Evaluating the predictive value of dynamic contrast-enhanced MRI and six diffusion models for prognostic factors in mass-type invasive ductal carcinoma of the breast

Accurate prognostic stratification of invasive ductal carcinoma (IDC) of breast is essential for personalized breast cancer treatment, but single-index DWI fails to capture the complex non-Gaussian diffusion of intratumoral water molecules. This study aims to evaluate the diagnostic efficacy of combining dynamic enhanced magnetic resonance imaging (DCE-MRI) features and quantitative parameters from six diffusion models in predicting prognostic factors of mass-type breast invasive ductal carcinoma (IDC). 100 patients with mass-type IDC who underwent routine breast MRI and multi-b-value DWI examinations in our hospital were enrolled. All patients were confirmed pathologically and had immunohistochemical results for Ki-67 and tumor-infiltrating lymphocytes (TILs). Among them, 15 cases exhibited low Ki-67 expression, while 85 cases showed high Ki-67 expression; 54 cases were categorized in the low TILs group, and 46 cases were classified in the medium-high TILs group. Clinical features, routine MRI features, and parameters from six diffusion models, including continuous-time random walk (CTRW), fractional order calculus (FROC), stretched exponential model (SEM), intravoxel incoherent motion magnetic resonance imaging (IVIM), and diffusion kurtosis imaging (DKI), were recorded. Independent predictive factors were identified through multivariate regression analysis. Receiver operating characteristic (ROC) curves were constructed to evaluate and compare the diagnostic efficacy of both independent and combined parameters. The Delong test was used to compare the diagnostic efficacy of each model. Multivariate logistic regression analysis identified αCTRW as an independent predictor of Ki-67 and TILs levels, with AUCs of 0.732 and 0.649, respectively. Additionally, the combination of time signal curve (TIC) and αCTRW for predicting TILs levels yielded an AUC of 0.700. There was no statistically significant difference between the AUCs of αCTRW and the combined diagnostic model (αCTRW + TIC). CTRW-α is a promising non-invasive imaging biomarker for predicting Ki-67 and TILs levels in mass-type IDC. By capturing intratumoral water molecule diffusion heterogeneity, it supports accurate preoperative prognostic stratification and personalized treatment decision-making, highlighting the value of multi-exponential DWI models in breast cancer imaging. Not applicable.