Published online Oct 7, 2014. doi: 10.3748/wjg.v20.i37.13325
Revised: June 4, 2014
Accepted: June 26, 2014
Published online: October 7, 2014
Processing time: 343 Days and 16.2 Hours
Core tip: Biomarkers are statistically identified as significant by: (1) classical statistical tests where each biomarker is independent from the others; and (2) multivariate methods that take into consideration the correlation among the biomarkers. This last approach provides pools of biomarkers with superior diagnostic and prognostic performances. Multivariate techniques are often applied with variable selection procedures to provide the smallest set of biomarkers with the best predictive ability. The exhaustive identification is instead a valid alternative since it can provide comprehensive information about the etiology of the pathology. The most recent applications of the omics approaches to the identification of biomarkers for PC are presented, with particular regard to the statistical methods adopted.