4.7. Statistical Analysis

Statistical analysis was carried out with nonparametric tests such as the Kruskal–Wallis and Mann–Whitney tests to evaluate differences among the groups of study in all exosome isolation methods.

Receiver operating characteristic (ROC) curve analysis of the RQ values was used to distinguish the samples showing malignancy of prostate tumour. Accuracy was measured as the area under the ROC curve (AUC) [25]. The threshold value was determined by Youden’s index, calculated as sensitivity plus specificity-1 [26]. A multivariate binary logistic regression analysis was used to assess whether previously described miRNA combined models [8] were useful as PCa biomarkers from sEVs obtained from the different EV isolation methods. An enter method was used to include in the equation all the variables of each nested model. The binary logistic regression model provides the following estimation of the logit function:

where p = P (presence of prostate cancer), Logit (p) = log(p/(1 − p)) = log(odds), B = logOR and Xn = the expression values of the miRNAs. Therefore, if we use this estimated model as a prediction model, with the standard classification cutoff of 0.5, we would classify individuals with a positive Logit function estimation as “positive for PCa” and individuals with negative Logit function estimation as “negative for PCa”.

Only p-values ≤ 0.05 were considered significant. All data analyses were performed using SPSS software, version 15 (SPSS Inc, Chicago, USA).