Methodological and analytical aspects of proliferation assessment in breast cancer

Abstract

Proliferation is a major determinant of prognosis in breast cancer and plays a key role in individualised treatment of the disease. While patients with tumours showing low proliferative activity can be spared chemotherapy in the early stages of the disease, a more aggressive treatment should be considered in patients with highly proliferating breast cancers. The most commonly used markers of proliferation are histological grade (including mitotic count), and Ki67- and PHH3-Index. However, a generally accepted and standardised proliferation assessment method has not yet been established for clinical decision making, and little is known about factors that could potentially compromise its standardisation. Therefore, we tested the impact of different Ki67 assessment methods on resulting Ki67 levels, investigated the extent of intratumoral heterogeneity of Ki67 expression and surveyed the status quo of interlaboratory variability of Ki67 staining in breast cancer. Then, as tumour proliferation in core needle biopsies had been previously reported to be lower than in subsequent surgical excisions, we also investigated whether increased biopsy volume would result in higher concordance rates between the specimens. In addition, we tested the performance of four commercially available immunohistochemical PHH3 antibodies for marking mitotic figures in a series of highly proliferating breast cancers. Our data show that Ki67 levels are highly influenced by both laboratory-specific analytic variables and selection of assessment protocols. For the latter, the major cause for differing results was intratumoral heterogeneity of Ki67 expression, which exists across all breast cancer subtypes and exceeded even variability between tumours. Differences in Ki67 staining performance between pathology labs, however, may be attributed to a lack of a morphologic correlate for proliferating cells (except for mitoses), which could be used as an internal control for Ki67 staining. Interestingly, the reliability of histological grade in core needle biopsies improved with increasing biopsy sample size while reliability of Ki67 levels in core biopsies was unaffected by biopsy volume but did not show higher concordance between core biopsies and subsequent surgical excisions than for histological grade in general. Sensitivity and specificity of the tested PHH3 antibodies to detect mitoses varied substantially in our study, showing insufficient performance in two of four antibodies. In conclusion, our data support the recommendation of international expert panels to use Ki67 levels for therapeutic decisions only in the context of laboratory specific reference values and in full awareness of analysed specimen type. Furthermore, our results emphasize the importance of correlation with histomorphology when immunohistochemical stains are used and interpreted for proliferation assessment. Our findings may contribute to a better understanding of methodological issues in individualised patient care that are currently highly debated, and may help to develop an international standard for proliferation assessment in breast cancer.