Potentially eligible locally advanced esophageal squamous cell carcinoma (LA-ESCC) patients were recruited at Shandong Cancer Hospital and Institute from June 2021 to July 2022 ( Table 1 ). All patients volunteered to participate in this study, and the local ethics committee of Shandong Cancer Hospital and Institute approved the prospective study.

Patients were enrolled based on the following criteria: (1) histopathologically confirmed esophageal squamous cell carcinoma(T3~4N0~3M0); (2) age ≥ 18 years; (3) presence of measurable primary tumors;(4) PD-L1 expression assay was conducted and (5) ¹⁸F-FAPI-04 PET/CT scanning was performed.

The exclusion criteria included: (1) pregnancy or breastfeeding; and (2) unwillingness to participate or withdraw. The flow chart of research and design is shown in Figure 1 .

¹⁸F-FAPI-04 was synthesized as described previously (18). Patients were not required to fast or under blood glucose measurement before scanning. After intravenous injection of ¹⁸F-FAPI-04 (4.81 MBq/kg), the patients needed to rest for approximately 1 h. Scanning was then performed with two different PET/CT(GE MINI TF Big Bore; Philips Healthcare, Cleveland, OH, USA). Whole-body CT scans were obtained using a low-dose protocol (300 mAs, 120 kV, a 512 × 512 matrix, rotation time of 1.0 s, and pitch index of 0.688; reconstructed with a soft-tissue kernel to a slice thickness of 2 mm) for attenuation correction. PET data were acquired in three-dimensional mode using a 200 × 200 matrix with an imaging time of 1 min per bed position. During image acquisition, the patients maintained normal shallow breathing. Subsequently, after attenuation and correction (Biograph 3D iterative reconstruction software, time of flight [TOF] correction), we viewed attenuation-corrected PET images, CT images, and PET/CT fusion images.

The attenuation-corrected CT images, PET images, and fused PET/CT images were displayed in coronal, sagittal, and transaxial slices, which were viewed and analyzed on the Nuclear Medicine Information System (Beijing Mozi Healthcare Ltd, Beijing, China). Two experienced PET/CT physicians (J.Z. and J.R., with 18 and 6 years, respectively, of nuclear oncology experience) visually assessed the ¹⁸F-FAPI-04 PET/CT images and reached a consensus regarding the image interpretations for primary tumors. Regions of interest were drawn around tumor lesions with higher uptake in transaxial sections, and ¹⁸F-FAPI-04 PET/CT parameters were generated by an automated 3-dimensional contouring program with a 30% isocontour. The uptake values in the region of interest were normalised to the injected dose per kilogram of patient body weight, and the standardised uptake values were derived according to the following formula: [measured radioactivity concentration (Bq/mL) × body weight (g)]/injected radioactivity concentration (Bq). Regions of interest were drawn around the primary tumor lesion, and the obtained parameters, including SUVmax, SUVmean, SUVpeak, SUVsd, metabolic tumor volume (MTV), and total lesion FAP expression (TLF), were generated by an automated contouring program provided by the vendor. TLF (total lesion FAP expression) was calculated as the product of the SUVmean of the lesion and the MTV (TLF= SUVmean× MTV). We also measured the SUVmean of 1 cm³ areas in the ascending aorta, liver and Lumbar 5 (L5) vertebrae. The circular region of interest (ROI) of 1cm3 was drawn in the normal regions of segments VII and VIII of the liver. The average of the liver SUVmean was calculated. The ratio of the SUVmax of the primary tumor to the SUVmean of the normal tissue (blood, liver and L5 vertebrae) is then calculated and is called the tumor to background ratio (TBRblood, TBRliver and TBRbone). For controversial lesions, discussion among the imaging experts with consideration of the results from other imaging modalities proceeded until a final consensus was reached.

In our study, PD-L1 expression in all patients was obtained by gastroscopic biopsy for pathological tissue of esophageal cancer.

PD-L1 expression was assessed by CPS, which was defined as the number of PD-L1 stained cells (tumor cells, lymphocytes and macrophages) divided by the total number of surviving tumor cells multiplied by 100. The maximum CPS is defined as 100. All other cells, such as tumor-associated plasma cells, neutrophils, normal/non-neoplastic cells, and necrotic cells, were excluded from the evaluation. The cutoff value was determined according to an FDA-approved test and the guidelines of pembrolizumab treatment and separated into two classifications: negative (CPS < 10) and positive expression (CPS ≥ 10) (5, 17). Patients without sufficient viable tumor cells (<100) were excluded. Each slide was blindly given a CPS for PD-L1 expression by two experienced pathologists. Both hematoxylin–eosin (HE) staining and PD-L1 IHC staining were assessed to reach a final CPS value. Two experienced pathologists (D.Z. and H.J., with 25 and 22 years, respectively, of oncology experience) evaluated pathological slides. Each case has a final consistent result after discussion.

Statistical analyzes were performed using SPSS software (version 27.0 for Windows; SPSS INC.). Continuous data were described as the mean ± standard deviation (mean ± SD) or median and interquartile, depending on whether they followed a normal distribution. and non-normally distributed data (including MTV) was expressed as the median and interquartile. Comparisons of normally distributed data between the two groups were performed using a paired two-sample t test, and comparisons of non-normally distributed data between the two groups were performed using the Mann–Whitney U test. Binary logistic regression analyses were performed to ascertain the relationships between ¹⁸F-FAPI-04 PET/CT parameters, tumor location, degree of differentiation and PD-L1 expression. Receiver operating characteristic (ROC) curve analysis was used to determine the threshold values with the maximum Youden index of ¹⁸F-FAPI-04 PET/CT parameters for PD-L1 positive expression. Spearman rank correlation coefficients were calculated to assess the relationship between ¹⁸F-FAPI-04 PET/CT parameters and PD-L1 expression. All tests were two-sided, and a probability of less than 0.05 was considered statistically significant.

From June 2021 to July 2022, 24 patients diagnosed with LA-ESCC based on histological examinations at Shandong Cancer Hospital and Institute were enrolled in this study. The characteristics of the patients are presented in Table 1 . Among all patients, 17 patients were classified as negative expression (CPS < 10), and 7 patients as positive expression (CPS ≥ 10). Figure 2 shows representative ¹⁸F-FAPI-04 PET/CT imaging results for two cases classified as positive and negative PD-L1 expression.

The quantitative ¹⁸F-FAPI-04 PET/CT parameters SUVmax, SUVmean, SUVpeak, SUVsd, MTV and TLF are shown in Table 2 for all patients, negative (CPS < 10) and positive expression (CPS≥10) patients. SUVmax, SUVmean, SUVpeak, and SUVsd were significantly higher in positive expression patients than in negative (14.13 ± 4.41 vs. 10.61 ± 2.77, p = 0.027; 8.67 ± 1.97 vs. 5.74 ± 1.60, p<0.001; 11.15 ± 2.90 vs.7.77 ± 2.31, p = 0.006 and 2.57 ± 0.48 vs. 1.73 ± 0.59, p = 0.003) ( Table 2 ). None of the other parameters showed a significant difference between negative and positive PD-L1 expression.

ROC curves were generated to evaluate the predictive accuracy of ¹⁸F-FAPI-04 PET/CT parameters for identifying negative and positive expression patients ( Table 3 ; Figure 3 ). The AUC value for SUVmean (AUC = 0.882) was higher than those for SUVsd (AUC =0.874), SUVpeak (AUC =0.840) and SUVmax (AUC=0.765) ( Table 3 ), while the AUC values for all four parameters were significant (p = 0.004, p = 0.005, p = 0.010 and p = 0.045, respectively) ( Table 3 ). The cutoff values for SUVmean, SUVsd, SUVpeak and SUVmax, based on the Youden indexes, were 7.38, 1.90, 9.67 and 13.71, respectively ( Table 3 ).

We found a moderate correlation between SUVsd, SUVean, SUVPeak, SUVmax and PD-L1 expression (rs=0.584,p=0.003; rs=0.571,p=0.004; rs=0.511,p=0.011; rs=0.462,p=0.024, respectively). The correlation between ¹⁸F-FAPI-04PET/CT biomarkers extracted from tumor lesions and PD-L1 expression is shown in Table 4 .

According to univariate logistic regression analyses, SUVmean (p = 0.026), SUVpeak (p = 0.031), and SUVsd (p = 0.024) were independently associated with the PD-L1 expression in LA-ESCC patients ( Table 5 ). Due to the moderate positive correlation among SUVmean, SUVsd and SUVpeak, we only included SUVsd (the largest correlation coefficient) and TBRliver (p = 0.059) in the multivariate logistic analysis. Finally, only SUVsd (p = 0.035) was an optimum predictor of PD-L1 expression in these patients ( Table 5 ).