Background: The Ki67 labeling index (LI) reflects the proliferative activity of breast cancers and defines luminal A and B tumors; however, no detailed method to measure Ki67 has been standardized. Here, we propose a fast and easy way to evaluate Ki67.
Methods: Immunohistochemical staining of estrogen receptor (ER), progesterone receptor (PgR), HER2 and Ki67 (MIB-1) was performed on 235 primary invasive ductal carcinomas. For each sample, a hot spot with many Ki67+ cells was identified using a low-power field (40×, 4× objective). Three independent areas in high-power field (400×) were selected at the hot spot, and all cancer cells in the 3 areas were manually counted to calculate LI (% Ki67+ cells). Alternatively, micrographs taken at 100× and 200× fields including the hot spot were shown to 2 pathologists, who visually assessed percentages of Ki67+ cells in 10 % intervals at a glance (Eye-10).
Results: Eye-10 and LI were strongly correlated (r = 0.9412, P < 0.0001). All cases of Eye-10 ≥ 30 % had LI > 14 %; most of those <10 % had LI < 14 %. Of 170 ER+/HER2- tumors, Eye-10-based subtypes matched 87 % of LI-driven subtypes, and interobserver agreement was good (κ = 0.705).
Conclusion: Eye-10 is far easier than counting many cancer cells and useful for classifying breast cancers. Eye-10 can exclude obviously high and low Ki67 cases, leaving a "gray zone" around a cutoff point. Combining Eye-10 and manual counting is a good candidate for a standard method to evaluate Ki67.