B65
Evaluation of KI67 as a marker of Luminal B tumours in breast cancer
Fiona M Blows1, Kristy E Driver1, Sarah-Jane Dawson1, Elena Provenzano2, John Le Quesne3, Carlos Caldas3, Paul D P Pharoah1
1University of Cambridge, UK, 2Addenbrookes Hospital NHS Trust, Cambridge, UK, 3Cancer Research UK Cambridge Research Institute, Li Ka Shing Centre, Cambridge, UK
Background
Breast cancer is a heterogeneous disease and gene-expression profiles are currently regarded as the gold standard for classification into five major categories. Standard IHC markers can be used to recapitulate this classification. However, the best markers to use have not yet been identified. In particular, markers to divide the hormone receptor positive tumours into the luminal A and luminal B categories are imperfect. Ki67 has been shown to be an indicator of proliferative activity and has been proposed as a marker for the luminal B subtype of breast cancer and the aim of this study was to evaluate the role of Ki67 in subclassifying ER positive tumours in a large case series.
Method
We used tissue microarrays to stain 1596 tumours from the SEARCH study for Ki67. Data on tumour size, grade, node status, and PR, ER, HER2 status were also available. A tumour was KI67 positive if more than 33 percent of nuclei stained positive.
Results
Thirteen percent of tumours (n=215) were KI67 positive. In the luminal subgroup of tumours (ER or PR positive) that are HER2 negative (n=1,079) seven percent (71) were KI67 positive. KI67 was associated with a poorer prognosis in this subgroup (HR=1.8, 95%CI 0.96 3.3, P=0.07), but the Cox proportional hazards assumption was violated (P=0.037). KI67 was therefore treated as a time varying covariate (log hazard ratio varying linearly with log time) and found to have a significant association with prognosis (P=0.04) with the associated hazard ratio decreasing with increasing time. This effect was markedly attenuated and no longer significant after adjusting for tumour size, node status and tumour grade.
Conclusion
KI67 identifies a poor prognostic sub group of ER positive tumours, but its effect is significantly attenuated over time and is much weaker when adjusted for other tumour characteristics.
