Evaluating Tumor‐Infiltrating Lymphocytes in Breast Cancer Using Preoperative MRI‐Based Radiomics

Abstract

Background Evaluating tumor‐infiltrating lymphocytes (TILs) in patients with breast cancer using radiomics has been rarely explored. Purpose To establish a radiomics nomogram based on dynamic contrast‐enhanced (DCE) magnetic resonance imaging (MRI) for preoperatively evaluating TIL level. Study Type Retrospective. Population A total of 154 patients with breast cancer were divided into a training cohort (N = 87) and a test cohort (N = 67), who were further divided into low TIL (<50%) and high TIL (≥50%) subgroups according to the histopathological results. Field Strength/Sequence 3.0 T; axial T2‐weighted imaging (fast spin echo), diffusion‐weighted imaging (spin echo‐echo planar imaging), and the volume imaging for breast assessment DCE sequence (gradient recalled echo). Assessment A radiomics signature was developed from the training dataset and independent risk factors were selected by multivariate logistic regression to build a clinical model. A nomogram model was built by combining radiomics score and risk factors. The performance of the nomogram was assessed using calibration curves and decision curves. The area under the receiver operating characteristic (ROC) curve, accuracy, sensitivity, and specificity were calculated. Statistical Tests The least absolute shrinkage and selection operator, univariate and multivariate logistic regression analysis, t‐tests and chi‐squared tests or Fisher's exact test, Hosmer–Lemeshow test, ROC analysis, and decision curve analysis were conducted. P < 0.05 was considered statistically significant. Results The radiomics signature and nomogram model exhibited better calibration and validation performance in the training (radiomics: area under the curve [AUC] 0.86; nomogram: AUC 0.88) and test (radiomics: AUC 0.83; nomogram: AUC 0.84) datasets compared with clinical model (training: AUC 0.76; test: AUC 0.72). The decision curve demonstrated that the nomogram model exhibited better performance than the clinical model, with a threshold probability between 0.15 and 0.9. Data Conclusion The nomogram model based on preoperative MRI exhibited an excellent ability for the noninvasive evaluation of TILs in breast cancer. Level of Evidence 4 Technical Efficacy Stage 2

Discriminating malignant from benign testicular masses using machine-learning based radiomics signature of appearance diffusion coefficient maps: Comparing with conventional mean and minimum ADC values

“…Herein, we mined high-throughput quantitative features, including intensity-based, structural, texture-based, and wavelet transform-based features, from ADC maps. Meanwhile, we selected the most effective features and developed a classification model using the machine learning algorithm [27]. Interestingly, our results showed that among the six radiomics features enrolled in the radiomics signature, minimum ADC values yielded the highest absolute value of the coefficient.…”

“…However, histogram analysis extracts the distribution of densities only. It cannot fully explore the potential value of imaging, for instance, the shape features and the spatial heterogeneity of the lesions [27,28]. Herein, we mined high-throughput quantitative features, including intensity-based, structural, texture-based, and wavelet transform-based features, from ADC maps.…”

Radiomic models based on magnetic resonance imaging predict the spatial distribution of CD8+ tumor-infiltrating lymphocytes in breast cancer

“…Several RMs based on MRI or mammography have been reported to predict the infiltration of stromal TILs in breast cancer (17)(18)(19)25). Our data suggest that the immune-excluded tumors showed worse survival outcomes than tumors with other immunophenotypes, despite the similar level of stromal TILs…”

“…Since dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is usually acquired for the initial diagnosis of breast cancer, RMs using the DCE-MRI can be clinically used. Although a few studies have predicted the level of TILs using RMs (17)(18)(19), radiomics has not been applied to discriminate the spatial contexture of the tumor-immune microenvironment.…”

Prognostic value of tumor-infiltrating lymphocytes in DCIS: a meta-analysis

“…Meanwhile, investigating whether TILs in DCIS could be used to stratify patients with different risk along with traditional risk factors such as tumor grade, tumor size, and age is also meaningful. Aside from pathological detection, according to Tiantian Bian et.al’s study, the preoperative MRI-based radiomics signatures are valuable in evaluating the TILs level in breast cancer [ 34 ]. Currently, NCT03495011 trial is trying to investigate the MRI-based radiomics signatures that correlate with pathologic markers in DCIS.…”