Relationship of Ki67 to tumor size and lymph node metastasis in breast cancer.
e21076 Background: Ki67, a tumor proliferation marker, has demonstrated usefulness in breast cancer prognosis. Prior work with BrdU labeling for cell proliferation in breast cancer has not settled the question whether cell proliferation labeling is independent of other tumor features like tumor size and presence of axillary metastases, see Rew (1992) vs. Thor et al (1999). Methods: We analyzed retrospectively our database of 379 patients for correlation between the and tumor size, presence of axillary lymph node metastases and the percentage of Ki67 positve cells. We used linear and parabolic regression to correlate tumor size with the Ki67 index and receiver operator characteristics curve to correlate the presence of axillary lymph node metastases with the Ki67 index. Results: A very weak linear relationship was detected between tumor size and Ki67 index. The R-squared coefficient was 0.03, indicating that tumor size explained only 3% of the variability in measured Ki67 indexes. The ROC analysis, looking at the correlation between Ki67 and lymph node metastasis, yielded an area under the curve (Az) of 0.53. This indicates a very weak correlation. No relationship between axillary lymph node metastasis would yield an Az of 0.5 and a perfect correlation would yield an Az of 1. Conclusions: Tumor size and axillary lymph node metastasis explain less than 10% of observed Ki67 index variability. Therefore, in breast cancer, the prognostic value of Ki67 is additive to that of tumor size and lymph node involvement.