Pretreatment 18F-FDG PET/CT-Derived Parameters in Predicting Clinical Outcomes of Locally Advanced Upper Third Esophageal Squamous Cell Carcinoma After Definitive Chemoradiation Therapy

The aim of this study was to investigate whether standard uptake values (SUVs) of pretreatment 18F-FDG PET/CT were the surrogate parameters for predicting the outcomes in locally advanced esophageal squamous cell carcinoma patients treated with definitive chemoradiotherapy. Sixty patients with esophageal squamous cell carcinoma underwent pretreatment 18F-FDG PET/CT and received definitive chemoradiotherapy. 18F-FDG metabolic parameters including SUVmax, SUVmean, SULpeak, total lesion glycolysis (TLG), and metabolic tumor volume (MTV) of primary tumor were calculated. The receiver-operating characteristic (ROC) curve was used to determine the optimal cutoff value of FDG PET/CT-derived parameters that associated with treatment response. Estimating progression-free survival (PFS) and overall survival (OS) was analyzed by using Kaplan–Meier methods. Univariate and multivariate analysis for PFS and OS was performed using Cox regression. Complete response was achieved in 38.3%. The 4-year OS and PFS rates were 48.6% and 44.4%, respectively. SUVmean with a cutoff value of 6.1 could predict complete response with sensitivity of 69.6%, specificity of 78.4%, and accuracy of 75%. Cox multi-factor regression analyses revealed SUVmean > 6.1 as an independent prognostic factor for OS (HR = 6.74, p = 0.02) and PFS (HR = 6.53, p < 0.001). Our study suggests that SUVmean of the primary tumor in pretreatment 18F-FDG PET/CT may be used as an independent predictor in esophageal squamous cell carcinoma patients treated with definitive chemoradiotherapy.


Introduction
Esophageal cancer, one of the most common cancers in the world, has a poor prognosis and high mortality rate [1]. Among the histopathologic types of esophageal cancer, squamous cell carcinoma is more common than adenocarcinoma. In addition, squamous cell carcinoma has been shown highly sensitive to radiotherapy and chemotherapy [2,3]. Upper third esophageal was defined as cervical and upper thoracic esophageal tumor by American Joint Committee on Cancer (AJCC) 7th edition [4] which represents 10% of esophageal cancer [5]. Preoperative chemoradiation and surgery are considered curative treatments for the middle and lower third esophageal cancer. However, radical surgery for upper third esophageal cancer is challenging because of the high prevalence of complication and death. In such a scenario, definitive chemoradiation therapy (dCRT) has become a standard treatment for upper third esophageal cancer stage II and III with improving survival in comparison with surgery or radiation therapy alone [6][7][8][9][10]. 18 F-fluoro-2-deoxy-glucose positron emission tomography/computed tomography ( 18 F-FDG PET/CT) has been applied as a useful tool for staging in esophageal squamous cell carcinoma [11]. Functional imaging with 18 F-FDG PET/ CT has provided characterization of tumor metabolism as well as predicting treatment outcome after CRT. Quantitative parameters derived from 18 F-FDG PET/CT such as maximal standard uptake value (SUVmax), mean standard uptake value (SUVmean), metabolic volume (MTV), and total lesion glycolysis (TLG) seem to have potential in prediction for recurrence and progression in esophageal cancer [12]. SUVmax is the most common parameter used in clinical practice although its role in prediction of outcomes after CRT remains still controversial [13][14][15]. Some studies suggested MTV was more important than SUVmax and associated with poor prognosis [16,17]. Other studies reported that TLG might be the best parameter in prediction of recurrence and long-term survival of esophageal cancer [12,18]. However, few studies showed no relationship of SUVmax, MTV, and TLG with survival [19][20][21]. The aim of this study was to assess the value of 18 F-FDG PET/CTderived parameters SUVmax, SUVmean, SUVpeak, MTV, and TLG in prediction for treatment response and survival in advanced upper third esophageal squamous cell carcinoma.

Patients
All patients who were diagnosed with esophageal cancer at our institution from May 2017 to July 2021 were retrospectively reviewed. The inclusion criteria were as follows: (1) upper edge of esophageal tumor located between the lower borders of cricoid cartilage and carina, (2) tumor biopsy confirmed squamous cell carcinoma, (3) stage II-III according to AJCC 7, (4) age from 18 to 75, and (5) ECOG 0-2. The exclusion criteria were set as follows: (1) previous radiotherapy or chemotherapy, (2) other serious co-morbidities, and (3) inadequate follow-up data. Finally, 60 patients were included for study analysis.

Staging 18 FDG-PET/CT Before Chemoradiotherapy
All patients in this study underwent whole-body 18 F-FDG PET/CT for staging of esophageal carcinoma in the Department of Nuclear Medicine, the 108 Military Central Hospital, within 14 days before the first day of treatment. PET/CT scan was performed, using GE Discovery 710 (GE Healthcare, Milwaukee, WI, USA), according to the European Association of Nuclear Medicine (EANM) guidelines, version 2.0 [22]. For patient preparation, the serum glucose level was checked to exclude hyperglycemia. Afterward, the patients rested in the waiting room before intravenous injection of 2.5 MBq/kg body weight (± 10%) of 18 F-FDG. The parameters of the low dose CT scan were as follows: 120 kVp, modulated mAs, the helical slice thickness of 3.75 mm, and 0.5 s/rotation. PET images were reconstructed using an iterative algorithm with attenuation correction with CT.

Quantitative 18 F-FDG Metabolic Assessment
18 F-FDG PET/CT images were evaluated by two nuclear medicine physicians and the consensus was made. The volume of interest (VOI) was set manually to exclude adjacent physiological 18 F-FDG-avid structures on attenuationcorrected PET images at the AW workstation version 4.7 (GE Healthcare, Milwaukee, WI, USA). Then, the region of interest (ROI) in the esophageal lesions was assessed with reference to patient's symptoms, endoscopy, and CT imaging. The tumor volume was determined by iterative adaptive threshold segmentation provided by vendor (PETVCAR software, GE Healthcare). The iterative algorithm used a slope gradient vector algorithm which found a threshold value that separated the tumor from the background tissue by weighting the SUV maximum value within the bounding box by a "w" weight factor (where 0 ≤ w ≤ 1 with default value of 0.5). The tumor border was then automatically contoured and metabolic tumor volume (MTV) was obtained as tumor volume. Tumor length was measured on the sagittal plane of 18 F-FDG PET/CT. SUVmax and SUVmean were defined as the maximum and mean value of SUV in the tumor volume. SUVpeak was average SUV of a 1cm 3 spherical centered on the hottest point within the tumor. Total lesion glycolysis (TLG) was calculated as SUVmean multiplied by MTV. All 18 F-FDG PET/CT-derived parameters were computed by PETVCAR software (version 4.7, GE Healthcare, Milwaukee, WI, USA) [23].

Chemoradiation Therapy
Definitive chemoradiation therapy was approved by the tumor board in Oncology Institute of the 108 Military Central Hospital. Gross tumor volumes were identified by the combination of contrast-enhanced CT and PET/CT. Intensity-modulated radiation therapy (IMRT) with simultaneous integrated boost technique delivers a total dose of 60 Gy to the primary tumor and active lymph nodes and 50.4 Gy to regional lymph nodes in 28 fractions. Chemotherapy was administered with cisplatin 75 mg/m 2 day 1 plus 5FU 750 mg/m 2 from day 1 to 4 (weeks 1,5,9,13) or paclitaxel 50 mg/m 2 plus carboplatin AUC2 (days 1,8,15,22,29).

Treatment Response and Follow-up
Treatment response was assessed for 3 months after completion of CRT by computed tomography (CT) and endoscopy. Clinical complete response was defined as no evidence of tumor on endoscopy and CT scan (esophageal wall thickening ≤ 5 mm, lymph node < 10 mm in short distance, and no evidence of distant metastases) [24]. During the first 2 years after therapy, patients were followed up every 3 months with clinical examination, esophageal endoscopy, and chest abdominal CT. After 2 years, the patients underwent followup every 6 months. The primary endpoint of this study was overall survival (OS) and the secondary endpoint was progression-free survival (PFS). OS was defined as the time of the beginning of dCRT to death from any cause or the last day of clinical follow-up. PFS was defined as the time from the beginning of dCRT to the day of disease progression, death from any cause, or the last day of clinical follow-up. Toxicity was evaluated using National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0 [25].

Statistical Analysis
Commercial software packages were used for statistical analysis (SPSS v.20.0, IBM Corp). Categorical values were compared using the chi-square test or Fisher's exact test. Continuous variables with normal distribution were compared using paired Student t test or repeated measure ANOVA. Variables not following normal distribution were compared using Mann-Whitney U test [26]. The optimal cutoff value, sensitivity, and specificity of each 18 F-FDG PET/CT parameter for response prediction were calculated using receiver-operating characteristic (ROC) curve analysis, and diagnostic power was assessed using the area under the curve (AUC). The statistically significance of the difference between the areas under the ROC curve was analyzed using the DeLong method. Logistic regression analysis was used to determine the significant parameters to predict OS and PFS by Cox model. The predictive value of clinical factors and 18 F-FDG PET/CT parameters on the outcomes was assessed by univariate analysis. Quantitative PET parameters were displayed as binary variables. Variables with p < 0.2 in univariate analysis were further used in multivariate analysis. Estimating PFS and OS was analyzed by using Kaplan-Meier method. The significance threshold was set at p < 0.05 [27].

The Study Population
Patient characteristics are summarized in Table 1. Sixty consecutive patients, 100% male with mean age of 57.9 ± 6.9 (range 45-75), were included in this study. The proportion of histopathological grade 1-2 and clinical stage II-IIIA was

Discussion
This study focused on the treatment response status and prognosis of upper third esophageal squamous cell carcinoma patients who were indicated for dCRT as a standard treatment. In our study, the combination of modern radiotherapy technique (IMRT) and chemotherapy regimens paclitaxel/carboplatin may improve survival. The rate of local recurrence and death in our study was lower than those in other studies which including upper and middle thoracic esophageal cancer [28,29]. Pretreatment metabolic parameters to predict response and outcome after dCRT are important in management of upper third esophageal squamous cell carcinoma by estimating the local radiation dose escalation; additional salvage chemotherapy or follow-up could be estimated [30,31].  The present study demonstrated that SUVs (SUVmean, SUVpeak, and SUVmax) could predict treatment response to dCRT in squamous cell esophageal carcinoma. The tumor response status plays an important role in decision making after dCRT. There was evidence that tumor response status is an independent factor for predicting survival of cancer patients [32]. Previous study has proven that PET/CT is a useful tool for the therapeutic monitoring of esophageal cancer [33]. SUVmax and SUVpeak presented quantitative glucose metabolic of a part of tumor. Otherwise, SUVmean reflects the mean metabolic uptake of the whole tumor volume that could be more reliable than SUVmax and SULpeak in prediction of tumor response after dCRT. Mantziari et al. found that SUVmax and SUVmean could predict therapeutic response; meanwhile, the values of MTV and TLG were still controversial [36].
There is still debate about which primary parameters derived from FDG PET/CT should be used to predict survival for esophageal cancer. Our study reported that SUVmean could predict PFS and OS of upper third esophageal squamous cell carcinoma. In addition, the increase of SUVmean seen in the aggressive and hypoxia tumor cells suggests that SUVmean could predict the tumor progression, OS, and PFS as well [33,34]. In addition, the segmentation method may provide different values of SUVmean. In our study, the iterative adaptive segmentation method showed the improvement in reducing the errors and the variability of SUVmean rather than manual delineation or using fix threshold. Several studies have shown that TLG and MTV are more promising than SUVs (SUVmax, SUVmean, and SUVpeak) in survival prediction [16,[34][35][36]. TLG represents the entire metabolic volume of tumor and it might provide more accurate values than those of SUVmax which is measured by maximal uptake within a single voxel. However, Atsumi et al. reported that SUVmax was an independent factor in prediction of treatment response, PFS, OS, and local control in esophageal cancer [14]. The value of quantitative metabolic parameters from FDG PET/CT may depend on histopathologic features, necrosis, location, volume, length, and heterogeneity of tumor [17,37,38]. High SUVmax uptake level and MTV might be seen on T3/4 stage and higher level of TLG depends on T4 and N stage [39]. The limitation of our study is that the quantitative parameters of whole-body lesions were not accessed. A certain number of positive lymph node on PET/CT could not be biopsied and accounted for quantitative analysis. Hence, the whole-body tumor ROIs including lymph node were not applied in this study.

Conclusion
Pretreatment SUVmean derived from 18 F-FDG PET/CT may be a predictive value of OS and PFS in squamous cell carcinoma of upper third of esophagus after dCRT. In addition, SUVmean can predict treatment response with high sensitivity and specificity. Our study suggests the cutoff SUVmean of 6.1 can be used to predict clinical treatment response and stratify prognostic outcomes in locally advanced esophageal squamous cell carcinoma patients treated with dCRT.