Accurate Colon Cancer Nodal Staging is essential for determining optimal treatment strategies, particularly the crucial decision regarding neoadjuvant therapy. However, traditional imaging methods, like standard Computed Tomography (CT), often struggle to precisely identify lymph node metastasis (LNM) before surgery. Researchers have consequently explored advanced techniques and combined models to improve diagnostic accuracy. This new study highlights the potent synergy between Dual-Energy CT (DECT) parameters and simple clinical markers to achieve superior preoperative prediction.
Limitations of Standard Preoperative Imaging
Standard contrast-enhanced CT is a mandatory test in current Indian guidelines for colon cancer staging, and furthermore it helps to identify distant metastases. Nevertheless, the accurate differentiation between benign and malignant lymph nodes remains a persistent challenge for standard CT and MRI. In many studies, CT’s accuracy for determining N-stage has proven limited, often falling below 70%. Therefore, treatment planning for locally advanced disease often relies on a high index of suspicion or post-surgical pathological confirmation. Since preoperative LNM status is crucial for deciding on neoadjuvant versus upfront surgical approaches, a more reliable predictive tool is necessary. Clinical nomograms and machine learning models are consequently emerging to fill this critical diagnostic gap.
Optimizing Colon Cancer Nodal Staging with a Combined Model
The retrospective study enrolled 205 patients with pathologically confirmed colon cancer. Researchers randomly divided these patients into training and test sets. The team evaluated DECT quantitative parameters, including extracellular volume fraction (ECV) and dual-energy index in venous phase (DEIV), alongside clinical characteristics such as Carbohydrate Antigen 125 (CA125) and Carbohydrate Antigen 242 (CA242). Univariable and multivariable logistic regression analysis were used to build three separate predictive models: the DECT model, the Clinical model, and the Combined model. This method directly compared the utility of imaging, clinical data, and their synergy.
The results were highly encouraging. CA125, CA242, DEIV, and ECV were all identified as independent factors for predicting pathological nodal positivity (pN+). Furthermore, the Combined model, which integrates all these parameters, demonstrated significantly superior diagnostic performance. It achieved an excellent Area Under the Curve (AUC) of 0.906 in the training set. Specifically, the Combined model significantly outperformed the Clinical model (AUC 0.825) and the DECT model alone (AUC 0.865). Moreover, in the independent test set, the Combined model proved superior to the existing Node Reporting and Data System (Node-RADS), with an AUC of 0.906 versus 0.706. This performance confirms the clinical value of combining radiomic and serological data.
Clinical Impact and Applications in Treatment Planning
The excellent performance of the combined DECT and clinical feature model has crucial implications for preoperative risk stratification. Accurate preoperative identification of pN+ status allows clinicians to tailor treatment more effectively. Thus, a patient identified as high-risk by this model could be prioritized for neoadjuvant chemotherapy or total neoadjuvant therapy, especially in cases of locally advanced colon cancer. Since current Indian guidelines emphasize the role of CEA and CECT for staging, the inclusion of CA125, CA242, and advanced DECT parameters offers a powerful refinement. Therefore, this model provides a highly practical, non-invasive tool to better inform multidisciplinary team discussions and improve patient outcomes.
Frequently Asked Questions
Q1: What specific DECT parameters were found to be independent predictors of lymph node metastasis?
The Dual-Energy CT (DECT) parameters found to be independent predictors of pathological nodal positivity (pN+) were the extracellular volume fraction (ECV) and the dual-energy index in the venous phase (DEIV).
Q2: How much better was the Combined model compared to the standard Node-RADS system?
In the independent test set, the Combined model achieved an Area Under the Curve (AUC) of 0.906, which was significantly superior to the AUC of 0.706 achieved by the Node-RADS system.
Q3: Besides DECT parameters, which clinical features were used in the predictive model?
The clinical features identified as independent predictors for pathological nodal positivity (pN+) were Carbohydrate Antigen 125 (CA125) and Carbohydrate Antigen 242 (CA242).
References
- He C et al. Preoperative colon cancer nodal staging using dual-energy CT and clinically derived features. Eur Radiol. 2026 Jan 16. doi: 10.1007/s00330-025-12294-6. PMID: 41543562.
- Liu Y et al. Predicting the risk of lymph node metastasis in colon cancer: development and validation of an online dynamic nomogram based on multiple preoperative data. Front Oncol. 2023 May 8. doi: 10.3389/fonc.2023.1189437.
- Agarwal S et al. Indian Council of Medical Research consensus document for the management of colorectal cancer. Indian J Med Paediatr Oncol. 2020;41(4):460-474.