Clinical Proteomics of Breast Cancer Unravels a Novel Layer of Breast Cancer Classification

Introduction

Breast cancer classification has been in the focus of numerous large worldwide efforts, analyzing the molecular basis of breast cancer subtypes and aiming to associate them with clinical outcome and improve the current diagnostic routine. Genomic and transcriptomic profiles of breast cancer have been well established, however the proteomic contribution to these profiles is yet to be elucidated.

Material and method

In this work, we performed mass-spectrometry (MS)-based proteomic analysis of primary breast tumors, and integrated with two existing datasets, all generated in our lab using the same approach. The entire cohort consisted of 212 macro-dissected formalin-fixed paraffin-embedded (FFPE) samples originating from different breast cancer subtypes, as well as healthy tissue. Out of the entire cohort, 131 samples were used for downstream analysis, which allowed us to challenge genomics-based classifications.

Results and discussion

Unsupervised analysis identified four proteomic clusters, among them a novel luminal breast cancer subtype, which combines luminal features and key basal signaling components. Interestingly, integration with other datasets provided the clinical relevance and suggested that this subtype is associated with impaired response to therapy, due to elevated reciprocal activity of the estrogen receptor and cancer signaling pathways, such as the PI3K pathway. Comparing our findings to independent published data, we recreated the same classification on reverse phase protein array data. We found that using proteomics was essential to recovering this subtype, which could not be identified with RNA-based approaches.

Conclusion

Altogether, our results both support and challenge the known breast cancer classification, as well as demonstrate the importance of deep proteomic analysis, which may affect cancer treatment decision making.