Ki 67 in Breast Cancer: How Many Tumor Cells Do We Need To Count?
Hala F Faragalla, Linda Feeley, Anita Bane, Anna Marie Mulligan. University Health Network, Toronto, ON, Canada; Cork University Hospital, Cork, Ireland; Juravinski Cancer Center, McMaster University, Hamilton, ON, Canada
Background: Ki67 has been shown to be a prognostic marker in breast cancer with high levels being shown to be associated with poor outcome. A major obstacle to its routine use is the lack of a standardized assay approach which hinders its use as a tool for selecting patient to chemotherapy or assigning them to a particular risk group. The aim of this study is to study how many tumor cells need to be counted on actual visual count for accurate assessment of Ki67 in invasive breast cancer.
Design: 112 invasive breast cancer tumor blocks were selected and stained with MIB1 antibody for assessment of Ki67 labeling index. The Ki67 labeling index was estimated by visual counting of 1000 tumor cells from three different areas of each tumor block including hotspots and invasive tumor front. The count per each 100 is recorded. Statistical analysis was performed using paired student t-test with a P-value < 0.05 considered significant.
Results: Our results show that on visual counting Ki 67 LI is higher in the initial 200-300 and 500 tumor cells, which tends to gradually decrease due to dilution effect. There was a statistically significant difference on counting 200, 300 and 500 tumor cells at a P value of 0.001, 0.02 and 0.03 respectively. However, there was no statistically significant difference on counting 700 and 800 versus 1000 tumor cells. Dichotomization into categories of high and low using 20% as a cut off value resulted in 7 discrepant results with Ki67 higher at 300 than 1000 tumor cells. The mean Ki67 proliferation value for the initial 300 tumor cells are 21.9 and for final 1000 tumor cells are 20.9 respectively.
Conclusions: In this study there was a decrease in proliferation value with increasing number of evaluated tumor cells. This is due to dilution effect. This decrease is statistically significant. The dilution effect is important in clinical practice if samples are dichotomized as only proliferation values near a chosen cut-off would be affected.
Tuesday, March 5, 2013 1:00 PM
Poster Session IV # 53, Tuesday Afternoon