Background
A big challenge of breast cancer prevention lies in the discovery and validation of biomarkers to identify individuals at the highest risk and to evaluate the impact of intervention strategies. We are looking for biomarkers of risk linked to the epigenome, that is at the heart of cancer onset and progression. The epigenome is highly responsive to exposures, making it an excellent resource for risk assessment. However, its responsiveness to environmental factors requires distinguishing actual evidence of elevated risk from unrelated epigenetic changes. For this project, we established novel in silico and in vitro model-based identification strategies. We focused on epigenetically-regulated microRNAs (miRs) to be used as biomarkers, because of their stability in the bloodstream, and their ability to reflect almost in real time the functionality of tissues in general, and the progression of cancers in particular.
Objectives
We hypothesize that with the handling of appropriate models, miRs can be identified as biomarkers for breast cancer risk stratification and prevention.
Our objectives are (1) to establish lists of potential biomarkers from in silico analyses and 3D cell culture models; (2) to assess the differential expression of these biomarkers in the plasma samples from cohorts built in different countries.  
Methods
An in silico analysis from the TCGA breast cancer (TCGA-BRCA) database has enabled us to select 11 miRs, related to pathways enriched in breast cancer, that are differentially expressed in tumors between breast cancer-free and breast cancer Stage I patients between 20 and 50. We also used 3D cell culture-based breast cancer risk progression through polarity disruption and chronic reactive oxygen species (ROS) exposure, and performed miR sequencing. Seven miRs with expression modified by risk increase were selected following differential expression analysis. 
Expression analysis of these 18 miRs was performed by RT-qPCR in three cohorts: an early risk cohort with sampling 5 years before diagnosis (from the USA), and stage I and metastatic breast cancer cohorts (from France). Aged-paired controls without breast cancer were also included. Normalization was performed against housekeeping genes (RNUG5 or UniSp2, depending on cohorts).
Results
Eighty-three percent (15/18) of the miRs are differentially expressed in Stage 1 breast cancers, and 10 of these miRs showed significantly different expression also in metastatic breast cancers. Within this set of miRs, two (including oncomiR-182-5p) were also identified as differentially expressed in women who developed breast cancers 5 years following sampling.
Conclusions and perspectives
Current findings from this project suggest that targeted in silico analysis and 3D cell culture models are effective for preclinical identification of potential biomarkers for risk stratification. They also illustrate the biological connection between cancer initiation and progression, indicating the value of integrating this continuum in prevention strategies.  We are currently working with Taiwan and Uruguay, and low- and middle-income countries, Lebanon, and Ghana, to identify which of these biomarkers are common and which are population-specific, possibly reflecting a particular context for breast cancer development in these countries. Ultimately, the preclinical identification process may be tailored to exposomes to identify environment-specific biomarkers of breast cancer risk.