The Long-term Prognostic Performance of Ki-67 in Primary Operable Breast Cancer and Evaluation of Its Optimal Cutoff Value.

Markers of proliferation are considered to have prognostic importance in breast cancer (BC). The Ki-67 index has been reported as a prognostic factor, but standardized cutoff values and counting methods are not yet established. We assayed the Ki-67 labeling index (LI) of 589 consecutive operable BC patients who underwent surgical resection. Ki-67 immunostaining was performed, and the LI was manually counted using an image processing program. We also compared the manual cell count (MCC) of Ki-67 to the whole-section eyeballed estimate count (EEC). Univariate survival analysis showed statistically significant differences in long-term BC-specific survival in the following factors: tumor size, histologic grade, nuclear grade, lymph node metastasis stage, estrogen receptor, progesterone receptor, human epidermal growth factor-2 (HER2) status, and intrinsic type (P<0.05). The MCC and EEC Ki-67 evaluations were statistically well correlated (Pearson correlation=0.683, P<0.001). Their agreement rate was highest at a 20% cutoff (κ-coefficient=0.464). With cutoff values of 14% and 20%, high Ki-67 LI was associated with poor BC-specific survival (P=0.028 and 0.012, respectively), and a 20% cutoff had a higher hazard ratio. High Ki-67 LI with a 20% cutoff was also associated with poor survival in the hormone receptor-positive and lymph node-positive subgroups (P=0.015 and 0.016, respectively). In conclusion, Ki-67 LI had prognostic significance, especially in hormone receptor-positive and lymph node-positive BC patients. EEC was relatively reliable counting method with a higher cutoff value. We suggest that the 20% cutoff value be the preferable value in clinical practice.