Abstract
BACKGROUND: Two conflicting hypotheses as to how breast cancer (BC) accesses the systemic circulation dominated the 20th century and affected surgical treatment. We hypothesized that tumor lymphovascular invasion (LVI) at the primary tumor site favors lymphatic and not blood vessel, capillaries, and systemic metastases (Smets) are dependent upon regional lymph node (RLN) mets. METHODS: Data from BC patients undergoing RLN biopsy was professionally abstracted and maintained in a prospective, precisely managed, single-institution database. Associations of RLN, LVI, and Smets were estimated by univariate and multivariate backward logistic regression models and patient-affiliated demographic, clinicopathologic, treatment type, and molecular marker data. RESULTS: Of 3329 patients, followed 1-22 years (mean 7.8), 463 of 3329 (13.9%) showed LVI, 742 of 3329 (22.3%) had RLN mets, and 262 of 3329 (7.9%) had Smets. Smets occurred in 52 of 252 (21% with LVI+/RLN+); 116 of 2301 (5% with LVI-/RLN-); 65 of 465 (14% with LVI-/RLN+); and 17 of 207 (8% with LVI+/RLN-), p = 0.021 for association between LVI and Smets for RLN+ patients but not for RLN- patients (p = 0.051). Positive RLN, larger tumor size, and higher grade (all p < 0.001) were predictive of Smets by the multivariable model, whereas positive LVI was not. CONCLUSIONS: LVI predicts RLN mets in BC. RLN is critical to Smets from BC, whereas LVI on its own is not. Smets occur significantly more commonly when both LVI and RLN mets occur together. LVI is, thus, likely to be primarily lymphatic invasion, and rarely, blood vessel invasion, supporting the Halsted paradigm. LVI and RLN together predict clinical outcome better than either alone.