC and 1 was submandibular gland salivary ductal carcinoma. Thus, HPV may offer limited information when concerning various pathologic types rather than SCC. Furthermore, virus infection status can't provide direct anatomical information for further biopsy. With regard to the detection of regional and distant metastases, the performance of [ 68 Ga]Ga-DOTA-FAPI-04 PET/CT varies among different researches [24,27]. In our study, There are some limitations in this study. First, the main limitation is the relatively small number of patients, and the number of pathologic types is uneven. In the future, larger population cohort studies with more cancer types need to take into account. Second, not all the metastatic lymph nodes are con rmed by pathology. Authors' contributions All authors contributed to the study conception and design. Patients were enrolled by Xiaomin Ou, Zuguang Xia and Qing Guan. Material preparation, data collection and analysis were performed by Bingxin Gu, Xiaoping Xu, Ji Zhang and Silong Hu. The rst draft of the manuscript was written by Bingxin Gu. The nal manuscript was revised by Zhongyi Yang and Shaoli Song. All authors read and approved the nal manuscript.
Compliance with ethical standards Con ict of interest The authors declare that they have no con ict of interest.
Ethics approval All procedures involving human participants were carried out in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This article does not contain any experiments with animals.
Consent to participate Informed consent was obtained from all individual participants included in the study.
Consent to publication Not applicable.
Availability of data and material Not applicable.
Code availability Not applicable.     with metastatic adenocarcinoma of the right neck.
[18F]FDG PET/CT was negative for the detection of the primary. On [68Ga]Ga-DOTA-FAPI-04 PET/CT, there was an intensive uptake in the right submandibular gland (d-f, red arrow; SUVmax = 15.80), while low background uptake was seen in the left submandibular gland (SUVmax ratio = 6.87). Subsequent surgery con rmed salivary ductal carcinoma. Black and white arrows indicate the metastatic lymph nodes. (d-f, red arrow; SUVmax = 3.60), while low background uptake was seen in the left sinus piriformis (SUVmax ratio = 3.27). Subsequent biopsy con rmed SCC. Black and white arrows indicate the metastatic lymph node.