As an adverse prognostic factor in patients with invasive breast cancer, lymphovascular invasion (LVI) is closely associated with axillary lymph node (ALN) involvement, disease recurrence, and distant metastasis. This study aims to investigate the relationship of preoperative MRI characteristics with LVI and disease-free survival (DFS).

This retrospective study included patients with available LVI status who underwent preoperative MRI examination at two hospitals from January 2013 to August 2020. Univariable and multivariable logistic regression analyses were performed to evaluate the independent predictors of LVI. A prediction model was established and a scoring system was carried out according to the odds ratio of multivariable analysis in the primary set, and the results were further verified in the validation set. The Kaplan-Meier method and log-rank test were applied to analyze the association of the risk score with DFS.

A total of 575 women (median age, 50 years; range, 24-79 years) were included. LVI-positive status was confirmed in 26.9% (104 of 386) and 27.5% (52 of 189) of the primary and validation set. At the primary set, Ki-67 index, MRI-ALN status, breast edema score, multicentricity or multifocality showed significantly independent values in predicting LVI status (P lt; 0.05). The area under the receiver operating characteristic curve of the model using these predictors in the primary and validation set was 0.75 (95% confidence interval [CI]: 0.69, 0.81) and 0.75 (95% CI: 0.67, 0.83), respectively. Across the entire study population, higher grades of simplified scores tended to have a higher proportion of positive LVI status. Furthermore, patients with high-score were significantly associated with a worse DFS (HR for high vs. low, 3.08; 95% CI: 1.32-7.18, P = 0.006).

The prediction model incorporating Ki-67 index, MRI-ALN status, breast edema score, multicentricity or multifocality showgood predictive performance for LVI statuand iassociated with DFin patientwith invasive breast cancer.

A feasible predictive model combining preoperative breast MRI characteristics and clinicopathological factors can help identify the presence of lymphovascular invasion in patients with invasive breast cancer. Furthermore, the variables used in our model are readily available in routine clinical practice and could help optimize clinical workflows.