Late recurrence in ER+ breast cancer - refining risk prediction beyond 5 years

Abstract

ER positive (ER+) breast cancer has the potential to recur many years following diagnosis. Predicting who is most at risk of recurrence and when this may occur is complex. Risk prediction incorporates clinical, pathological and molecular modelling, has the potential to alter management and markedly affects the survivorship of women treated for early breast cancer. The research detailed in this thesis was aimed at testing and developing risk prediction methods for women with early breast cancer focussing purely on their prognostic potential for late recurrence. Using retrospective validation studies and prospective utility studies, an existing clinical model was tested in cohorts of women with high stage disease and also women outside of clinical trials. This model was further tested for its utility in a real time MDT meeting setting. Immunohistochemical biomarkers were explored as potential 'add-ons' to clinical models in order to improve precision. Lastly a molecular study involving DNA and RNA sequencing, aimed at discovering biomarkers for very late recurrence, was conducted. As a result of this work, the prognostic value of existing risk prediction models has been confirmed in a real-life cohort and its applicability broadened to include high stage post-menopausal women. The prognostic value of molecular biomarkers for late recurrence in different patient populations has been explored and questions regarding extrapolation of models in different populations have been addressed. Finally, this work identified clinical and molecular candidate biomarkers for very late recurrence including DNA mutations, high RNA expression of proliferation and cell cycle genes and baseline tumour indices such as size and lymph node burden. These would be suitable for ongoing research and development.