Aims Personalised breast cancer therapy requires pathological characterisation of tumours. The proliferative index, based on Ki67, is pivotal, but a standard method has not been established. Here we look for an easy and practical way to evaluate Ki67.

Methods Immunohistochemical staining of estrogen receptors, progesterone receptors, HER2 and Ki67 (MIB-1) was performed on resected specimens from 406 primary invasive ductal carcinomas. Ki67 labelling index (LI) from manual counting was compared with visual assessment using a 5-grade scale (Eye-5). Next, 10 pathologists evaluated 100 samples with marked hot spots by using Eye-5. Another 100 samples without marking were also assessed by eight pathologists. One year later, two pathologists reviewed 222 cases with Eye-5. Prognosis was analysed among estrogen receptor-positive cases with postoperative endocrine therapy.

Results Eye-5 showed good correlation to LI. All 136 cases of score 4–5 had LI >20% and all 56 cases of score 1 had LI<20%, which means that manual counting was not necessary for about half of the cases. Interobserver and intraobserver variability was low even when a hot spot was not fixed. Eye-5 also correlated with histological grade and lymph node metastasis. Combining Eye-5 and histological grade created a new algorism to predict LI, which allows 80% of all cases (74% of luminal cases) without manual counting. Cases of Eye-5 score 1–2 had significantly better survival than score 3–5.

Conclusions Visual assessment of Ki67 by a 5-grade scale (Eye-5) is fast, easy, and reliable with acceptably low interobserver and intraobserver variability. Eye-5 can replace LI in many luminal tumours, and is a strong candidate as a standard method of evaluating Ki67.