Development of a predictive model for lymph node metastasis in esophageal cancer using ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (FDG PET/CT) semi-quantitative parameters and tumor biomarkers

Introduction

As a malignant tumor, esophageal cancer (EC) ranks seventh in global incidence and sixth in cancer-related mortality, thus becoming a pressing public health problem (1). Based on histological features, EC can primarily be categorized into two types: esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC). ESCC represents about 90% of worldwide EC instances, exhibiting notable differences in regional distribution with higher rates found in East Africa, sub-Saharan Africa, and Central Asia. Conversely, EAC tends to be more common in European nations and affluent countries in North America. Its incidence has increased exponentially (by 300%) over the past four decades, a phenomenon closely related to the prevalence of risk factors such as gastroesophageal reflux disease and obesity (2,3). Due to the insidious early symptoms of EC, most patients are diagnosed at the middle or advanced stages. Approximately 30% (stage T3) to 50% (stage T4) of cases cannot achieve radical (R0) resection (4). Clinical studies have confirmed that compared with surgery alone, comprehensive treatment (including surgery combined with neoadjuvant/adjuvant chemoradiotherapy) can significantly improve the survival outcomes of patients with locally regional EC and has become the standard treatment model for resectable EC (5). Despite the continuous innovation of treatment methods, the 5-year overall survival (OS) rate of EC patients still remains at a relatively low level of 15% to 20% (6), indicating an urgent need to explore more effective prognostic prediction and treatment strategies.

Lymph node metastasis (LNM) is a key driving factor for the progression and poor prognosis of EC. Research data show that the incidence of LNM varies significantly among different pathological types and tumor stages. In EAC, the incidence of LNM for intramucosal (pT1a) lesions is less than 5%, while it surges to 26% for submucosal (pT1b) lesions. For ESCC patients, the risk of LNM at stage pT1a is approximately 4%, but it can reach as high as 30% at stage pT1b, which is significantly higher than that in EAC (7). In addition, the number of positive lymph nodes (≥4) and the proportion of positive lymph nodes (≥0.2) have been proven to be independent prognostic indicators for OS. The 5-year survival rates of patients at stage pN0 and stage pN+ are 63% and 30%, respectively (7,8), highlighting the core position of LNM in the prognostic assessment of EC. Therefore, accurately identifying LNM at an early stage and formulating individualized treatment plans are crucial steps for improving patients’ prognosis.

Prior studies have developed predictive models for EC-related LNM, but their performance and applicability remain limited. Models relying solely on clinical factors such as tumor diameter and tumor (T) stage yield low discriminative power, with area under the curve (AUC) values ranging from 0.65 to 0.80 (9,10), as they fail to capture the biological aggressiveness of tumors that drives metastasis. Single tumor biomarkers [e.g., carcinoembryonic antigen (CEA), carbohydrate antigen (CA) 19-9] also show modest predictive accuracy (AUC =0.70–0.75), due to low specificity in early-stage EC and potential false positives from benign conditions (10,11). The advanced molecular imaging technique known as ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) is crucial for staging tumors, defining radiotherapy target volumes, assessing treatment effectiveness, and tracking recurrence (12,13). In the assessment and management of EC, the use of ¹⁸F-FDG PET/CT can influence treatment choices for approximately one-third of individuals. Its ability to detect distant metastasis shows marked improvement in both sensitivity and specificity compared to conventional imaging techniques, making it a vital supportive tool for EC staging (14,15). However, the clinical values of metabolic parameters derived from PET/CT, such as the maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG), in the prognostic prediction and assessment of EC are still controversial (16). Although these parameters have been proposed as prognostic markers, the standardization and effectiveness of their clinical applications still need further verification.

Notably, no prior study has systematically integrated ¹⁸F-FDG PET/CT semi-quantitative parameters with multiple tumor biomarkers to develop an LNM predictive model for EC. This gap exists because standalone predictors are inherently incomplete, PET/CT parameters capture local tumor metabolic burden, while tumor biomarkers reflect systemic tumor activity that correlates with occult metastatic risk. Combining these two predictor classes addresses the shortcomings of standalone models, thereby maximizing predictive accuracy and clinical utility. In light of the aforementioned context, this research seeks to assess the predictive significance of metabolic parameters from ¹⁸F-FDG PET/CT and the variability in metabolism concerning LNM in EC. We present this article in accordance with the TRIPOD reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-545/rc).

Methods

Patient selection

This retrospective study aimed to develop and internally validate a predictive model for EC-related LNM using ¹⁸F-FDG PET/CT semi-quantitative parameters and tumor biomarkers. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Changzhou Cancer Hospital [2024(SR)NO.036] and individual consent for this retrospective analysis was waived. This study reviewed the cases of 200 patients who were pathologically diagnosed with primary EC. The inclusion criteria were as follows: (I) patients had not received any oncological treatment, such as surgery, radiotherapy, or chemotherapy, prior to ¹⁸F-FDG PET/CT examination; (II) EC and lymph node (LN) status was diagnosed via endoscopic or surgical pathology, and there was no history of other malignancies; (III) complete clinical data were available; (IV) the SUVmax of the primary EC lesion was ≥2.5. The exclusion criteria were: (I) patients who had undergone tumor-related interventional therapy before ¹⁸F-FDG PET/CT; (II) those with poor-quality ¹⁸F-FDG PET/CT images or negative ¹⁸F-FDG uptake in the primary lesion; (III) patients with concurrent other primary tumors or a previous history of malignant tumors; (IV) those without pathological confirmation of EC after PET/CT; (V) patients with low metabolic levels in PET images (SUVmax <2.5). Based on the “10 events per independent variable” principle for sample size estimation, a total of 200 patients were enrolled, consisting of 43 non-LNM cases and 157 LNM cases.

¹⁸F-FDG PET/CT scanning

The UNITED IMAGING uMI780 PET/CT scanner (112-ring PET, 128-slice CT) was used. ¹⁸F-FDG radiotracer was supplied by Shanghai Atomic KeXing Pharmaceutical Co., Ltd. (Shanghai, China), with radiochemical purity >95%. Patients fasted for 6–8 hours, with blood glucose levels between 2.9–9.1 mmol/L. They received an intravenous injection of ¹⁸F-FDG (173.382–347.615 MBq, 3.70–5.55 MBq/kg), patients with blood glucose 7.0–9.1 mmol/L rested 60–70 minutes, while others rested 45–60 post-injection, then, drinking 200 mL of water every 15 minutes (2–3 times). Before scanning, patients emptied their bladders and drank 500 mL of water to distend the stomach. Scanning included a 120 kV, 130 mA CT scan in the supine position during quiet breathing, followed by PET acquisition of 4–7 bed positions (skull vertex to upper-middle thigh), 2–3 minutes per bed. PET images were reconstructed with ordered subset expectation maximization (OSEM) (3 iterations, 24 subsets, 192×192 matrix, 3.0 mm slice thickness) plus point spread function (PSF)/time-of-flight (TOF).

Clinical biomarkers

Fasting venous blood (5 mL) was collected 1 week pre-PET/CT. Serum was separated via centrifugation (3,000 rpm, 10 min) and stored at −80 ℃. Biomarkers (CEA, CA19-9, CA125, CA724) were measured via chemiluminescent immunoassay (Roche Cobas e601, Penzberg, Germany); abnormal values exceeded reference ranges (CEA <5 ng/mL, CA19-9 <37 U/mL, CA125 <35 U/mL, CA724 <6.9 U/mL). LNM status was confirmed by surgical pathology: resected lymph nodes were formalin-fixed, paraffin-embedded, hematoxylin and eosin (HE)-stained, and reviewed by two pathologists (disagreements resolved by a senior pathologist). LNM was defined as malignant cells in lymph nodes.

Analysis of PET/CT images

All the data processing was completed on the workstation of UNITED IMAGING. Three physicians with PET/CT diagnostic experience at or above the attending physician level jointly analyzed the images. Using an absolute threshold of SUV =2.5 for the lesions, the software automatically delineated the SUVmax, MTV, and TLG of the primary EC lesions. To assess inter-observer variability, 30 cases (15% of the total 200-cohort) were randomly selected, and Cohen’s kappa coefficient was calculated for MTV and TLG (key volumetric parameters for LNM prediction) to quantify agreement. For the patients in this group, the radioactive uptake of the lesions (SUVmax ≥2.5) was used as the diagnostic criterion for malignancy.

Statistical analysis

All statistical analyses were performed employing SPSS 21.0 software (SPSS Inc., Chicago, IL, USA). Data were expressed as “mean ± standard deviation (SD)” and compared between LNM and non-LNM groups via two-sided independent samples t-test; categorical data were presented as frequencies (%) and analyzed with two-sided chi-square test (P<0.05 was statistically significant). To identify LNM-related risk factors, two-sided univariate logistic regression (screening threshold: P<0.10) was applied. Predictors with P<0.10 were included in two-sided multivariate logistic regression (backward elimination, P>0.05 for removal) to confirm independent predictors. Three sets of combined predictive models were built based on multivariate results and clinical relevance, with CA724 retained for its high receiver operating characteristic (ROC) curve (AUC) among biomarkers. Predictive accuracy was assessed via ROC curves [AUC, 95% confidence interval (CI]; optimal cutoffs for continuous predictors (e.g., MTV) were determined by Youden index. Two-sided Delong test compared AUC differences between models to evaluate parameter incremental value. All tests (t-test, chi-square, logistic regression, Delong test) were two-sided; P<0.05 was statistically significant.

Results

Univariate analysis of clinical data related to LNM

A total of 200 patients diagnosed with EC were enrolled in this study. Among them, 157 patients exhibited LNM, while 43 patients were not [non-lymph node metastasis (NLNM)]. Moderate-to-high agreement was also observed for SUVmax (kappa =0.82, 95% CI: 0.69–0.91) and SUVmean (kappa =0.80, 95% CI: 0.67–0.89). These results verified the reliability of PET/CT parameter measurements in the study. Rigorous statistical analyses were performed on the comprehensive clinical datasets of these patients. Results showed that there were significant differences in tumor diameter and T stage between the LNM group and the NLNM group. However, there were no significant differences in terms of age, gender, smoking and drinking history, the prevalence of diabetes and hypertension, family history of EC, histological type, and tumor location (Table 1).

Univariate analysis of PET/CT parameters related to LNM

We performed a comparative assessment of the PET/CT metrics of the primary lesions from both groups. The findings indicated that among patients with LNM, the levels of tumor markers—including CEA, CA19-9, CA125, and CA724—alongside semi-quantitative measures of ¹⁸F-FDG PET/CT such as SUVmax, SUVmean, MTV, and TLG, were all notably elevated compared to those in the NLNM group, with these differences being statistically significant (Table 2).

Analysis of risk factors for LNM in patients with EC

Among the 200 enrolled patients, 157 had LNM upon pathological diagnosis, while 43 did not. To elucidate the factors associated with LNM in EC, univariate logistic regression analysis was performed on 19 potential risk factors. The results showed that tumor diameter, T stage, CEA, CA19-9, CA125, CA724, SUVmax, SUVmean, MTV, and TLG were significantly associated with LNM in EC (all P values were less than 0.05). Subsequently, these significant variables were incorporated into a multivariate logistic regression model. The analysis indicated that CEA [odds ratio (OR) =1.326; 95% CI: 1.055–1.666; P=0.02], SUVmax (OR =1.351; 95% CI: 1.065–1.713; P=0.01), SUVmean (OR =22.391; 95% CI: 3.415–146.818; P=0.001), and MTV (OR =1.198; 95% CI: 1.062–1.351; P=0.003) were independent risk factors for LNM, while tumor diameter, T stage, CA19-9, CA125, CA724, and TLG were not significantly associated with LNM in the multivariate model (Table 3).

ROC curve analysis was systematically conducted to assess the predictive utility of multiple key parameters for LNM in EC. The results revealed that the T stage and tumor diameter exhibited relatively limited discriminatory power, with the AUC values both falling below 0.7. Among tumor markers, CA72-4 demonstrated superior predictive performance, with an AUC of 0.756, 95% CI: 0.691–0.814, surpassing CEA: AUC =0.726, 95% CI: 0.659–0.787, CA19-9: AUC =0.748, 95% CI: 0.681–0.806, and CA12-5: AUC =0.713, 95% CI: 0.645–0.775. Notably, several ¹⁸F-FDG PET/CT metabolic parameters and serological markers showed remarkable discriminatory ability. The SUVmax achieved an AUC of 0.810 (95% CI: 0.749–0.862), with a sensitivity of 85.35% and specificity of 65.12%. The SUVmean yielded an AUC of 0.827 (95% CI: 0.767–0.876), corresponding to a sensitivity of 92.99% and specificity of 69.77%. Among these parameters, MTV emerged as the most powerful predictor, with an AUC of 0.878 (95% CI: 0.824–0.920), sensitivity of 91.72%, and specificity of 65.12%. According to Youden’s index, an optimal cutoff value of 11.88 was determined for MTV. Additionally, TLG also demonstrated substantial predictive value, with an AUC of 0.811 (95% CI: 0.750–0.863), sensitivity of 75.80%, and specificity of 74.42%. These findings highlight the potential clinical significance of integrating metabolic and serological parameters for enhanced preoperative prediction of LNM in EC (Table 4).

To optimize the predictive model for LNM, several predictive models were constructed in the study to assess the risk of LNM in EC. Each model was based on different combinations of semi-quantitative parameters of ¹⁸F-FDG PET/CT and the tumor marker CA724 (the highest AUC of tumor markers in this study). In the first group of models, Model 1 combined CA724 and SUVmean, with AUC of 0.907 (95% CI: 0.858–0.943), a sensitivity of 91.08%, and a specificity of 74.42%. Model 2 incorporated MTV on this basis, and the AUC increased to 0.963 (95% CI: 0.926–0.984), with the sensitivity increasing to 96.18% and the specificity reaching 83.72%. Model 3 further added TLG, with an AUC of 0.964 (95% CI: 0.928–0.985), a sensitivity of 94.90%, and a specificity of 86.05% (Figure 1). The second group of models was constructed with CA724 and SUVmax as the basis. Model 1 had an AUC of 0.892 (95% CI: 0.840–0.931), a sensitivity of 73.89%, and a specificity of 90.70%. After adding MTV to Model 2, the AUC increased to 0.953 (95% CI: 0.913–0.978), with the sensitivity reaching 91.08% and the specificity remaining at 90.70%. Model 3 incorporated TLG, with an AUC of 0.955 (95% CI: 0.916–0.976), a sensitivity of 91.08%, and a specificity of 88.37% (Figure 2). In the third group of models, Model 1 was composed of CA724, SUVmean, and SUVmax, with an AUC of 0.929 (95% CI: 0.884–0.961), a sensitivity of 87.90%, and a specificity of 83.72%. After introducing MTV into Model 2, the AUC reached 0.964 (95% CI: 0.928–0.985), with a sensitivity of 96.82% and a specificity of 83.72%. Model 3 finally integrated TLG, with an AUC of 0.965 (95% CI: 0.930–0.986), a sensitivity of 94.90%, and a specificity of 86.05% (Figure 3).

Figure 1 Receiver operating characteristic curves of identified models for evaluated lymph node metastasis. Model 1: CA724 + SUVmean: AUC: 0.907, 95% CI: 0.858–0.943. Model 2: CA724 + SUVmean + MTV: AUC: 0.963, 95% CI: 0.926–0.984. Model 3: CA724 + SUVmean + MTV + TLG: AUC: 0.964, 95% CI: 0.928–0.985. AUC, area under the curve; CA724, carbohydrate antigen 724; CI, confidence interval; MTV, metabolic tumor volume; SUVmean, mean standardized uptake value; TLG, total lesion glycolysis.

Figure 2 Receiver operating characteristic curves of identified models for evaluated lymph node metastasis. Model 1: CA724 + SUVmax: AUC: 0.892, 95% CI: 0.840–0.931. Model 2: CA724 + SUVmax + MTV: AUC: 0.953, 95% CI: 0.913–0.978. Model 3: CA724 + SUVmax + MTV + TLG: AUC: 0.955, 95% CI: 0.916–0.979. AUC, area under the curve; CA724, carbohydrate antigen 724; CI, confidence interval; MTV, metabolic tumor volume; SUVmax, maximum standardized uptake value; TLG, total lesion glycolysis.

Figure 3 Receiver operating characteristic curves of identified models for evaluated lymph node metastasis. Model 1: CA724 + SUVmean + SUVmax: AUC: 0.929, 95% CI: 0.884–0.961. Model 2: CA724 + SUVmean + SUVmax + MTV: AUC: 0.964, 95% CI: 0.928–0.985. Model 3: CA724 + SUV mean + SUVmax + MTV + TLG: AUC: 0.965, 95% CI: 0.930–0.986. AUC, area under the curve; CA724, carbohydrate antigen 724; CI, confidence interval; MTV, metabolic tumor volume; SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; TLG, total lesion glycolysis.

In each combination of models, the gradual incorporation of semi-quantitative parameters from ¹⁸F-FDG PET/CT resulted in an increasing trend in the AUC values of the models, which greatly enhanced the accuracy of predicting LNM in EC. Notably, the models that integrated both MTV and TLG exhibited exceptional performance regarding predictive efficiency, demonstrating high sensitivity and specificity. This combination offers a robust resource for clinicians in the prediction of LNM in EC.

Discussion

LNM is a critical independent prognostic factor in malignant tumors, playing a decisive role in disease progression and outcomes for EC patients (7,8). Clinical evidence demonstrates that LNM significantly increases the risks of disease progression, recurrence, and mortality, profoundly impacting OS (8). Current diagnostic and therapeutic strategies for EC heavily rely on clinical staging, with radical surgery remaining the cornerstone of treatment for resectable cases. Consequently, accurate preoperative assessment of LNM is paramount, as it not only guides treatment selection but also serves as a vital prognostic indicator. Traditional imaging modalities primarily rely on lymph node size for LNM evaluation, yet this approach suffers from substantial limitations in clinical practice. While CT and magnetic resonance imaging (MRI) provide excellent anatomical visualization of lymph nodes, they lack sufficient specificity in distinguishing benign from malignant nodes (17,18). Conversely, the unique ability of ¹⁸F-FDG PET/CT for metabolic imaging has positioned it as an effective method for staging tumors, evaluating treatment responses, and predicting outcomes in EC (12-14). The superiority of PET/CT over conventional size-based assessment was demonstrated by Yoshimura et al. (19) in a prospective study of 1,073 lymph nodes. When analyzed by nodal stations, PET showed modest sensitivity (28.6%) but high specificity (96.7%). Remarkably, node-by-node analysis significantly improved diagnostic performance (sensitivity 94.7%, specificity 78.7%). This finding has been corroborated by multiple meta-analyses highlighting the prognostic value of PET/CT metabolic parameters (16,20,21). A pooled analysis of 1,294 EC patients revealed that MTV and TLG were strong predictors of OS [hazard ratio (HR) 2.26 and 2.23, respectively] (20). Similar predictive value was observed for event-free survival, with HRs of 2.03 for MTV and 2.57 for TLG (18). These findings have been replicated across various malignancies, including cervical and lung cancers, supporting the broad applicability of PET/CT parameters in oncology (22,23).

The aim of our study was to examine how effective the semi-quantitative parameters of ¹⁸F-FDG PET/CT are in predicting LNM in EC. Among the evaluated metrics (SUVmax, SUVmean, MTV, and TLG), MTV demonstrated superior diagnostic performance (AUC =0.878; 95% CI: 0.824–0.920; sensitivity 91.72%; specificity 65.12% at cutoff 11.88). As a composite measure integrating both metabolic intensity and tumor volume, MTV provides a more comprehensive assessment of tumor burden compared to isolated SUV measurements. TLG also showed robust predictive capability (AUC =0.811; 95% CI: 0.750–0.863), offering additional value by quantifying total tumor glucose uptake while mitigating the limitations of SUVmax affected by metabolic heterogeneity and partial volume effects. We constructed multiple predictive models to systematically evaluate parameter combinations incorporating PET/CT metrics and the tumor marker CA724. All models exhibited excellent discriminatory ability (AUC >0.89), with performance improving progressively through parameter addition. However, the performance of the models showed a stepwise improvement with the gradual inclusion of parameters. Taking the first group of models as an example, Model 1, which only included CA724 and SUVmean, already had a relatively high AUC (0.907; 95% CI: 0.858–0.943), suggesting that the combination of these two parameters has a certain predictive value for LNM. When the MTV was added to form Model 2, the AUC increased to 0.963 (95% CI: 0.926–0.984), and the sensitivity and specificity also increased accordingly. This indicates that MTV can capture the overall metabolic information of the tumor, complementing the deficiency of SUVmean, which only reflects the local metabolic intensity, and effectively improving the model’s ability to identify LNM. Model 3, which further incorporated the TLG, achieved an AUC of 0.964 (95% CI: 0.928–0.985), maintaining high sensitivity while further optimizing the specificity. This validated the distinct importance of TLG in measuring the total glucose metabolic activity within the tumor, while its combined impact with MTV considerably improved the model’s robustness. Yet TLG requires additional post-processing compared to SUVmax/MTV, and this extra complexity may lack justification in routine clinical settings that prioritize rapid decisions. Future studies should therefore incorporate cost-effectiveness analyses or clinician-reported usability evaluations to define TLG’s practical value versus simpler parameters like MTV.

By comparing the models with different parameter combinations, it was found that the overall performance of the second group of models based on SUVmax was slightly lower than that of the models based on SUVmean. This may be attributed to the fact that SUVmax is easily affected by factors such as metabolic heterogeneity within the tumor, partial volume effects, and physiological uptake of surrounding tissues, leading to deviations in reflecting the true metabolic state of the tumor. As an indicator of the average metabolic level, SUVmean reduces the above interferences to a certain extent and can more stably improve the predictive performance when combined with other parameters. In addition, the third group of models included both SUVmean and SUVmax. Although it performed well at the Model 1 stage (AUC =0.929; 95% CI: 0.884–0.961), the improvement after adding MTV and TLG subsequently was similar to that of the first two groups of models. This suggests that there may be information overlap between SUVmean and SUVmax in the multi-parameter model, while MTV and TLG are the key parameters to break through the bottleneck of predictive performance.

The results of this study are consistent with previous conclusions regarding the role of ¹⁸F-FDG PET/CT metabolic parameters in tumor metastasis prediction (12-14). MTV and TLG serve as significant markers that represent both tumor metabolic activity and burden, demonstrating crucial relevance in the assessment of staging and prognosis for EC as well as other types of malignant tumors (16,20,21). Research has shown a significant correlation between MTV and TLG with the OS and progression-free survival (PFS) of individuals diagnosed with EC (14,15). High MTV and TLG values indicate higher tumor invasiveness and poorer prognosis. This study further applied them to the prediction of LNM and found that they can effectively distinguish between metastatic and non-metastatic cases, providing a quantitative basis for preoperative risk stratification.

There are still some limitations in this study. Firstly, the model construction is based on retrospective data and single center, which may lead to selection bias and affect the external applicability of the model. Secondly, factors such as partial volume effects and individual metabolic differences, especially the LNM imbalance, among patients may interfere with the accuracy of PET/CT parameters. In addition, the study did not explore the interaction between other potential influencing factors (such as the tumor microenvironment and gene markers) and the existing parameters. Future research can verify the robustness of the model through multicenter, prospective cohort with a balanced LNM ratio, and combine new technologies such as radiomics and liquid biopsy to construct a more accurate prediction system for LNM in EC, providing more reliable support for clinical personalized treatment decisions.

Conclusions

In summary, ¹⁸F-FDG PET/CT semi-quantitative metrics are clinically valuable for predicting LNM in EC. Multi-parameter combined models outperform single indicators, with the best diagnostic performance seen in models integrating CA724 and metabolic parameters (MTV, SUVmean, TLG). These models offer a robust preoperative assessment tool, enabling precise risk stratification and personalized treatment for EC patients: low-risk patients (MTV <11.88 + normal CA724) may undergo minimally invasive surgery with limited lymphadenectomy (reducing morbidity), while high-risk patients could benefit from neoadjuvant therapy.

Acknowledgments

None.

Footnote

Reporting Checklist: The author has completed the TRIPOD reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-545/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-545/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-545/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-545/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Changzhou Cancer Hospital [2024(SR)NO.036] and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Fernández-Montes A, Alcaide J, Alsina M, et al. SEOM-GEMCAD-TTD Clinical Guideline for the diagnosis and treatment of esophageal cancer (2021). Clin Transl Oncol 2022;24:658-69. [Crossref] [PubMed]
3. Puhr HC, Prager GW, Ilhan-Mutlu A. How we treat esophageal squamous cell carcinoma. ESMO Open 2023;8:100789. [Crossref] [PubMed]
4. Obermannová R, Alsina M, Cervantes A, et al. Oesophageal cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2022;33:992-1004. [Crossref] [PubMed]
5. Shah MA, Altorki N, Patel P, et al. Improving outcomes in patients with oesophageal cancer. Nat Rev Clin Oncol 2023;20:390-407. [Crossref] [PubMed]
6. Deboever N, Jones CM, Yamashita K, et al. Advances in diagnosis and management of cancer of the esophagus. BMJ 2024;385:e074962. [Crossref] [PubMed]
7. Nobel T, Sihag S. Advances in Diagnostic, Staging, and Restaging Evaluation of Esophageal and Gastric Cancer. Surg Oncol Clin N Am 2024;33:467-85. [Crossref] [PubMed]
8. Zhang Y, Cao Y, Zhang J, et al. Lymph Node Ratio Improves Prediction of Overall Survival in Esophageal Cancer Patients Receiving Neoadjuvant Chemoradiotherapy: A National Cancer Database Analysis. Ann Surg 2023;277:e1239-46. [Crossref] [PubMed]
9. Wang J, Zhang X, Gan T, et al. Risk factors and a predictive nomogram for lymph node metastasis in superficial esophageal squamous cell carcinoma. World J Gastroenterol 2023;29:6138-47. [Crossref] [PubMed]
10. Li Y, Wang JX, Yibi RH. Prediction of lymph node metastasis in early esophageal cancer. World J Gastrointest Surg 2023;15:2294-304. [Crossref] [PubMed]
11. Yan HJ, Mao WJ, Yu RX, et al. Preoperative Clinical Characteristics Predict Recurrent Laryngeal Nerve Lymph Node Metastasis and Overall Survival in Esophageal Squamous Cell Carcinoma: A Retrospective Study With External Validation. Front Oncol 2022;12:859952. [Crossref] [PubMed]
12. Liu D, Lin D, Lin Z, et al. Development and validation of a nomogram for incorporating (18)F-FDG PET/CT spleen uptake for predicting prognosis in elderly esophageal cancer patients treated with radiotherapy. J Thorac Dis 2024;16:7853-65. [Crossref] [PubMed]
13. Brenner B, Kundel Y, Cohen Z, et al. Early prediction of residual disease after neoadjuvant chemoradiotherapy and cetuximab for locally advanced esophageal cancer using 18F-FDG PET-CT imaging: a prospective cohort study. J Gastrointest Oncol 2022;13:2721-35. [Crossref] [PubMed]
14. Martínez A, Infante JR, Quirós J, et al. Baseline (18)F-FDG PET/CT quantitative parameters as prognostic factors in esophageal squamous cell cancer. Rev Esp Med Nucl Imagen Mol (Engl Ed) 2022;41:164-70. [Crossref] [PubMed]
15. Karahan Şen NP, Aksu A, Çapa Kaya G. Volumetric Evaluation of Staging (18)F-FDG PET/CT Images in Patients with Esophageal Cancer. Mol Imaging Radionucl Ther 2022;31:216-22. [Crossref] [PubMed]
16. Huang M, Wang W, Wang R, et al. The prognostic value of pretreatment [18F]FDG PET/CT parameters in esophageal cancer: a meta-analysis. Eur Radiol 2025;35:3396-408. [Crossref] [PubMed]
17. Van Coillie S, Van San E, Goetschalckx I, et al. Targeting ferroptosis protects against experimental (multi)organ dysfunction and death. Nat Commun 2022;13:1046. [Crossref] [PubMed]
18. Santos FS, Verma N, Marchiori E, et al. MRI-based differentiation between lymphoma and sarcoidosis in mediastinal lymph nodes. J Bras Pneumol 2021;47:e20200055. [Crossref] [PubMed]
19. Yoshimura S, Takahashi M, Aikou S, et al. One-by-One Comparison of Lymph Nodes Between 18F-FDG Uptake and Pathological Diagnosis in Esophageal Cancer. Clin Nucl Med 2020;45:741-6. [Crossref] [PubMed]
20. Krengli M, Ferrara E, Guaschino R, et al. 18F-FDG PET/CT as predictive and prognostic factor in esophageal cancer treated with combined modality treatment. Ann Nucl Med 2022;36:450-9. [Crossref] [PubMed]
21. Han S, Kim YJ, Woo S, et al. Prognostic Value of Volumetric Parameters of Pretreatment 18F-FDG PET/CT in Esophageal Cancer: A Systematic Review and Meta-analysis. Clin Nucl Med 2018;43:887-94. [Crossref] [PubMed]
22. Jiang CZ, Zheng K, Zhang YY, et al. (18)F-FDG PET/CT semi-quantitative parameters combined with SCC-Ag in predicting lymph node metastasis in stage I-II cervical cancer. Front Oncol 2024;14:1278464. [Crossref] [PubMed]
23. Yu X, Wang J, Huang L, et al. Predictive value of 18F-FDG PET/CT metabolic parameters for lymph node metastasis of non-small cell lung cancer. Biomark Med 2025;19:35-41. [Crossref] [PubMed]