Information Sciences in Imaging 2012 Seminar Series: December


Genomics of breast cancer and personalized cancer medicine


Today, breast cancer is appreciated as a group of molecularly distinct neoplastic disorders. Genomics has shown the potential to revolutionize the diagnosis and management of breast cancer by offering an unprecedented comprehensive view of the molecular underpinnings of pathology. 

In this talk, I will illustrate that gene expression signatures have tremendous power to identify new cancer subtypes and to predict clinical outcomes. To utilize the combined strength from individual gene signatures, we developed an analytical framework to aggregate the predictive power of multiple signatures. By capturing a broad spectrum of biological aspects in heterogeneous breast tumors, our strategy improves risk stratification for cancer patients.

Furthermore, we show that risk stratifications within each ER stratum are driven by distinct molecular mechanisms. There are needs for designing robust prognostic tools separately for ER-positive and ER-negative disease. Interestingly, in the context of long-term survival, incorporating characteristics for tumor aging may help to alleviate the time-dependency effect of signatures and achieve a lasting and persistent performance in survival prediction. 

Through this work, we have tried to underscore the importance of computational analysis in transforming the masses of genomic data into new biological insights, which in turn guides the analysis. The ultimate goal is to provide a more comprehensive understanding of cancer and to forecast outcomes and therapeutic efficacy in the clinic.