Objective To develop and validate a novel model based on preoperative MRI features and multiregional apparent diffusion coefficients (ADCs) to improve the prediction of pN stage in resectable rectal adenocarcinoma (RA). Methods Two hundred fifty-four consecutive patients (median age [interquartile range], 67 [56–74] years; 156 males) with resectable RA were retrospectively collected at two medical centers from January 2017 to December 2023 and were divided into the training (n = 139), internal validation (n = 60), and external validation (n = 55) sets. All patients underwent preoperative MRI scans. Univariate and multivariate logistic regression analyses were conducted on the MRI features and multiregional (RA, peritumoral tissue, and tumor-adjacent rectal wall) ADCs to construct a nomogram model for preoperative predicting pN stage in the training set. Receiver operating characteristic (ROC) analysis was used to evaluate the predictive performance of the nomogram model vs the conventional MRI-assessed N (mriN) stage. The ROC curves were compared using the DeLong test. Results The predictors incorporated in the nomogram model comprised gross tumor volume, categories of short diameter of maximum node, extramural vascular invasion, mesorectal fascia involvement, and ADCs of RA and peritumoral tissue. This model yielded a better prediction of the pN stage compared to the mriN stage in training (AUC, 0.848 vs 0.672; p < 0.001), internal validation (AUC, 0.843 vs 0.699; p = 0.008), and external validation (AUC, 0.857 vs 0.723; p = 0.01) sets. Conclusion This novel model based on the preoperative MRI features and multiregional ADCs can improve the prediction of the pN stage in RA. Key Points <Emphasis Type="BoldItalic">Question</Emphasis> Accurate preoperative assessment of the pN stage is important for determining an appropriate therapeutic strategy in rectal cancer, but the conventional mriN stage has low sensitivity. <Emphasis Type="BoldItalic">Findings</Emphasis> Utilization of certain MRI features and multiregional ADCs improves preoperative assessment of the pN stage in RA when compared with conventional MRI assessment. <Emphasis Type="BoldItalic">Clinical relevance</Emphasis> The novel model, based on preoperative MRI features and multiregional ADC values, can improve the prediction of the pN stage compared to the mriN stage in RA. The combination of this model with the mriN stage helps personalize treatment plans to improve patient prognosis. Graphical Abstract

A novel pN stage prediction model for resectable rectal adenocarcinoma based on preoperative MRI features and multiregional apparent diffusion coefficients

期刊论文
发表日期：

2025年3月25日

查看全文

Hui Luo,
Yue-qin Gou,
Yue-su Wang,
Hui-lin Qin,
Hai-ying Zhou,
Xiao-ming Zhang,
Tian-wu Chen

摘要

Objective

To develop and validate a novel model based on preoperative MRI features and multiregional apparent diffusion coefficients (ADCs) to improve the prediction of pN stage in resectable rectal adenocarcinoma (RA).

Methods

Two hundred fifty-four consecutive patients (median age [interquartile range], 67 [56–74] years; 156 males) with resectable RA were retrospectively collected at two medical centers from January 2017 to December 2023 and were divided into the training (n = 139), internal validation (n = 60), and external validation (n = 55) sets. All patients underwent preoperative MRI scans. Univariate and multivariate logistic regression analyses were conducted on the MRI features and multiregional (RA, peritumoral tissue, and tumor-adjacent rectal wall) ADCs to construct a nomogram model for preoperative predicting pN stage in the training set. Receiver operating characteristic (ROC) analysis was used to evaluate the predictive performance of the nomogram model vs the conventional MRI-assessed N (mriN) stage. The ROC curves were compared using the DeLong test.

Results

The predictors incorporated in the nomogram model comprised gross tumor volume, categories of short diameter of maximum node, extramural vascular invasion, mesorectal fascia involvement, and ADCs of RA and peritumoral tissue. This model yielded a better prediction of the pN stage compared to the mriN stage in training (AUC, 0.848 vs 0.672; p < 0.001), internal validation (AUC, 0.843 vs 0.699; p = 0.008), and external validation (AUC, 0.857 vs 0.723; p = 0.01) sets.

Conclusion

This novel model based on the preoperative MRI features and multiregional ADCs can improve the prediction of the pN stage in RA.

Key Points

Question Accurate preoperative assessment of the pN stage is important for determining an appropriate therapeutic strategy in rectal cancer, but the conventional mriN stage has low sensitivity.

Findings Utilization of certain MRI features and multiregional ADCs improves preoperative assessment of the pN stage in RA when compared with conventional MRI assessment.

Clinical relevance The novel model, based on preoperative MRI features and multiregional ADC values, can improve the prediction of the pN stage compared to the mriN stage in RA. The combination of this model with the mriN stage helps personalize treatment plans to improve patient prognosis.

Graphical Abstract