Ma Weili, Hou Chuanling, Yang Minxia, Wei Yuguo, Mao Jiwei, Guan Le, Zhao Zhenhua
Department of Radiology, Shaoxing People's Hospital, Key Laboratory of Functional Molecular Imaging of Tumor and Interventional Diagnosis and Treatment of Shaoxing City, Shaoxing, China.
Department of Pathology, Shaoxing People's Hospital, Shaoxing, China.
Front Oncol. 2025 Apr 28;15:1509207. doi: 10.3389/fonc.2025.1509207. eCollection 2025.
This study aimed to develop and evaluate multiple machine learning models utilizing contrast-enhanced T1-weighted imaging (T1-CE) to differentiate between low-/high-infiltration of total T lymphocytes (CD3) in patients with rectal cancer.
We retrospectively selected 157 patients (103 men, 54 women) with pathologically confirmed rectal cancer diagnosed between March 2015 and October 2019. The cohort was randomly divided into a training dataset (n=109) and a test dataset (n=48) for subsequent analysis. Seven radiomic features were selected to generate three models: logistic regression (LR), random forest (RF), and support vector machine (SVM). The diagnostic performance of the three models was compared using the DeLong test. Additionally, Kaplan-Meier analysis was employed to assess disease-free survival (DFS) in patients with high and low CD3+ tumor-infiltrating lymphocyte (TIL) density.
The three radiomics models performed well in predicting the infiltration of CD3+ TILS, with area under the curve (AUC) values of 0.871, 0.982, and 0.913, respectively, in the training set for the LR, RF, and SVM models. In the validation set, the corresponding AUC values were 0.869, 0.794, and 0.837, respectively. Among the radiomics models, the LR model exhibited superior diagnostic performance and robustness. The merged model, which integrated radiomics features from the SVM model and clinical features from the clinical model, outperformed the individual radiomics models, with AUCs of 0.8932 and 0.8829 in the training and test cohorts, respectively. Additionally, a lower expression level of CD3+ TILs in the cohort was independently correlated with DFS ( = 0.0041).
The combined model demonstrated a better discriminatory ability in assessing the abundance of CD3+ TILs in rectal cancer. Furthermore, the expression of CD3+ TILs was significantly correlated with DFS, highlighting its potential prognostic value.
This study is the first attempt to compare the predictive TILs performance of three machine learning models, LR, RF, and SVM, based on the combination of radiomics and immunohistochemistry. The MRI-based combined model, composed of radiomics features from the SVM model and clinical features from the clinical model, exhibited better discriminatory capability for the expression of CD3+ TILs in rectal cancer.
本研究旨在开发并评估多种机器学习模型，利用对比增强T1加权成像（T1-CE）来区分直肠癌患者总T淋巴细胞（CD3）的低/高浸润情况。
我们回顾性选取了2015年3月至2019年10月期间病理确诊的157例直肠癌患者（103例男性，54例女性）。该队列被随机分为训练数据集（n = 109）和测试数据集（n = 48）用于后续分析。选取七个放射组学特征生成三个模型：逻辑回归（LR）、随机森林（RF）和支持向量机（SVM）。使用DeLong检验比较这三个模型的诊断性能。此外，采用Kaplan-Meier分析评估高和低CD3 +肿瘤浸润淋巴细胞（TIL）密度患者的无病生存期（DFS）。
这三个放射组学模型在预测CD3 + TILS浸润方面表现良好，在训练集中，LR、RF和SVM模型的曲线下面积（AUC）值分别为0.871、0.982和0.913。在验证集中，相应的AUC值分别为0.869、0.794和0.837。在放射组学模型中，LR模型表现出卓越的诊断性能和稳健性。整合了SVM模型的放射组学特征和临床模型的临床特征的合并模型优于单个放射组学模型，在训练和测试队列中的AUC分别为0.8932和0.8829。此外，队列中较低的CD3 + TILs表达水平与DFS独立相关（P = 0.0041）。
联合模型在评估直肠癌中CD3 + TILs丰度方面表现出更好的区分能力。此外，CD3 + TILs的表达与DFS显著相关，突出了其潜在的预后价值。
本研究首次尝试基于放射组学和免疫组织化学的组合比较三种机器学习模型LR、RF和SVM预测TILs的性能。由SVM模型的放射组学特征和临床模型的临床特征组成的基于MRI的联合模型在直肠癌中CD3 + TILs表达方面表现出更好的区分能力。
