Prospective Head-to-Head Comparison of ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT in De Novo and Recurrent Epstein-Barr Virus–associated Nasopharyngeal Carcinoma

Abstract

Purpose

To compare gallium 68 (⁶⁸Ga) fibroblast activation protein inhibitor (FAPI) PET/CT with plasma Epstein-Barr virus (EBV)–DNA load measurement, contrast-enhanced MRI/CT, and fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) PET/CT for nasopharyngeal carcinoma (NPC) initial assessment and recurrence surveillance.

Materials and Methods

Participants with NPC who underwent both ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT examinations within 1 week for initial assessment (treatment-naive cohort) and recurrence surveillance (posttreatment cohort) were enrolled in this prospective study conducted between July 2020 and November 2023 (Chinese Clinical Trial Registry identifier no. ChiCTR2100054163). The data of plasma EBV DNA load and contrast-enhanced MRI/CT were analyzed. Diagnostic performance was compared using the McNemar test. The relationships between plasma EBV DNA and the two tracers were assessed by using simple linear regression analysis.

Results

Sixty-five participants (median age, 50 years [IQR, 41.50–61 years]; 50 male) were included, with 31 in the treatment-naive cohort and 34 in the posttreatment cohort. In participant-level analysis, ⁶⁸Ga-FAPI PET/CT demonstrated the highest accuracy rate in identifying the participants with de novo (100% [31 of 31] vs 84% [26 of 31], 77% [24 of 31], and 84% [26 of 31], P = .06, .02, and .06, respectively) and recurrent (94% [32 of 34] vs 53% [18 of 34], 68% [23 of 34], and 85% [29 of 34], P = .001, .02, and .25, respectively) NPC, compared with plasma EBV DNA, contrast-enhanced MRI/CT, and ¹⁸F-FDG PET/CT. Furthermore, compared with contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT, ⁶⁸Ga-FAPI PET/CT led to upstaging in five of 31 (16%) and downstaging in four of 31 (13%) treatment-naive participants. Lesion uptake at ⁶⁸Ga-FAPI PET/CT did not correlate with plasma EBV DNA load (P > .05).

Conclusion

⁶⁸Ga-FAPI PET/CT outperformed plasma EBV DNA level, contrast-enhanced MRI/CT, and ¹⁸F-FDG PET/CT in initial assessment and recurrence surveillance of NPC.

Keywords: Head/Neck, PET/CT, Nasopharyngeal Carcinoma, Epstein-Barr Virus, Fibroblast Activation Protein Inhibitor, FDG

Chinese Clinical Trial Registry identifier no. ChiCTR2100054163

Supplemental material is available for this article.

© RSNA, 2026

See also commentary by Basha and Almalki in this issue.

Summary

Gallium 68 fibroblast activation protein inhibitor PET/CT outperformed conventional methods (plasma Epstein-Barr virus DNA level, contrast-enhanced MRI/CT, and fluorine 18 fluorodeoxyglucose PET/CT) in nasopharyngeal carcinoma primary evaluation and recurrence monitoring, suggesting its potential clinical utility.

Key Points

• ■ Gallium 68 (⁶⁸Ga) fibroblast activation protein inhibitor (FAPI) PET/CT demonstrated the highest accuracy rate in identifying the participants with de novo (100% [31 of 31] vs 84% [26 of 31], 77% [24 of 31], and 84% [26 of 31], P = .06, .02, and .06, respectively) and recurrent (94% [32 of 34] vs 53% [18 of 34], 68% [23 of 34], and 85% [29 of 34], P = .001, .02, and .25, respectively) nasopharyngeal carcinoma, compared with plasma Epstein-Barr virus DNA load, contrast-enhanced MRI/CT, and fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) PET/CT.

• ■ Compared with contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT, ⁶⁸Ga-FAPI PET/CT led to upstaging in five of 31 (16%) and downstaging in four of 31 (13%) treatment-naive participants, respectively.

• ■ Lesion uptake at ⁶⁸Ga-FAPI PET/CT did not correlate with plasma EBV DNA level (P > .05).

Introduction

Nasopharyngeal carcinoma (NPC) is a malignant tumor arising from nasopharyngeal epithelium. It is endemic to Southeast Asia, particularly southern China (1,2). Due to occult onset, most participants with NPC are diagnosed at a locoregionally advanced stage, resulting in poor prognoses (3). Nonkeratinizing carcinoma, accounting for almost 95% of cases, is the most common histologic type of NPC and is closely associated with Epstein-Barr virus (EBV) infection (4). EBV plays a vital role in the occurrence, development, and recurrence of NPC, and the measurement of plasma EBV DNA load has been widely used in diagnosis and surveillance (5). However, because more than 90% of the global population is infected with EBV and remains persistently seronegative, recurrent infections may complicate the clinical management of NPC when plasma EBV DNA levels are used for monitoring (6).

As a noninvasive imaging modality, MRI is the reference standard for diagnosis, primary tumor (T) and lymph node staging, and surveillance of NPC (3). However, its limited ability in the detection of distant metastases, underdiagnosis of early T-stage disease, and difficulty in identifying regional metastatic lymph nodes smaller than 10 mm in short-axis diameter can hamper its diagnostic performance (7,8). Flourine 18 (¹⁸F)–labeled fluorodeoxyglucose (FDG) PET/CT, a whole-body scanning modality, demonstrates better performance than MRI in detecting regional lymph nodes and distant metastases of NPC but is less accurate for T staging (9). Moreover, false-positive ¹⁸F-FDG uptake caused by infection or chronic inflammation can obscure the distinction between primary tumor and metastatic lymph nodes (10). Therefore, novel reliable and reproducible imaging techniques are urgently needed to improve NPC diagnosis and surveillance.

Fibroblast activation protein (FAP), which is overexpressed in the tumor stroma, particularly in cancer-associated fibroblasts in more than 90% of epithelial carcinomas, has emerged as a promising target for pan-tumor PET/CT imaging, including in NPC (11). Recent studies have demonstrated the potential utility of gallium 68 (⁶⁸Ga)–labeled FAP inhibitor (FAPI) PET/CT for detecting both primary and metastatic lesions in newly diagnosed NPC (12–14). Zhao et al (12) reported that ⁶⁸Ga-FAPI PET/CT showed higher uptake in primary NPC tumors, regional lymph nodes, and bone and visceral metastases compared with ¹⁸F-FDG PET/CT. Zheng et al (13) further found that ⁶⁸Ga-FAPI PET/CT outperformed both MRI and ¹⁸F-FDG PET/CT in assessing the primary tumor in NPC. However, few studies have evaluated the role of ⁶⁸Ga-FAPI PET/CT for NPC surveillance. Surveillance imaging after radiation therapy for NPC is challenging because treatment-induced changes like inflammation and fibrosis cause ambiguous findings at CT/MRI and nonspecific uptake at ¹⁸F-FDG PET/CT (15). In contrast, FAP expression is typically not upregulated in pure inflammatory or fibrotic scar tissues. Therefore, ⁶⁸Ga-FAPI PET/CT may provide higher specificity by selectively visualizing the tumor microenvironment, potentially reducing false-positive results and improving recurrence detection.

Building on these promising findings, we hypothesized that ⁶⁸Ga-FAPI PET/CT would outperform ¹⁸F-FDG PET/CT in recurrent NPC surveillance. Therefore, this study aimed to compare the diagnostic performance of ⁶⁸Ga-FAPI PET/CT, plasma EBV DNA, contrast-enhanced MRI/CT, and ¹⁸F-FDG PET/CT for the detection and surveillance of NPC. A secondary objective was to compare tracer uptake between ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT.

Materials and Methods

Participants

This prospectively registered comparative trial (Chinese Clinical Trial Registry identifier no. ChiCTR2100054163) was conducted at Fudan University Shanghai Cancer Center from July 2020 to November 2023, and the data were analyzed cross-sectionally for the present study. The study was approved by the Fudan University Shanghai Cancer Center institutional review board (approval no. 2109243-25). Written informed consent was obtained from each participant. The inclusion criteria were as follows: (a) newly diagnosed or previously treated NPC, (b) underwent both ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT scans, (c) underwent contrast-enhanced MRI or CT, and (d) available plasma EBV DNA data. Participants with a second malignant tumor and unavailable clinical data were excluded.

The Standards for Reporting of Diagnostic Accuracy checklist is included in Appendix S1, and the flow diagram of participant inclusion and exclusion is shown in Figure 1. The 10 overlapping participants were originally reported in a prior study that broadly compared ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT across various head and neck cancers of unknown primary origin (16). The present study, in contrast, focuses exclusively on NPC and extends the comparison by incorporating serum EBV DNA level and contrast-enhanced MRI/CT alongside both PET tracers. This further analysis therefore addresses a distinct clinical question with an expanded methodologic scope in a dedicated NPC cohort.

Figure 1:

Standards for Reporting of Diagnostic Accuracy Studies (ie, STARD) flow diagram of participant selection. FAPI = fibroblast activation protein inhibitor, ¹⁸F = fluorine 18, FDG = fluorodeoxyglucose, ⁶⁸Ga = gallium 68.

EBV DNA Measurement

Plasma EBV DNA levels were quantified using quantitative real-time polymerase chain reaction with a commercial kit (Sansure Biotech), following the manufacturer’s instructions. A viral load > 500 copies/mL was defined as positive.

CT and MRI Scan Performance

Contrast-enhanced CT scans were performed using a 64-section multidetector CT scanner (Sensation 64; Siemens Medical Solutions). The acquisition parameters were set at 120 kVp and 200 mA, with a detector configuration of 64 × 0.6 mm. After an initial unenhanced scan, 100 mL of a nonionic iodinated contrast agent was administered intravenously at 3 mL/sec. Triple-phase enhanced scanning was then initiated at 30 seconds (arterial phase), 80 seconds (portal venous phase), and 5 minutes (delayed phase). All images were reconstructed with a section thickness of 1 mm and a 512 × 512 matrix.

MRI was performed using a 3-T scanner (MAGNETOM Skyra; Siemens Medical Solutions) equipped with a dedicated 16-channel head/neck coil. The protocol comprised axial T1-weighted turbo spin-echo (repetition time msec/echo time msec: 500/6.5), T2-weighted turbo spin-echo (2500/78), and contrast-enhanced fat-saturated T1-weighted three-dimensional fast low-angle shot sequences with an isotropic resolution of 0.7 × 0.7 × 3 mm³. Key parameters for the turbo spin-echo sequences included a section thickness of 4.5 mm with a 0.9-mm gap and field of view of 270 mm. A bolus of gadolinium-based contrast agent (Magnevist; Bayer HealthCare) was injected intravenously at 0.1 mmol per kilogram of body weight before the fast low-angle shot acquisition.

Radiopharmaceuticals and PET/CT Acquisition

¹⁸F-FDG and ⁶⁸Ga-FAPI-04 were synthesized at our center according to a previously described protocol (17). The final radiopharmaceutical products exhibited radiochemical purity greater than 95%.

For PET/CT acquisition (Biograph mCT Flow scanner; Siemens Medical Solutions), participants fasted for at least 6 hours before ¹⁸F-FDG administration, whereas fasting was not required for ⁶⁸Ga-FAPI administration. PET/CT scanning was performed approximately 60 minutes after intravenous injection of 244.00 MBq ± 33.90 (SD) of ¹⁸F-FDG or 156.42 MBq ± 20.57 of ⁶⁸Ga-FAPI. Both radiotracer PET/CT examinations were completed within 1 week according to the previously described protocol (17). Briefly, CT scans were first performed from the proximal thighs to the head using 120 kV, CareDOSE 4D, section thickness of 3.0 mm, and a 0.33-second rotation time and were reconstructed to a 512 × 512 matrix (voxel size, 0.98 × 0.98 × 3.0 mm³). Subsequently, PET scans were performed at a bed speed of 2 mm/sec. PET images were reconstructed using a TrueX+TOF algorithm (two iterations, 21 subsets, and 5-mm full width at half-maximum Gaussian filter), with CT data applied for attenuation correction.

Imaging Interpretation

Fusion PET/CT images were reviewed and analyzed on a multimodality workstation (Syngo; Siemens Medical Solutions). Three experienced nuclear medicine physicians (Z.Y. and S.S., with >15 years of experience in PET/CT and B.G., with >10 years of experience in PET/CT) analyzed and interpreted the ¹⁸F-FDG, ⁶⁸Ga-FAPI, contrast-enhanced MR, and CT images independently. Reviewers were blinded to both the results of the other imaging modalities and the clinical information of the participants. In cases of disagreement, consensus was reached through a multidisciplinary conference.

Lesions showing increased radiotracer uptake relative to surrounding normal tissue and exhibiting malignant features (eg, mass, invasion, and destruction) at PET/CT were classified as suspected malignant lesions. Lymph nodes were considered metastatic on contrast-enhanced MR or CT images if they met at least one of the following criteria: (a) presence of necrosis or extracapsular spread, (b) short-axis diameter ≥ 5 mm in the retropharyngeal region or ≥10 mm in other regions, and (c) cluster of ≥3 lymph nodes of borderline size. All lesions were verified with biopsy or 6-month follow-up. Follow-up imaging at approximately 3 months and 6 months consisted of contrast-enhanced MRI for the nasopharynx and neck, while contrast-enhanced CT/MRI or ¹⁸F-FDG PET/CT was used for distant sites as clinically indicated. A lesion was classified as benign if it met any one of the following criteria: (a) complete resolution without any antineoplastic therapy and (b) no substantial change in size over at least 6 months. Lesions were deemed malignant during follow-up if they met at least one of the following criteria: (a) demonstration of typical malignant features and (b) substantial reduction or progression in size after antineoplastic therapy confirmed with follow-up imaging (17). TNM staging was performed according to the eighth edition of the American Joint Committee on Cancer/Union for International Cancer Control staging system.

For PET/CT quantitative analysis, the maximum standardized uptake value (SUV_max), normalized to body weight, was manually computed by drawing a three-dimensional volume of interest around each malignant lesion. The mean SUV (SUV_mean) and metabolic tumor volume (MTV) were recorded using absolute and relative SUV thresholds of 2.5 and 40%, respectively. The choice of an absolute SUV threshold of 2.5 combined with a relative threshold of 40% of SUV_max for MTV delineation was based on established methods in NPC PET imaging literature (18). Total lesion glucose (TLG) or total lesion FAP (TLF) was calculated as: TLG/TLF = SUV_mean × MTV. The tumor-to-background ratio (TBR) was calculated as: TBR = SUV_max/bSUV_mean, where bSUV_mean denotes the mean SUV of the thoracic aorta.

Statistical Analysis

The McNemar test was used to compare differences in sensitivity, specificity, and accuracy. Differences of semiquantitative parameters between ¹⁸F-FDG and ⁶⁸Ga-FAPI were evaluated using paired Wilcoxon signed rank test. Simple linear regression analysis was applied to assess relationships between plasma EBV DNA and semiquantitative parameters of ¹⁸F-FDG and ⁶⁸Ga-FAPI. GraphPad Prism (version 10.2.0) was used for statistical analyses. A two-tailed P value < .05 was considered statistically significant.

Results

Participant Characteristics

A total of 65 participants with NPC (15 female and 50 male; median age, 50 years [IQR, 41.50–61 years]) were enrolled in this prospective study (Fig 1, Table 1). Thirty-one participants were newly diagnosed with NPC (treatment-naive cohort), while 34 participants were previously diagnosed and underwent standard treatment (posttreatment cohort). All participants were diagnosed with EBV before the initial treatment, confirmed by plasma EBV DNA level, serum antibody test, or EBV-encoded small RNA level. The median plasma EBV DNA level was 1530 copies/mL (IQR, 22–8115 copies/mL) for all participants. Interestingly, the treatment-naive cohort showed significantly higher plasma EBV DNA level than the posttreatment cohort (4710 vs 811.50 copies/mL, P < .001). The detailed characteristics of each participant are presented in Tables S1 and S2.

Table 1:

Clinical Characteristics of Participants with Nasopharyngeal Carcinoma

Characteristic	Total (n = 65)	Treatment-naive Cohort (n = 31)	Posttreatment Cohort (n = 34)
Age (y)*	50 [41.50–61]	49 [41–63]	50.50 [41.75–59.50]
Sex
Male	50/65 (77)	24/31 (77)	26/34 (76)
Female	15/65 (23)	7/31 (23)	8/34 (24)
BMI (kg/m2)†	22.67 (16.98–28.73)	23.05 (16.98–27.34)	22.32 (16.98–28.73)
Plasma EBV DNA (copies/mL)*	1530 [22–8115]	4710 [1680–24900]	811.50 [0–1507.50]

Note.—Unless otherwise indicated, values are numerators and denominators, with percentages in parentheses. BMI = body mass index (calculated as weight in kilograms divided by height in meters squared), EBV = Epstein-Barr virus.

^*Values are medians, with IQRs in brackets.

^† Values are means, with ranges in parentheses.

Comparison of Diagnostic Performance for Identification of NPC across Plasma EBV DNA Level, Contrast-enhanced MRI/CT, and ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT

All the 31 participants from the treatment-naive cohort were pathologically diagnosed with NPC, while 24 of 34 participants from the posttreatment cohort were confirmed as having recurrence with biopsy (five participants) or 6-month follow-up (19 participants). Compared with plasma EBV DNA level, contrast-enhanced MRI/CT, and ¹⁸F-FDG PET/CT, ⁶⁸Ga-FAPI PET/CT demonstrated the highest sensitivity and accuracy rate in identifying the participants with de novo (sensitivity, 100% [31 of 31] vs 84% [26 of 31], 77% [24 of 31], and 84% [26 of 31], P = .06, .02, and .06, respectively; accuracy rate, 100% [31 of 31] vs 84% [26 of 31], 77% [24 of 31], and 84% [26 of 31], P = .06, .02, and .06, respectively; Table 2, Fig 2, and Fig S1) and recurrent (sensitivity, 100% [24 of 24] vs 54% [13 of 24], 58% [14 of 24], and 96% [23 of 24], P < .001, P = .002, and .50, respectively; accuracy rate, 94% [32 of 34] vs 53% [18 of 34], 68% [23 of 34], and 85% [29 of 34], P = .001, .02, and .25, respectively; Fig 3) NPC.

Table 2:

Participant-based Diagnostic Performance of Plasma EBV DNA Level, Contrast-enhanced MRI/CT, and ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT in Identifying Participants with De Novo and Recurrent Nasopharyngeal Carcinoma

Test Characteristic by Cohort	Participant-based Analysis				Comparison with 68Ga-FAPI (P Value)
	EBV DNA	Contrast-enhancedMRI/CT	18F-FDG	68Ga-FAPI	EBV DNA	Contrast-enhancedMRI/CT	18F-FDG
Treatment-naive cohort
True positive	26	24	26	31
False positive	0	0	0	0
True negative	0	0	0	0
False negative	5	7	5	0
Sensitivity (%)	84 (26/31) [66, 94]	77 (24/31) [58, 90]	84 (26/31) [66, 94]	100 (31/31) [86, 100]	.06	.02	.06
Specificity (%)	NA	NA	NA	NA	NA	NA	NA
PPV (%)	100 (26/26) [84, 100]	100 (24/24) [83, 100]	100 (26/26) [84, 100]	100 (31/31) [86, 100]	…	…	…
NPV (%)	NA	NA	NA	NA	…	…	…
Accuracy rate (%)	84 (26/31) [66, 94]	77 (24/31) [58, 90]	84 (26/31) [66, 94]	100 (31/31) [86, 100]	.06	.02	.06
Posttreatment cohort
True positive	13	14	23	24
False positive	5	1	4	2
True negative	5	9	6	8
False negative	11	10	1	0
Sensitivity (%)	54 (13/24) [33, 74]	58 (14/24) [37, 77]	96 (23/24) [77, 100]	100 (24/24) [83, 100]	<.001	.002	.50
Specificity (%)	50 (5/10) [20, 80]	90 (9/10) [54, 99]	60 (6/10) [27, 86]	80 (8/10) [44, 96]	.45	>.99	.50
PPV (%)	72 (13/18) [46, 89]	93 (14/15) [66, 100]	85 (23/27) [65, 95]	92 (24/26) [73, 99]	…	…	…
NPV (%)	31 (5/16) [12, 59]	47 (9/19) [25, 71]	86 (6/7) [42, 99]	100 (8/8) [60, 100]	…	…	…
Accuracy rate (%)	53 (18/34) [35, 70]	68 (23/34) [49, 82]	85 (29/34) [68, 94]	94 (32/34) [79, 99]	.001	.02	.25

Note.—Unless otherwise indicated, values are numbers, data in parentheses are numerators and denominators, and data in brackets are 95% CIs. EBV = Epstein-Barr virus, FAPI = fibroblast activation protein inhibitor, ¹⁸F = fluorine 18, FDG = fluorodeoxyglucose, ⁶⁸Ga = gallium 68, NA = not applicable, NPV = negative predictive value, PPV = positive predictive value.

Figure 2:

PET/CT and MR images in a 66-year-old male participant (participant 18) with pathologically confirmed metastatic poorly differentiated carcinoma of left neck. The plasma EBV DNA load was 189 000 copies/mL. (A) ¹⁸F-FDG PET/CT image reveals multiple lymph node and bone metastases with intensive metabolic activity (black and white arrows) but no metabolic activity in the nasopharynx (red dashed circle). (B) ⁶⁸Ga-FAPI PET/CT image also demonstrates the lymph node and bone metastases with high ⁶⁸Ga-FAPI activity (black and white arrows). Interestingly, there was moderate uptake of ⁶⁸Ga-FAPI in the left pharyngeal recess (red arrow; SUV_max, 3.22). (C) T1-weighted, T2-weighted, contrast-enhanced (C+) T1-weighted, and DWI MR images present no evidence for primary tumor in the nasopharynx (red dashed circle). Subsequent nasopharyngeal biopsy findings confirmed undifferentiated nonkeratinizing nasopharyngeal carcinoma. DWI = diffusion-weighted imaging, EBV = Epstein-Barr virus, FAPI = fibroblast activation protein inhibitor, ¹⁸F = fluorine 18, FDG = fluorodeoxyglucose, ⁶⁸Ga = gallium 68, SUV_max = maximum standardized uptake value.

Figure 3:

PET/CT and MR images in a 66-year-old male participant (participant 54) with nasopharyngeal carcinoma completing treatment for 19 years. The plasma EBV DNA load was 0 copy/mL. (A) ¹⁸F-FDG PET/CT image reveals intensive metabolic activity of nasopharynx (black and white arrows) and moderate metabolic activity of submental lymph node (red arrow; SUV_max, 2.9). (B) ⁶⁸Ga-FAPI PET/CT image also demonstrates intensive uptake of ⁶⁸Ga-FAPI in the nasopharynx (black and white arrows) but no intensive uptake of ⁶⁸Ga-FAPI in the submental lymph node (red arrow). (C) MR images show enlarged submental lymph node (short diameter, 0.84 cm) with hypointensity on T1-weighted image and hyperintensity on T2-weighted image, contrast-enhanced (C+) T1-weighted image, and DW image (red arrow). However, biopsy of the submental lymph node demonstrated no evidence for metastasis. DWI = diffusion-weighted imaging, EBV = Epstein-Barr virus, FAPI = fibroblast activation protein inhibitor, ¹⁸F = fluorine 18, FDG = fluorodeoxyglucose, ⁶⁸Ga = gallium 68, SUV_max = maximum standardized uptake value.

Overall, 287 lesions (269 malignant lesions and 18 benign lesions) from the treatment-naive cohort and 79 lesions (64 malignant lesions and 15 benign lesions) from the posttreatment cohort were included in the analysis. ¹⁸F-FDG PET/CT revealed the highest sensitivity compared with contrast-enhanced MRI/CT and ⁶⁸Ga-FAPI PET/CT in identifying the malignant lesions for the treatment-naive cohort (Table 3; 97% [262 of 269] vs 81% [205 of 252] and 96% [257 of 269], P < .001 and P = .36, respectively), whereas ⁶⁸Ga-FAPI PET/CT showed the highest sensitivity for the posttreatment cohort (97% [62 of 64] vs 57% [35 of 61] and 83% [53 of 64], P < .001 and P = .02, respectively). Furthermore, ⁶⁸Ga-FAPI PET/CT showed the highest accuracy rate in detecting the primary and metastatic lesions of both cohorts compared with contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT (96% [275 of 287] vs 83% [223 of 270] and 91% [262 of 287], P < .001 and P = .047, respectively; 92% [73 of 79] vs 59% [47 of 79] and 67% [53 of 79], P < .001 and <.001, respectively).

Table 3:

Lesion-based Diagnostic Performance of Plasma EBV DNA Load, Contrast-enhanced MRI/CT, and ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT in Identifying Participants with De Novo and Recurrent Nasopharyngeal Carcinoma

Test Characteristic by Cohort	Lesion-based Analysis			Comparison with 68Ga-FAPI (P Value)
	Contrast-enhanced MRI/CT	18F-FDG	68Ga-FAPI	Contrast-enhanced MRI/CT	18F-FDG
Treatment-naive cohort
True positive	205	262	257
False positive	0	18	0
True negative	18	0	18
False negative	47	7	12
Sensitivity (%)	81 (205/252) [76, 86]	97 (262/269) [94, 99]	96 (257/269) [92, 98]	<.001	.35
Specificity (%)	100 (18/18) [78, 100]	0 (0/18) [0, 22]	100 (18/18) [78, 100]	NA	<.001
PPV (%)	100 (205/205) [98, 100]	94 (262/280) [90, 96]	100 (257/257) [98, 100]	…	…
NPV (%)	28 (18/65) [18, 40]	0 (0/7) [0, 44]	60 (18/30) [41, 77]	…	…
Accuracy rate (%)	83 (223/270) [77, 87]	91 (262/287) [87, 94]	96 (275/287) [93, 98]	<.001	.047
Posttreatment cohort
True positive	35	53	62
False positive	6	15	4
True negative	12	0	11
False negative	26	11	2
Sensitivity (%)	57 (35/61) [44, 70]	83 (53/64) [71, 91]	97 (62/64) [88, 99]	<.001	.02
Specificity (%)	67 (12/18) [41, 86]	0 (0/15) [0, 25]	73 (11/15) [45, 91]	.69	<.001
PPV (%)	85 (35/41) [70, 94]	78 (53/68) [66, 87]	94 (62/66) [84, 98]	…	…
NPV (%)	32 (12/38) [18, 49]	0 (0/11) [0, 32]	85 (11/13) [54, 97]	…	…
Accuracy rate (%)	59 (47/79) [48, 70]	67 (53/79) [56, 77]	92 (73/79) [84, 97]	<.001	<.001

Note.—Unless otherwise indicated, values are numbers, data in parentheses are numerators and denominators, and data in brackets are 95% CIs. EBV = Epstein-Barr virus, FAPI = fibroblast activation protein inhibitor, ¹⁸F = fluorine 18, FDG = fluorodeoxyglucose, ⁶⁸Ga = gallium 68, NA = not applicable, NPV = negative predictive value, PPV = positive predictive value.

With regard to TNM staging, ⁶⁸Ga-FAPI PET/CT led to upstaging of five of 31 (16%) participants and downstaging of four of 31 (13%) participants in the treatment-naive cohort compared with contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT, respectively (Tables 4, S1). In the posttreatment cohort, ⁶⁸Ga-FAPI PET/CT resulted in upstaging in nine of 34 (26%) and one of 34 (3%) participants and downstaging in three of 34 (9%) and five of 34 (15%) participants compared with contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT, respectively (Tables 4, S2). However, ⁶⁸Ga-FAPI PET/CT also underestimated the TNM stage in two of 31 (6%) participants from the treatment-naive cohort and overestimated in two of 34 (6%) participants from the posttreatment cohort compared with ¹⁸F-FDG PET/CT and contrast-enhanced MRI/CT, respectively.

Table 4:

Comparison of TNM Staging with Contrast-enhanced MRI/CT and ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT in Participants with De Novo and Recurrent Nasopharyngeal Carcinoma

Staging Change by Cohort	68Ga-FAPI vs Contrast-enhanced MRI/CT	68Ga-FAPI vs 18F-FDG
Treatment-naive cohort
Up	5/31 (16)	0/31 (0)
Down	0/31 (0)	4/31 (13)
Overestimation	0/31 (0)	0/31 (0)
Underestimation	0/31 (0)	2/31 (6)
None	26/31 (84)	25/31 (81)
Posttreatment cohort
Up	9/34 (26)	1/34 (3)
Down	3/34 (9)	5/34 (15)
Overestimation	2/34 (6)	0/34 (0)
Underestimation	0/34 (0)	0/34 (0)
None	20/34 (59)	28/34 (82)

Note.—Values are numerators and denominators, with percentages in parentheses. FAPI = fibroblast activation protein inhibitor, ¹⁸F = fluorine 18, FDG = fluorodeoxyglucose, ⁶⁸Ga = gallium 68.

Comparison of ¹⁸F-FDG and ⁶⁸Ga-FAPI Uptake in Primary and Metastatic Lesions

In the treatment-naive cohort, ⁶⁸Ga-FAPI PET/CT helped identify all the primary tumors with lower SUV_max but significantly higher TBR than ¹⁸F-FDG PET/CT (31 vs 26, 7.67 ± 4.37 vs 9.13 ± 8.48, P = .52; and 8.53 ± 6.63 vs 5.79 ± 5.03, P = .001, respectively; Fig 2, Table S3). However, ¹⁸F-FDG PET/CT helped detect more metastatic lymph nodes with significantly higher SUV_max and TBR compared with ⁶⁸Ga-FAPI PET/CT (172 vs 162, 12.38 ± 8.75 vs 7.47 ± 3.70, P < .001; and 8.62 ± 5.48 vs 8.18 ± 5.15, P = .79, respectively). In the posttreatment cohort, ⁶⁸Ga-FAPI PET/CT better demonstrated liver metastases compared with ¹⁸F-FDG PET/CT, with significantly higher SUV_max and TBR (15 vs seven, 4.90 ± 4.48 vs 2.64 ± 3.38, P = .001; and 3.67 ± 3.14 vs 1.44 ± 1.77, P < .001, respectively).

Among all the primary and metastatic lesions, ¹⁸F-FDG PET/CT showed significantly higher semiquantitative values of SUV _max, SUV_mean with an absolute threshold of 2.5 (SUV_mean2.5), TLG/TLF with an absolute threshold of 2.5 (TLG/TLF_2.5), and SUV_mean with a relative threshold of 40% (SUV_mean40%) than ⁶⁸Ga-FAPI PET/CT (9.50 ± 7.74 vs 6.87 ± 3.87, P < .001; 4.57 ± 2.51 vs 3.83 ± 1.39, P < .001; 33.82 ± 75.12 vs 31.99 ± 81.68, P = .04; 5.79 ± 4.85 vs 4.12 ± 2.31, P < .001, respectively; Table S4). ⁶⁸Ga-FAPI PET/CT outperformed ¹⁸F-FDG PET/CT in terms of TBR and MTV with a relative threshold of 40% (MTV_40%) (7.10 ± 5.13 vs 6.45 ± 5.04, P < .001; and 4.60 ± 6.66 vs 3.49 ± 4.49, P < .001).

Assessment of Correlation between Plasma EBV DNA Level and Semiquantitative Parameters Detected at ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT

As shown in Figure S2, the semiquantitative parameters SUV_max, TBR, SUV_mean2.5, TLG_2.5, and SUV_mean40% of the lesion with the highest SUV at ¹⁸F-FDG PET/CT were significantly correlated with plasma EBV DNA levels (R² = 0.07, P = .04; R² = 0.15, P = .002; R² = 0.08, P = .02; R² = 0.07, P = .03; and R² = 0.07, P = .03, respectively). In terms of the average semiquantitative values of the lesions detected at ¹⁸F-FDG PET/CT, only TBR was significantly associated with the plasma EBV DNA level (R² = 0.07, P = .03; Figure S3). Furthermore, the semiquantitative values of the lesions detected at ⁶⁸Ga-FAPI PET/CT showed no relationship with the plasma EBV DNA level (P > .05).

Discussion

In this prospective study, we compared the diagnostic performance of plasma EBV DNA level, contrast-enhanced MRI/CT, and ¹⁸F-FDG and ⁶⁸Ga-FAPI PET/CT in participants with either de novo or recurrent NPC. Our results demonstrated that ⁶⁸Ga-FAPI PET/CT presented the highest diagnostic accuracy in diagnosis (100% [31 of 31] vs 84% [26 of 31], 77% [24 of 31], and 84% [26 of 31], P = .06, .02, and .06, respectively) and surveillance (94% [32 of 34] vs 53% [18 of 34], 68% [23 of 34], and 85% [29 of 34], P = .001, .02, and .25, respectively) of NPC compared with plasma EBV DNA level, contrast-enhanced MRI/CT, and ¹⁸F-FDG PET/CT, suggesting the potential of ⁶⁸Ga-FAPI PET/CT as a promising imaging modality for initial assessment and recurrence surveillance of NPC.

Metastatic squamous cell carcinoma of the neck is the primary and most common manifestation of all head and neck cancers, for example, NPC, oropharyngeal carcinoma, and laryngocarcinoma (19). Combining plasma EBV DNA test and contrast-enhanced MRI, most cases of NPC could be diagnosed accurately in daily clinical practice. However, about 30% of patients with NPC present with occult primary tumors at contrast-enhanced MRI or ¹⁸F-FDG PET/CT examination, which manifest as small primary tumor with large metastases (12). In the current study, plasma EBV DNA level was false negative in five of 31 (16%) participants with de novo NPC, while contrast-enhanced MRI missed seven of 31 (23%) primary tumors. Moreover, in the posttreatment cohort, 11 of 34 (32%) and 10 of 34 (29%) false-negative and five of 34 (15%) and one of 34 (3%) false-positive cases were detected with plasma EBV DNA level and contrast-enhanced MRI, respectively. Previous research (20) showed that positive plasma EBV DNA levels were detected in only 61% of participants with recurrent NPC, which could be addressed by a small amount of tumor cell death that released low free EBV DNA in the plasma. Furthermore, limited to anatomic imaging, contrast-enhanced MRI/CT performs worse in identifying small lesions with diameter < 10 mm (7). Therefore, only combining the two examinations is far from meeting the individual requirement for participants with NPC.

Tumor stroma, accounting for a majority of solid tumor mass, plays an important role in tumor cell growth, invasion, and migration (21). By imaging tumor stroma rather than tumor, ⁶⁸Ga-FAPI PET/CT demonstrates superiority in identifying the primary and metastatic lesions over contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT (16,17). In line with previous studies (12–14), our results showed that ⁶⁸Ga-FAPI PET/CT helped detect all (100%, 31 of 31) primary tumors of NPC in the treatment-naive cohort, while contrast-enhanced MRI and ¹⁸F-FDG PET/CT missed the primary tumor in seven of 31 (23%) and five of 31 (16%) participants, respectively. Our representative images (Fig 2) showed that the primary tumor could manifest a paucity of ¹⁸F-FDG uptake even with multiple intensive ¹⁸F-FDG–avid metastases in participants with NPC. Furthermore, inflammatory ¹⁸F-FDG uptake may conceal small primary tumors, misguiding the following endoscopic biopsy (Figure S1). Nevertheless, ⁶⁸Ga-FAPI PET/CT could overcome these limitations in detecting the primary tumors of NPC. The false-positive ⁶⁸Ga-FAPI uptake observed in two posttreatment cases is likely attributable to postradiation therapy fibrosis. Radiation-induced tissue injury activates local fibroblasts, leading to FAP expression, which can be avidly targeted by the tracer. This highlights a key diagnostic challenge, as benign fibroblastic activity, a recognized pitfall of FAPI imaging, can contribute to false-positive interpretations in the posttreatment setting (22).

Cancer-associated fibroblasts could promote and compose metastasis by activating the JAK/STAT3 and TGF-β signaling pathways (23). Thus, imaging targeting FAP could significantly improve the detection rate of metastatic lesions compared with ¹⁸F-FDG PET/CT among various cancers (24–26). In the current study, ⁶⁸Ga-FAPI PET/CT helped detect most of the malignant lesions with highest sensitivity (97% [62 of 64] vs 57% [35 of 61] and 83% [53 of 64], P < .001 and P = .02, respectively) and accuracy rate (92% [73 of 79] vs 59% [47 of 79] and 67% [53 of 79], P < .001 and <.001, respectively) compared with contrast-enhanced MRI/CT and ¹⁸F-FDG PET/CT in the posttreatment cohort. In addition, the low background uptake of bone and liver could improve the detection accuracy of positive lesions with ⁶⁸Ga-FAPI compared with ¹⁸F-FDG PET/CT (Table S3) (27). The most clinically significant implication emerges when ⁶⁸Ga-FAPI PET/CT helps identify occult distant metastases missed with conventional imaging, resulting in disease upstaging from M0 to M1. This pivotal finding would fundamentally alter participant management, typically shifting the therapeutic goal from curative-intent, local-regional regimens (eg, definitive chemoradiotherapy) to palliative systemic therapy or metastasis-directed approaches. Conversely, downstaging (eg, from M1 to M0) with ⁶⁸Ga-FAPI PET/CT can validate and reinforce the decision to proceed with aggressive curative-intent strategies. Ultimately, the precise staging afforded by ⁶⁸Ga-FAPI PET/CT directly informs critical therapeutic decisions, optimizing management by aligning it with the true extent of disease (28).

Regarding the detection of regional lymph node metastases, the performance of ⁶⁸Ga-FAPI PET/CT varies among different studies (12–14,29). Zhao et al (12) reported that ⁶⁸Ga-FAPI PET/CT could help detect more metastatic regional lymph nodes (95% [115 of 121] vs 75% [91 of 121], P < .001) with significantly higher radiotracer uptake (11.12 vs 6.53, P < .001) than ¹⁸F-FDG PET/CT. However, our current study showed that ⁶⁸Ga-FAPI PET/CT helped identify fewer positive regional lymph nodes (162 vs 172) and lower radiotracer uptake (7.47 ± 3.70 vs 12.38 ± 8.75, P < .001) in the treatment-naive cohort compared with ¹⁸F-FDG PET/CT, which was in line with Qin et al’s research (29). Nevertheless, among these research studies, most of the lymph nodes were regarded as metastases by radiographic follow-up, which may cause diagnostic bias. In a pathology-proven prospective trial (30), ⁶⁸Ga-FAPI PET/CT demonstrated better agreement with surgical pathology in accurately diagnosing regional lymph nodes of head and neck squamous cell carcinoma than ¹⁸F-FDG PET/CT (69% [20 of 29] vs 31% [nine of 29], P = .001). Chen et al (31) also reported the superiority of ⁶⁸Ga-FAPI PET/CT in diagnosing metastatic neck lymph nodes with higher specificity (100% [17 of 17] vs 29% [five of 17], P < .001) and accuracy (97% [35 of 36] vs 67% [24 of 36], P = .003) over ¹⁸F-FDG PET/CT in participants with oral squamous cell carcinoma, which were well consistent with the pathologic results. The observed variance in nodal detection may be attributed to a key limitation of our imaging-based reference standard, which likely favored ¹⁸F-FDG. Its known inflammatory uptake could inflate apparent sensitivity, while ⁶⁸Ga-FAPI’s true-negative results in reactive nodes might be misclassified as false-negative results. This bias underscores why pathology-proven studies consistently report ⁶⁸Ga-FAPI’s superior specificity. In addition, our representative case (Fig 3) showed that ⁶⁸Ga-FAPI PET/CT presented true-negative uptake in the submental lymph node and led to downstaging compared with ¹⁸F-FDG PET/CT and contrast-enhanced MRI (32). Thus, the higher specificity of ⁶⁸Ga-FAPI in assessing the regional lymph nodes may avoid unnecessary treatment for participants with NPC.

The plasma EBV DNA load has been reported as an individual prognostic indicator for survival prediction in participants with NPC (33). In the current study, our results showed that the plasma EBV DNA load was significantly correlated with the semiquantitative parameters detected at ¹⁸F-FDG PET/CT but not with that of ⁶⁸Ga-FAPI PET/CT. This may be caused by the fact that EBV mainly affected the tumor cells rather than the tumor stroma cells (34). The observed “uncoupling” between ⁶⁸Ga-FAPI uptake and plasma EBV DNA load may translate into direct clinical complementarity. Crucially, ⁶⁸Ga-FAPI PET/CT can provide critical evidence of relapse in cases of EBV DNA–negative recurrence, where the viral marker may be undetectable due to biologic or assay limitations. Conversely, when EBV DNA load is positive but conventional imaging findings are ambiguous, ⁶⁸Ga-FAPI PET/CT’s high specificity for tumor stroma helps confirm active malignancy, guiding biopsy or therapy. Thus, integrating both modalities can optimize posttreatment surveillance for NPC. Additionally, ⁶⁸Ga-FAPI PET/CT holds potential utility beyond endemic, EBV-driven NPC. Its mechanism is agnostic to viral etiology, making it particularly relevant for surveillance in nonendemic regions and for non-EBV subtypes, where standard biomarkers fail. Thus, it could become a pivotal tool in these settings. Future studies are needed to validate its efficacy in etiologically and geographically diverse populations.

The major limitation of this study was that not all the lesions detected with ¹⁸F-FDG or ⁶⁸Ga-FAPI PET/CT were confirmed with pathology. This creates a potential for incorporation bias, which may lead to an overestimation of the diagnostic performance. Though radiographic follow-up is widely accepted for differential diagnosis between malignant and benign, contradictory results from similar research may sometimes bring confusion to clinical application (12,29). However, it is impossible to biopsy multiple lesions from the same participant. Thus, standard protocol of radiographic follow-up is urgently needed. A further limitation of our study was the lack of a formal a priori sample size calculation. The sample size was determined by feasibility and the available cohort during the study period. While our results provide promising preliminary evidence, future studies with power-based sample size calculations are needed to confirm our findings.

In conclusion, this prospective study reemphasizes the value of initial assessment for participants with NPC and highlights the promising role of ⁶⁸Ga-FAPI PET/CT in recurrent surveillance. Our results suggest that ⁶⁸Ga-FAPI PET/CT holds substantial promise in the diagnostic workup of NPC, especially in challenging clinical scenarios in which conventional imaging yields equivocal findings. Prospective multicenter studies of larger cohorts are needed to validate these findings.

Disclosures of conflicts of interest

Please see ICMJE form(s) for author conflicts of interest. These have been provided as supplemental materials.