Data analysis


Patients subpopulation for the analysis is formed by the eligible patients with surgical evaluation and availability of plasma sample for the requested RNA analyses.
Baseline characteristics will be described for overall sample population by means of standard summary statistics (absolute frequencies, mean median and extreme values for continous data, percentage for categorical data).
The association of variation between preneoadjuvant and postneoadjuvant expression levels of miRNA with pCR will be presented with contingency tables and analysed by mean of a logistic model. the odds ratio for the association and the AUC will be calculated, together with the corresponding 95% confidence intervals.
For secondary objective a) and c) the role of liquid biopsies in the monitoring of treatment will be investigated by using a semi-parametric survival model (Cox model) with time dependent variables, in order to incorporate modifications in the plasma measurements over time and their association with outcome, while for objective b) the same analysis will be applyed only on the subgroup of patients achieving pCR.