Prediction of Lymphovascular Invasion in Non-Mass Enhancement Breast Cancer Using DCE-MRI and Clinical-Pathological Features

Shiqi Guo¹Kai Zhao²Hu Xu¹Yujiao Xie^1,3Qingyang Li¹Junqing Liang⁴Siyi Chen¹Jiahong Sun¹Zhaofeng Gao^1,3Li Zhu^1*Jiandong Wang^1*

• ¹Department of General Surgery, The First Medical Center of Chinese PLA General Hospital, Beijing, China
• ²The First Medical Center of Chinese PLA General Hospital, Beijing, China
• ³School of Medicine, Nankai University, Tianjin, China
• ⁴Affiliated Hospital, Inner Mongolia Medical University, Hohhot, Inner Mongolia Autonomous Region, China

Abstract Objective: The present study explores the relationship between the distribution patterns of non-mass enhancement (NME) type invasive breast cancer in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and lymphovascular invasion (LVI). Methods: A retrospective analysis was conducted on 192 female patients with NME-type breast cancer who underwent DCE-MRI between January 2019 and December 2023. Based on postoperative pathological results, the patients were divided into two groups: LVI-positive [LVI(+)] (N=50) and LVI-negative [LVI(-)] (N=142). A generalized variance inflation factor (GVIF) analysis was used to identify variables with high multicollinearity. Multivariate logistic regression was used to analyze the risk factors associated with LVI. The performance of the Distribution of NME + ADC + Molecular subtype was evaluated using receiver operating characteristic (ROC) curves and the areas under the curve (AUC). A nomogram was built based on the predictive factors and internally evaluated using a bootstrap resampling method . The performance of the predictive model was evaluated by calibration curve and decision curve analysis (DCA). The DeLong test was applied to compare differences between AUC values, while net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were used to assess the predictive ability of adding the Distribution of NME to the basic model [apparent diffusion coefficient (ADC) + Molecular subtype)]. Results: Compared to focal distribution, patients with linear distribution of NME had a higher risk of LVI positivity (P = 0.030). Distribution of NME + ADC + Molecular subtype demonstrated a relatively strong ability to predict LVI status, with an AUC of 0.723. Compared to the performance of each risk factor alone in predicting LVI, the differences in AUC were statistically significant (P = 0.008, P = 0.006, P = 0.012, DeLong test). The inclusion of Distribution of NME could effectively improve the ability of basic model (ADC + Molecular subtype) to predict LVI, its NRI value was 0.389 (P = 0.013) and IDI value was 0.047 (P = 0.008). Conclusion: Distribution of NME + ADC + Molecular subtype was effective in predicting LVI status, with an AUC of 0.723. The inclusion of Distribution of NME significantly improved its predictive ability for LVI.