Liu Xiangchun, Wan Lijuan, Zhao Rui, Chen Shuang, Peng Wenjing, Yang Fan, Zhang Hongmei
Department of Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, #17 Panjiayuan Nanli, Chaoyang District, Beijing 100021, China.
Department of Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, #17 Panjiayuan Nanli, Chaoyang District, Beijing 100021, China.
Acad Radiol. 2025 Aug;32(8):4583-4594. doi: 10.1016/j.acra.2025.03.056. Epub 2025 May 2.
To develop and validate a predictive model for the pre-treatment evaluation of perineural invasion (PNI) status and to examine its prognostic stratification effectiveness in patients with stage I-III rectal cancer (RC) based on mismatch repair (MMR) status, clinical data, and magnetic resonance imaging (MRI) evaluated features.
This retrospective study included 815 patients with stage I-III RC who underwent MRI scans from January 2016 to November 2023 and were randomly assigned to the training and validation cohorts. MMR status, clinical data, and MRI-evaluated features associated with PNI status were identified as independent predictors for developing a predictive model by univariable and multivariable logistic regression analyses in the training cohort. The receiver operating characteristic curves and the area under the curves (AUCs) were utilized to evaluate the diagnostic performance of the prediction model in both the training and validation cohorts. The Kaplan-Meier survival curves and Cox proportional hazards regression analysis were utilized to evaluate the prognostic stratification value of the model in both the training and validation cohorts.
The predictive model developed with independent predictors, including deficient MMR (odds ratio [OR]=0.434, P=0.021), male gender (OR=1.578, P=0.013), MRI-evaluated tumor morphology (partly annular, OR=3.257, P<0.001; annular, OR=5.184, P<0.001), tumor stage (T3, OR=1.953, P=0.004; T4, OR=2.627, P=0.013), extramural vascular invasion (OR=1.736, P=0.041), tumor deposit (OR=3.902, P<0.001) and mesorectal fascia involvement (OR=2.679, P=0.023), achieved AUCs of 0.748 (95% confidence interval [CI]: 0.711-0.785, P<0.001) and 0.719 (95% CI: 0.640-0.798, P<0.001) in the training and validation cohorts, respectively. The Kaplan-Meier survival curves show effectively prognostic stratification for disease-free survival (DFS), distant metastasis-free survival (DMFS), and recurrence-free survival (RFS) between predicted PNI-positive and PNI-negative patients (both P<0.05). Cox regression analysis indicated that predicted PNI-positive status was a significant risk factor associated with inferior DFS and DMFS in both training and validation cohorts (both P<0.05). The predicted PNI-positive status was a significant risk factor associated with inferior RFS in the training cohort (P=0.002); however, no significant association was observed in the validation cohort (P=0.104).
The developed prediction model for evaluating the PNI status of RC prior to treatment showing acceptable performance and helping with prognostic stratification, which may assist in personalized treatment decisions.
开发并验证一种用于术前评估神经周围侵犯（PNI）状态的预测模型，并基于错配修复（MMR）状态、临床数据和磁共振成像（MRI）评估特征，检验其在I-III期直肠癌（RC）患者中的预后分层效果。
这项回顾性研究纳入了815例I-III期RC患者，这些患者在2016年1月至2023年11月期间接受了MRI扫描，并被随机分配到训练队列和验证队列。通过单变量和多变量逻辑回归分析，在训练队列中将与PNI状态相关的MMR状态临床数据和MRI评估特征确定为开发预测模型的独立预测因子。利用受试者工作特征曲线和曲线下面积（AUC）评估预测模型在训练队列和验证队列中的诊断性能。利用Kaplan-Meier生存曲线和Cox比例风险回归分析评估模型在训练队列和验证队列中的预后分层价值。
由独立预测因子（包括MMR缺陷（比值比[OR]=0.434，P=0.021）、男性（OR=1.578，P=0.013）、MRI评估的肿瘤形态（部分环形，OR=3.257，P<0.001；环形，OR=5.184，P<0.001）、肿瘤分期（T3，OR=1.953，P=0.004；T4，OR=2.627，P=0.013）、壁外血管侵犯（OR=1.736，P=0.041）、肿瘤结节（OR=3.902，P<0.001）和直肠系膜筋膜受累（OR=2.679，P=0.023））构建的预测模型，在训练队列和验证队列中的AUC分别为0.748（95%置信区间[CI]：0.711-0.785，P<0.001）和0.719（95%CI：0.640-0.798，P<0.
