Ki-67 Expression after Tamoxifen Given Preoperatively Predicts Disease Free Survival and Overall Survival in Women with Operable ER-Positive Breast Cancer.

Abstract

Abstract We have previously shown in a window of opportunity, 3-arm, double-blind phase II randomized trial conducted in a single Institution that tamoxifen given for 4 weeks between tru-cut biopsy and surgery at a daily dose of 5 or 1 mg day was equivalent to 20 mg in decreasing ki-67 expression. A total of 31 premenopausal and 89 postmenopausal women with operable stage I-II ER-positive breast cancer were recruited (Decensi A et al, JNCI 2003;95:779-90). Here we analyzed the prognostic and predictive significance of ki-67% after a median follow-up of 86 months (11-111) in the same cohort of 120 women who underwent adjuvant treatment according to St. Gallen guidelines. At the current follow-up time DFS was 76% and OS was 92%. The median (IQ range) post-treatment ki-67% value was 12% (7-17) in women free of recurrence versus 20% (11-29) in women who relapsed (p=.0014). Post-treatment (surgery) ki-67% was positively associated with worse prognosis in univariate analysis (log-rank test =.0009, with ki-67% categorized in three groups: <7 (n=24), 7-19 (n=63), ≥20 (n=29); Figure).Post-treatment ki-67% was significantly associated with breast cancer death as well (p=.005, ki-67% <20 versus ≥20). In multivariate Cox analysis of DFS, both pre- and post-treatment ki-67 used as continuous variables were significantly associated with a higher risk of recurrence independent of adjuvant treatment, estrogen receptor level and pre-surgical tamoxifen dose (p=.002 and .04, respectively). Women with post-treatment ki-67% in the middle tertile (7-19%) and top tertile (≥20%) had a HR=2.1 (0.5-9.7) and HR=6.7 (1.5-30.2) of recurrence compared with women in the bottom tertile, respectively. There was no significant difference among tamoxifen dose groups on recurrence or death. In multivariate Cox analysis of OS, both pre- and post-treatment ki-67 used as continuous variables were significantly associated with a higher risk of breast cancer death (p=.0453 and .003, respectively). The risk of breast cancer death was 5 times higher in women with post-treatment ki-67 >20% than in women with ki-67<20% (HR=5.26, 1.22-23, p=.03). Our findings indicate that ki-67 is a prognostic and predictive marker in women with operable ER-positive breast cancer. The modulation of Ki-67% after a few weeks of hormonal intervention before surgery is a promising surrogate biomarker for outcome prediction and provides the opportunity to screen active agents in a cost-effective manner. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 406.