On-treatment Biomarkers Can Improve Prediction of Response to Neoadjuvant Chemotherapy in Breast Cancer

Richard J. Bownes; Arran K. Turnbull; Carlos Martinez-Perez; David A. Cameron; Andrew H. Sims; Olga Oikonomidou


Breast Cancer Res. 2019;21(73) 

In This Article


Determining molecular differences between tumours to select the most effective treatment is the defining feature of precision oncology. Accurately predicting which patients will respond to treatment before exposure relies on a highly specific target. In breast cancer, ER status is a good indicator of response to endocrine treatment, but resistance, both primary and acquired, is common. Chemotherapy is an unselective treatment, relying on cancer cells growing faster than normal cells. The results presented here, along with others,[7,8] suggest on-treatment biomarkers have improved value in predicting whether tumours respond to treatment and are associated with the outcome. Changes in gene expression in sequential patient-matched were fairly consistent in response to chemotherapy across two independent datasets, regardless of the response status. Identifying molecular markers between responding and non-responding tumours was much more challenging. We previously demonstrated that lobular and ductal breast cancers respond to endocrine treatment in the same way, despite clear histological and molecular distinctions that are apparent and maintained on-treatment,[22] demonstrating that pre-treatment variations do not necessarily lead to differences in response. The results of this study are somewhat exploratory, rather than definitive, but further illustrate the considerable potential value of on-treatment sampling.

There are no universally agreed-upon markers predictive of response to chemotherapy, and the few that have been investigated in the neoadjuvant setting typically centre around established markers including ER, P53 HER2, and Ki-67;[23] thus, the introduction of new novel biomarkers can expand the currently available clinical options for physicians. A study published over a decade ago stated that the differences in gene expression between responders and non-responders to neoadjuvant chemotherapy must be rather subtle.[12] The results presented here confirm this statement; however, our results suggest that on-treatment biomarkers may provide important information for predicting response.

As cancer is inherently a proliferative disease, measuring the change in markers of proliferation on-treatment is logical and genes like ki-67 have been demonstrated previously to be potentially a new clinical tool for disease prognosis and prediction.[24,25] It is therefore all the more interesting that the potentially novel biomarker identified in this study, AAGAB is not tightly correlated with known markers of proliferation. AAGAB has primarily been studied for its role in punctate palmoplantar keratoderma[26] and the role of adaptin in the clathrin-independent endocytosis of epidermal growth factors. The level of AAGAB was found to be prognostic of response (p < 0.001) in renal cancers (favourably) and in thyroid cancers (unfavourably) from the TCGA study, and expression is elevated in breast cancer, relative to the normal breast (p < 0.001). However, the exact role of AAGAB in breast cancer is currently unclear and potentially warrants further investigation. Clearly, further validation of the role of AAGAB in breast cancer is warranted and will be performed as new neoadjuvant chemotherapy datasets become available. This study supports the use and identification of genes or markers from on-treatment biopsies as a tool for improving patient response classification. We propose that the use of on-treatment samples offers valuable insight into the dynamic changes correlated with response, and submit our findings as support for continued neoadjuvant sampling, and novel biomarker generation.