MicroRNA Signature Distinguishes Early from Late Biochemical Failure in Prostate Cancer
Samantha J Wala, Zsuzsanna Lichner, Annika Fendler, Carol Saleh, Aurfan N Nasser, Dina Boles, Sahar Al-Haddad, Peter Kupchak, Moyez Dharsee, Paulo S Nuin, Kenneth R Evans, Klaus Jung, Carsten Stephan, Neil Fleshner, George M Yousef. Keenan Research Centre in the Li Ka Shing Knowledge Institute of St. Michael's Hospital, Toronto, Canada; University of Toronto, Toronto, Canada; Free University, Berlin, Germany; Ontario Cancer Biomarker Network, MaRS Centre, Toronto, Canada; University Hospital Charité, Berlin, Germany; University Health Network, Toronto, Canada
Background: The introduction of PSA testing led to over-diagnosis and over-treatment of prostate cancer. There is currently no biomarker that can predict disease aggressiveness at time of surgery. miRNAs are short non-coding RNA molecules that were shown to regulate cancer pathogenesis by direct regulation of their target mRNAs. Here we show that expression of miRNAs can be used in a statistical model to accurately predict the risk for biochemical failure.
Design: Study set patients were dichotomized to 'high risk' group (n=26, biochemical failure within 24 months after prostatectomy) and 'low risk' group (n=15, did not have biochemical failure for at least 35 months). miRNAs were ranked from a pool of 754 mature miRNAs by significance using the non-parametric ROC AUC analysis and the permutation t-test. Differential expression of candidate miRNAs was validated on the study set and on an independent patient set by qRT-PCR. Linear regression models were built to predict biochemical failure and were ranked by ROC area under curve.
Results: We identified 25 miRNAs that were significantly differentially expressed between the two risk groups. Based on the expression of 2 or 3 candidate miRNAs, we developed three logistic regression models with a positive predictive value (PPV) of 98.7%. We also validated the best performing model on an independent cohort of patients. Two miRNAs featured in the models, miR-152 and miR-331-3p, were further characterized at molecular level. miR-152 is downregulated in the 'high risk' group and it's ectopic expression reduced cell proliferation of PCa3 and DU145. Target prediction indicated that miR-331-3p and miR-152 can directly target ErBB3 and ErBB2. We validated the interaction by transient transfection of the miRNA mimics followed by qRT-PCR quantification of the predicted targets. Importantly, ERBB2 and ERBB3 present an important alternative pathway for AR activation.
Conclusions: miRNAs are potentially useful biomarkers of prostate cancer progression. Differential expression of miR-331-3p and miR-152 could contribute to the androgen independent activation of AR by targeting the ERBB family.
Category: Genitourinary (including renal tumors)
Monday, March 4, 2013 1:00 PM
Poster Session II # 142, Monday Afternoon