Background: Differentiation by multigene signatures with excellent performance of hormone receptor (HR)-positive BC largely related to their proliferation genes. Despite questions about its usefulness, there is increasing evidence that Ki67 is a valuable prognostic marker. Standardization of Ki67 pathological assessment is the main problem to interpret reported trials.

The aims of the study are to evaluate the role Ki67 as prognostic marker to predict relapse in HR-positive BC patients (luminal A and B) in adjuvant setting using prospective patients’ cohort. In addition, cut-off value and significant level of Ki67 were investigated comparing with other biomarkers including HER2 in luminal BCs.

Method: We retrospectively analyzed the clinicopathologic characteristics of 1,070 postoperative breast cancer patients including Ki67 and clinical outcomes in terms of relapse free survival (RFS) between 2004 and 2007 at the Samsung Medical Center. Ki67 labelling index was measured in quantitative and semiquantitative method, independently. The percentage of positive nuclei stained for Ki67 was calculated each section based on the approximately 1,000 carcinoma cell nuclei. In addition, Ki67 was graded on a scale from 0 to 4, where 0 = staining of 0-4% of tumor cells, 1 = staining of 5-25% of tumor cells, 2 = 26-50% of tumor cells, 3 = staining of 51-75% of tumor cells, and 4 = staining of more than 76% of tumor cells. ROC curve was drawn to evaluate the usefulness of Ki67 index to get AUC then, find out the proper cut-off value of Ki67 to predict relapse. Multivariate analyses with Cox-regression model were performed. Results: Among 1,564 patients who received curative surgery for invasive breast cancer from January 2004 to June 2007, 1,070 patients with HR-positive were included in this analysis excluding 494 with HER2-enriched or triple negative breast cancer patients. Median follow-up duration was 56.9 months (range 36-77 months). Median age was 46 years (range 22-83 years). Ki67 threshold >19.5%, corresponding to a sensitivity 78.3%, a specificity 51.6% was chosen as cut-off value for relapse in adjuvant patients’ cohort. The AUC was 0.689 (P<0.0001 by Mann-Whitney U test). Overall relapse rate was 5.6%. In univariate analysis by log-rank test for relapse, ER negativity (p=0.010), HER2 positivity (p=0.017), histologic (p=0.001), and nuclear grade (p=0.012), lymphovascular invasion (P<0.0001), TNM stage (0.001), and Ki67 (P<0.0001) were identified risk factors to predict relapse. However, ER negativity, Ki67, and stage were identified as independent risk factors for relapse (Hazard Ratio (HR) 2.7, p=0.031 for ER negativity, HR 3.4, p <0.0001 for Ki67, HR 1.6, p=0.017 for stage) in Cox-regression multivariate analysis. Conclusion: Ki67 could strongly predict clinical outcomes for patients with luminal. 19.5% may be useful cut-off value of Ki67 labelling index. Ki67 may be a better biomarker to predict clinical outcomes than HER2 expression in luminal BCs. ER-/PR+ subset probably has different biology with other luminal BCs. Prospective clinical trials to choose therapeutic option using Ki67 are warranted.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P2-06-01.