Ki-67

Various indices of cellular proliferative activity have been investigated. Mitotic counts only detect cells in the M phase, are dependent on the period of time between surgical removal and fixation of the specimen, and suffer from heterogeneous distribution and confusion between mitoses and nuclear pyknosis and karyorrhexis. Counting nucleolar organiser regions (AgNORs) and 5-bromodeoxyuridine or tritiated thymidine labelling are also not practical in a routine situation. Ki-67 recognises a proliferation specific nuclear antigen10.

Immunohistochemical expression

Ki-67 is expressed by proliferating cells in late G1, S, G2 and M phases, but not in resting cells in G0. Staining is commonly nucleolar or perinucleolar. In conjunction with Ki-S2, it may be used to derive a cell cycling ratio.

The prototypic monoclonal antibody works only in frozen sections. Subsequently, antibodies against formalin-resistant epitopes, MIB-1 and NCL-Ki-67-MM1 (MM1), have been raised. Polyclonal antibodies such as NCL-Ki-67p and rabbit anti-human (Rah) Ki-67, are also immunoreactive in paraffin sections. Staining is nuclear and may be diffuse or granular, or a combination1. There are good correlations between immunoreactivity using monoclonal Ki-67 in frozen section, polyclonal antibodies in paraffin sections and other indices of cell proliferation. In one study1, MIB-1 gives a higher labelling index than other antibodies, with good inter-reading reproducibility, although this was not reproduced in a second study2, which showed considerable variation between readings. It is therefore important to identify the antibody used in every study.

	prototypic Ki-67	MIB-1		MM1		NCL-Ki-67p		Rah Ki-67
tonsillar squamous epithelium1		granular & diffuse	diffuse only	granular & diffuse	diffuse only	granular & diffuse	diffuse only	granular & diffuse	diffuse only
		30.8 to 30.9%1	18.2 to 18.3%1	13.9 to 14.50%1	5.8, 5.8%1	20.2 to 21.2%1	7.2 to 7.2%1	16.6 to 17.9%1	7.7 to 7.8%1
glioblastoma	5.7%1	6.1 to 9.4%(granular staining)1		7.9%(granular staining)1		5.9%(granular staining)1		5.2 to 10.9%(granular staining)1

Immunoreactivity has therefore been shown not to be fixation dependent, an advantage over PCNA.

Lymphoproliferative disorders may show characteristic patterns of proliferation as assessed by the Ki-67 staining7:

	diagnosis	pattern of staining with Ki-67
	Resting lymphoid tissue	The proliferation rate is low. Distinction from follicular lymphoma may be made by the negativity of the follicles for CD10 and bcl-6. (bcl-2 may be positive.)
	Follicular hyperplasia	Follicles show a high level of staining, with a polarity; there is almost 100% staining of nuclei in the dark zone, which tends to form a meniscus. The borders of the follicles are sharply demarcated from the mantle which has a low proliferation rate. Some follicles may lack the typical pattern, so multiple follicles should be assessed: this is the case if the section is purely through the light zone.
	Mixed pattern hyperplasia	Follicles show a high level of staining and interfollicular areas a low to moderate level
	Progressive transformation of germinal centres	There are large follicles within which there are small aggregates of proliferating cells: the non-proliferating cells are T-cells or mantel zone B-cells. The pattern overlaps with that of the floral variant of follicular lymphoma. The few large follicles showing PTGC are present alongside obviously benign reactive follicles.
	Castleman's disease	A nodular pattern of staining of low intensity, similar to follicular lymphoma, but the nodules are smaller and more widely spaced.
	Follicular lymphoma	Follicles show a lower level of staining than in follicular hyperplasia and polarity is lacking. Margins are ill-defined. Grade 3 lymphomas may have a proliferation rate approaching that of benign follicles but are distinguished by the uniformity of the pattern among many follicles.
	Floral variant of follicular lymphoma	There are large follicles within which there are small aggregates of proliferating cells: the non-proliferating cells are centrocytes. The pattern overlaps with that of progressive transformation of germinal centres. In addition to the floral type follicles, there are typical neoplastic follicles.
	Marginal zone lymphoma	Diffuse low level of (interfollicular) staining, with or without residual follicles. Colonisation of follicles gives a pattern similar to that of follicular lymphoma.
	Small lymphocytic lymphoma	Ki-67 accentuates the proliferation centres, which may resemble the pattern of neoplastic follicles.
	Burkitt and Burkitt-like lymphoma	proliferation index approaching 100%
	Transformation of indolent lymphomas	May be suggested by an anomalously high proliferation rate
	T-cell rich B-cell lymphoma	The neoplastic cells are highlighted
	Hodgkin lymphoma
	Angioimmunoblastic T-cell lymphoma

Hyalinising trabecular tumour of the thyroid shows cytoplasmic staining with MIB-1
Breast carcinoma: the value of Ki-67 as a prognostic marker is controversial. Breast carcinoma: tumors with high Ki-67 proliferation indices (> 20%) were associated with a higher 4-year probability of relapse of disease and death when compared with tumors with low Ki-67 values. In addition, this proliferative parameter maintained its prognostic significance when the patients were stratified according to lymph node involvement, menopausal status, and nuclear estrogen receptor content.6 In another study, it proved an independent prognostic marker on multivariate analysis along with nodal status, tumour size and histological grade9. It was not a statistically significant prognostic factor if mitotic count is included in the analysis. However, there are cases where the Ki-67 fraction and mitotic count give disparate results: a low standardized mitotic index combined with a high Ki-67 fraction is associated with a favorable prognosis9.

Diagnostic utility

The rate of cell proliferation as assessed by Ki-67 immunoreactivity has been studied as a prognostic indicator in numerous malignant neoplasms and shown to correlate with tumour grade and clinical course. This includes:

Non-Hodgkin lymphoma: low grade lymphomas have a labelling index of less than 20%, high grade lymphomas greater than 20%. Low grade lymphomas with a labelling index in excess of 5% have a worse prognosis than those with an index of less than 5%. In particular, grade 1 and 2 follicular lymphomas with a higher than usual proliferation index have similar clinical behaviour to high grade follicular lymphoma8. In small lymphocytic lymphoma, cell proliferation assessed by MIB1 staining may correlate with more aggressive behaviour11. However, high grade lymphoma with a very high index (>80%) tend to have a better prognosis than high grade lymphomas with a lower index, being less likely to relapse if remission is achieved.3 In Burkitt and Burkitt-like lymphoma, the nuclear labelling by Ki67 is nearly 100% and this can be used as a diagnostic criterion6. The pattern of proliferation as assessed by Ki-67 may be diagnostically useful in lymphoproliferative disorders (vide supra)7.
Glioma: the indices ranged from 0% to 4.5% for 16 low grade astrocytomas; from 0.7% to 7.4% for 8 anaplastic astrocytomas; and from 1.7% to 32.2% for 27 glioblastomas. The differences among the means of each group are statistically significant. Five patients with malignant gliomas with an index of less than 2.5 had survival times of more than 40 weeks.4
Soft tissue sarcoma: Ki-67 index (the number of Ki-67-positive tumor cells/10 HPF) positively correlated with mitotic count , cellularity and the histological grade. The Ki-67 low index group (less than 50/10 HPF) showed a more favorable prognosis than the high index group (more than 50/10 HPF). Three cases with low mitotic count and unfavorable prognosis were proved to be in the Ki-67 high index group (142-382/10 HPF).5
In combination with p16 in the identification of anal intraepithelial neoplasia.