Population-based study of peritumoral lymphovascular invasion and outcome among patients with operable breast cancer.

BACKGROUND Lymphovascular invasion has been associated with poor prognosis in women with breast cancer, but it is unclear whether the presence of lymphovascular invasion should be considered sufficient to reclassify breast cancer patients who are at a low risk of recurrence into a high-risk category. METHODS Of the 16,172 patients with operable breast cancer who were entered into the Danish Breast Cancer Cooperative Group Registry from January 1, 1996, to December 31, 2002, lymphovascular invasion was classified at primary diagnosis in 16,121 patients as present (n = 2453, 15%) or as absent (n = 13,206, 82%). Patients with at least one of the risk criteria (positive lymph nodes, tumor size > 2 cm, high grade, hormone receptor-negative tumor, or younger than 35 years) were assigned to the high-risk group; the other patients were assigned to the low-risk group. All procedures, including report forms, central review, and querying, were specified in advance. Kaplan-Meier analyses were used to estimate disease-free intervals and overall survival rates among patients with and without lymphovascular invasion, and multivariable analysis was used to adjust for differences in baseline characteristics and therapy. All statistical tests were two-sided. RESULTS Complete follow-up was achieved for 15,659 patients. The median estimated potential follow-up was 6.4 years for invasive disease-free interval and 7.7 years for overall survival. Invasive disease-free interval and overall survival were statistically significantly associated with lymphovascular invasion within the high-risk group (hazard ratio [HR] for invasive disease = 2.29, 95% confidence interval [CI] = 2.14 to 2.45, P < .001; and HR for death = 2.42, 95% CI = 2.25 to 2.61, P < .001) but not within the low-risk group. At 5 years after surgery, 65.4% (95% CI = 63.5% to 67.3%) and 85.2% (95% CI = 84.5% to 85.9%) of those in the high-risk group with and without lymphovascular invasion were alive; 98.1% (95% CI = 87.6% to 99.7%) and 94.1% (95% CI = 93.2% to 94.8%) of those in the low-risk group with and without lymphovascular invasion were alive. These differences persisted in a multivariable analysis, and for overall survival, a statistically significant interaction (P = .03) was observed between lymphovascular invasion and risk group. CONCLUSIONS In this prospective population-based study, lymphovascular invasion was not an independent high-risk criterion. Lymphovascular invasion should not by itself be considered sufficient to move patients from a low-risk group to a high-risk group.