Summaries of logistic regression models for individual MiRNAs in the discovery and validation cohorts accounting for age, PSA, stage and biopsy characteristics.

*MiRNAs were represented in models as binary indicators, with cut-offs selected using a classification tree.**Estimated odds ratio from a logistic regression also controlling for age, PSA, stage and degree of biopsy involvement.†Estimated area under the ROC curve from a logistic regression also controlling for age, PSA, stage and degree of biopsy involvement.‡P-value comparing AUC for model including for miRNA, to model including only age, PSA, stage and degree of biopsy involvement.