MRI-based Nomogram to Preoperatively Predict Early Recurrence of Adenocarcinoma of the Esophagogastric Junction: a two-center study

峰 秦, Ruirui Song, Peng Shen, Jing Zhang, Junbo Wu, Rong Jin, Jun Zhao, Shuai Quan, Jiayi Zhang, Guoxin Fan, Chenyue Xu, Jin Wen, Yingchun Liu, Wenqing Hu, Yongai Li, Xiaotang YangResearch Square 2025 · DOI

This study presents a significant step forward in personalized oncology for a challenging type of cancer. By leveraging preoperative MRI data, researchers have developed a tool that predicts cancer recurrence more accurately than the current "gold standard" clinical staging system.

TL;DR

Researchers developed an MRI-based nomogram to predict the early recurrence (within two years) of Adenocarcinoma of the Esophagogastric Junction (AEG) before surgery. By combining tumor length, diffusion-weighted imaging data, and lymph node counts, the model significantly outperforms traditional TNM staging, allowing for better-tailored treatment plans.

What Problem Does This Paper Address?

Adenocarcinoma of the Esophagogastric Junction (AEG)—cancer occurring where the esophagus meets the stomach—is on the rise globally, particularly in East Asia. While surgery is the primary treatment, "early recurrence" (returning within two years) is a major hurdle, drastically reducing survival rates.

The current clinical "gold standard" for prognosis is the TNM (Tumor, Node, Metastasis) staging system. However, TNM staging is often most accurate after surgery when a pathologist can examine the tissue. Preoperatively, it struggles to detect micrometastases in lymph nodes. There is a critical need for a non-invasive, preoperative method to identify high-risk patients who might benefit from more aggressive initial treatments, such as neoadjuvant chemoradiotherapy.

Methods

The researchers conducted a retrospective two-center study involving 138 patients. They analyzed preoperative contrast-enhanced MRI scans, focusing on both morphological and functional imaging features.

The study's "secret sauce" involves three specific preoperative variables:

  1. Tumor Length: Measured via MRI.
  2. Relative ADC (rADC): Derived from Diffusion-Weighted Imaging (DWI). The Apparent Diffusion Coefficient (ADC) measures water molecule diffusion; lower values typically indicate denser, more aggressive tumors. To account for differences in MRI hardware between the two hospitals, the authors used a "relative" value—the ratio of the tumor’s ADC to the spleen’s ADC.
  3. Regional Lymph Node Count: The total number of regional lymph nodes visible on high b-value DWI scans.

These variables were integrated into a nomogram—a graphical calculating device that allows clinicians to input patient data and receive a statistical probability of 2-year survival/recurrence.

Datasets

The study utilized data from two medical centers to ensure the model wasn't overfitted to a single population:

  • Training Set: 81 patients from Center 1 (Changzhi People's Hospital).
  • External Validation Set: 57 patients from Center 2 (Shanxi Cancer Hospital).
  • Timeframe: Data was collected between March 2014 and October 2023.

Key Insights and Ideas

  • The Power of rADC: Using the spleen as a reference point for ADC values (rADC) is a clever methodological choice. It normalizes the data, making the model robust even when using different MRI scanners (1.5T vs. 3.0T) across different institutions.
  • Imaging vs. Pathology: The study found that these three MRI markers were more predictive of early recurrence than the actual post-operative TNM stage. This suggests that the biological aggressiveness captured by MRI (like water diffusion) might be more important for early recurrence than the physical extent of the tumor alone.
  • Lymph Node Quantity Matters: Rather than just looking for "positive" or "negative" nodes, the total count of visible regional nodes on DWI served as a strong independent predictor of recurrence.

Key Results

The nomogram demonstrated exceptional predictive power:

  • Accuracy: The model achieved an Area Under the Curve (AUC) of 0.927 in the training set and 0.896 in the validation set. For context, an AUC of 1.0 is a perfect prediction.
  • Superiority: The nomogram significantly outperformed the AJCC 8th edition TNM staging system, which had lower AUCs (0.815 and 0.717, respectively).
  • Clinical Benefit: Decision Curve Analysis (DCA) confirmed that using this nomogram to guide treatment provides a higher "net benefit" to patients than relying on traditional staging or treating everyone the same way.

Limitations and Future Work

While promising, the authors acknowledge several limitations:

  • Sample Size: With only 138 patients, larger multi-center studies are needed to further refine the model.
  • Dimensionality: ADC values were measured from a single axial slice. Future work could involve "whole-tumor" volume analysis for even greater precision.
  • Missing Markers: The study did not incorporate molecular or immunohistochemical markers (like PD-L1 or Her2 status), which are becoming increasingly vital in personalized oncology. Integrating these with MRI data could create an even more powerful "radiomics" profile.

Summary updated

Abstract

<title>Abstract</title> <bold>Purpose:</bold> This study was to develop a predictive model based on preoperative MRI for early recurrence in patients with AEG. <bold>Methods:</bold> A retrospective analysis was performed on a cohort of consecutive patients from two medical centers who underwent preoperative contrast-enhanced MRI between March 2014 and October 2023, with AEG confirmed via surgical intervention. Cox regression analysis was used to explore risk factors for early postoperative recurrence, utilizing clinical data, laboratory results, and MRI characteristics. A nomogram was subsequently constructed and its performance was evaluated against the postoperative TNM staging system. <bold>Results:</bold> The study included 138 patients, divided into a training set (Center 1, n=81) and a validation set (Center 2, n=57). In the training set, three preoperative variables—tumor length, relative ADC value, and total number of regional lymph nodes—were found to be independently associated with early recurrence. The nomogram model constructed using these variables, demonstrated AUC values of 0.927 (95% CI, 0.854–0.928) and 0.896 (95% CI, 0.796–0.899) in the training and validation sets, respectively, with C-indices of 0.868 and 0.819. The model’s accuracy was significantly superior to that of the TNM staging system (P&lt;0.05). <bold>Conclusion: </bold>The MRI-based nomogram demonstrates superior performance compared to the traditional TNM staging system in predicting early postoperative recurrence in AEG.
BibTeX
@inproceedings{2025mri,
  author = {峰 秦 and Ruirui Song and Peng Shen and Jing Zhang and Junbo Wu and Rong Jin and Jun Zhao and Shuai Quan and Jiayi Zhang and Guoxin Fan and Chenyue Xu and Jin Wen and Yingchun Liu and Wenqing Hu and Yongai Li and Xiaotang Yang},
  title = {{MRI-based Nomogram to Preoperatively Predict Early Recurrence of Adenocarcinoma of the Esophagogastric Junction: a two-center study}},
  booktitle = {Research Square},
  year = {2025},
  doi = {10.21203/rs.3.rs-6389616/v1},
}
Figure