Esophageal cancer is the seventh most prevalent malignancy worldwide, with the sixth leading cause of cancer-related mortality (1). Accurate staging is essential for treatment selection and prognosis prediction for patients with esophageal cancer. Imaging plays a critical role in tumor staging. Currently, the commonly used imaging methods for staging include computed tomography (CT), endoscopic ultrasonography (EUS), and positron emission tomography (PET)/CT. However, accurate preoperative staging remains a challenge (2).

In the description of the primary tumor (T staging) of esophageal cancer, it is difficult to distinguish the layers of the esophageal wall on CT due to the poor contrast of soft tissue. The application of CT is limited to distinguishing T3 and T4 tumors in the T staging of esophageal cancer (3). In the assessment of regional lymph nodes (N staging), the determination only depends on the size of the lymph nodes by CT, with low accuracy. EUS or EUS combined with fine-needle aspiration biopsy (FNAB) reveals high accuracy in T and N staging, but its application is limited by operator dependency, risk of hemorrhage (0.13% morbidity rate), and perforation (0.03%-0.07% morbidity rate), inability to pass through the stenosis (20%-30% morbidity rate), and the scope of the examination (4, 5). In clinical practice, ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET/CT has limitations in T staging of esophageal cancer due to its resolution but shows high specificity in N staging, with poor sensitivity and some false-positive results. With superior soft-tissue contrast, magnetic resonance imaging (MRI) may distinguish the layers of the esophageal wall and adjacent lymph nodes and has the features of multi-parametric and functional imaging. MRI revealed better accuracy in T staging and higher sensitivity in N staging than CT, but there were still some primary tumors and lymph nodes that were difficult to detect and accurately describe by MRI. Therefore, a more accurate and reliable noninvasive preoperative staging method is desired.

¹⁸F-FDG PET/MRI provides both metabolic and anatomical information about the tumor and combines the advantages of MRI’s superior soft-tissue resolution and multi-parametric imaging, which can detect more malignant lesions than PET/CT, leading to changes in TNM staging (6, 7). Previous studies revealed ¹⁸F-FDG PET/MRI to be superior to PET/CT in T staging and at least comparable to PET/CT in N and M staging of a variety of tumors (8, 9). In addition, imaging parameters such as the standardized uptake value (SUV) and apparent diffusion coefficient (ADC) of tumors may correlate with staging and prognosis (10). Preliminary studies showed that PET/MRI may overcome the inherent limitations of PET/CT and CT in T staging of esophageal cancer, and has advantages over other imaging methods in N staging (11). However, there are few studies of PET/MRI in the assessment of esophageal cancer, and the value needs to be further explored. Therefore, the purpose of our study was to compare the diagnostic efficiency of PET/MRI, PET/CT, MRI, and contrast-enhanced CT (CECT) in the preoperative assessment of primary tumors and regional lymph nodes of esophageal cancer and to explore the role of imaging and clinical parameters in predicting pathological stages.

The depth of tumor invasion is the key to treatment and surgical options. T1-2 tumors can be treated directly by surgery, while T3-4 tumors often need preoperative neoadjuvant therapy. Previous studies have demonstrated the high accuracy (higher than 80%) of EUS in T staging (14). However, EUS has some limitations, such as operator dependence, inability to pass through the stenoses, and the risk of hemorrhage and perforation, which limit its application. CT, with poor soft-tissue contrast, has limited ability to accurately distinguish T1 and T2 diseases from T3 diseases and is mainly used to distinguish T3 and T4 tumors. Due to the spatial resolution of PET and the low contrast of low-dose CT, PET/CT scans are unable to provide accurate information on esophageal wall stratification and have a limited role in T staging.

MRI, with superior soft-tissue contrast, can display stratification of the esophageal wall and observe the surrounding tissue structure. In vitro studies showed that the three layers of the esophageal wall could be distinguished on high-resolution T2WI (15). However, there are still some tumors that cannot be accurately staged by MRI, especially superficial diseases. Previous studies showed that PET/MRI could identify the esophageal wall layer, with acceptable accuracy slightly worse than EUS (11). In this study, we found that PET/MRI could distinguish the stratification of the esophageal wall with better accuracy than CECT in preoperative T staging of ESCC. The diagnostic accuracy of PET/MRI for T1, T2, and T3 diseases (86.7%, 77.8%, and 83.3%, respectively) was superior to that of CECT (40.0%, 44.4%, and 75.0%, respectively). Moreover, PET/MRI may improve the detection of superficial lesions by MRI. Thus, the description of tumor invasion depth may be significantly improved by PET/MRI. Given the poor performance of PET/CT in T staging of esophageal cancer, we believe that PET/MRI may be an effective supplement to PET/CT, MRI, and CECT in T staging of ESCC.

Lymph node staging is a critical prognostic factor for esophageal cancer. Accurate N staging determines the treatment and facilitates complete resection of all positive lymph nodes to improve long-term survival. Conversely, extended lymphadenectomy may increase postoperative complications. At present, the accuracy of common methods is not satisfactory in lymph node assessment. CT only takes the size of lymph nodes as a judgment standard, with low accuracy, as some small lymph nodes are metastatic, whereas inflammatory, reactive, and granulomatous lymph nodes may be enlarged. Previous studies have used different criteria for diagnosing lymph node metastasis, with most studies using a short diameter larger than 10 mm as a criterion (16). However, our study revealed that the short diameter of metastatic lymph nodes in all 23 stations was less than 10 mm, with a mean value of 5.8 ± 2.0 mm. Therefore, it is not accurate to take only the short diameter of lymph nodes as the standard for the diagnosis of metastatic lymph nodes, which has a high false-negative rate.

EUS combined with FNAB revealed high accuracy in the diagnosis of lymph node metastasis, but non-paraesophageal lymph nodes were easily missed due to the limitation of the detection range, and the passage was limited when the lumen was narrow. Furthermore, the lymph nodes near the tumor cannot be punctured. The advantage of PET/CT in N staging of esophageal cancer is of high specificity. However, its sensitivity is low, at only approximately 30%-60%, due to the spatial resolution. Therefore, the detection of small lymph node metastases and the differentiation of paraesophageal lymph nodes from primary tumors remain challenges. Lymph nodes can be detected sensitively on DWI images. Meanwhile, the intensity and internal structure of the lymph nodes can be observed on high-resolution T2WI (17, 18). Therefore, metastatic lymph nodes smaller than 10.0 mm can also be identified by MRI. However, there is an overlap of ADC values or intensity between metastatic and non-metastatic lymph nodes (19). In this condition, the differentiation of benign and malignant lymph nodes is difficult by MRI. PET/MRI may provide additional information for lymph node assessment. Our study found that PET/MRI had better diagnostic efficiency than PET/CT, MRI, and CECT in lymph node assessment of ESCC (AUC: 0.883, 0.745, 0.697, and 0.580, respectively; sensitivity: 78.3%, 52.2%, 47.8%, and 21.7%, respectively). Combining metabolism and morphology, PET/MRI diagnosed 13, 7, and 6 more stations of lymph node metastases than CECT, MRI or PET/CT, respectively, as well as excluded metastases in 8, 13, and 3 stations than CECT, MRI, or PET/CT, respectively. Subgroup analysis in our study also revealed that PET/MRI had more obvious advantages in lower-thoracic, poorly-differentiated, and T3 ESCC. Therefore, we believe that PET/MRI may improve the sensitivity, accuracy, and diagnostic confidence of lymph node assessment, play a complementary or further confirming role and may reduce the risk of biopsy or avoid other additional imaging examinations.

Previous studies have demonstrated that the SUV, MTV, TLG, and ADC may be valuable prognostic factors for esophageal cancer (10, 20). However, whether these parameters can be used for the prediction of the pathological stage is still controversial. For the treatment of esophageal cancer, the muscularis propria is an important dividing line. Our study revealed that SUV_max, TLG, and tumor wall thickness may be useful for the differentiation of T1-2 and T3 tumors. However, we found that MTV, ADC_min, ADC_mean, △HU, and clinical parameters play a limited role in differentiating T1-2 and T3 tumors, and all the imaging parameters of the primary tumor and clinical parameters play a limited role in differentiating N0 and N+ patients. Therefore, whether these parameters can be used to predict T and N staging remains to be further explored.

Our findings are consistent with those of several previous studies (21), but some studies have found significant differences in ADC and MTV between high and low T-stage tumors (22), and some studies involving PET/CT have reported a significant correlation between tumor SUV_max and N stage (23). In the present study, we found that the SUV_max and D_max of lymph nodes may be useful in differentiating metastatic and non-metastatic lymph nodes. The differences between the results of different studies may be attributed to differences in clinicopathological characteristics or treatment of patients or differences in sample size. In general, the use of imaging parameters to accurately predict the staging of esophageal cancer still requires further study. Furthermore, the application of new techniques may help to improve the efficiency of PET/MRI in predicting T and N staging of esophageal cancer (24–28).

This study has several limitations. First, the limited number of cases included in the analysis may affect the power of the statistical analysis. Second, only a few patients underwent EUS examinations before surgery in our study. Therefore, the diagnostic efficiency of PET/MRI, PET/CT, MRI, and CECT cannot be compared with EUS at the same time. This may be explained by the guidelines published by the National Institute for Health and Care Excellence (NICE) in 2018, which recommend that EUS should be performed only if there is a potential change in treatment after PET/CT (29). Third, since most of our patients were elderly and could not tolerate multiple breath-hold acquisitions, we did not include breath-hold sequences in MRI acquisition, which may improve the observation of lesions. Finally, our patients were surgical patients without distant metastasis; therefore, the value of PET/MRI in M staging was not analyzed.

In conclusion, ¹⁸F-FDG PET/MRI has advantages over ¹⁸F-FDG PET/CT, MRI, and CECT in the preoperative assessment of the primary tumor and regional lymph node of resectable ESCC, especially in the description of the depth of tumor invasion and the sensitivity of lymph node assessment. Furthermore, PET/MRI-derived imaging parameters also contribute to the prediction of T staging and lymph node status. ¹⁸F-FDG PET/MRI may be a potential supplement or alternative imaging method for preoperative staging of ESCC.

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by Ethics Committee of the Peking University Cancer Hospital & Institute. The patients/participants provided their written informed consent to participate in this study.

FW, ZY, and NL contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by FW, RG, and NL. Image acquisition was performed by YZ and BY. The statistical methods were reviewed by XM, HK, and YY. The first draft of the manuscript was written by FW and was revised by NL and ZY. All authors contributed to the article and approved the submitted version.

This study was funded by the National Natural Science Foundation (No. 81871387; No. 81871386), Beijing Natural Science Foundation (No. 7202027).

Authors HK and YY were employed by Beijing United Imaging Research Institute of Intelligent Imaging, UIH Group.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.