Prognostic analysis 1. Forest plots from univariate and multivariate prognostic analysis for OS (A-B) and DFS (C-D), including major risk factors, MRD results, and risk-oriented therapy. The multivariate model analysis was performed on data from 149 patients with no missing values. All covariates were evaluated in univariate models and all relevant variables with univariate association within P < .15 were considered in the multivariate models. To compare the prognostic ability of multivariate models with the contribution of each variable, the Akaike information criterion was used to compare the models’ goodness of fit with the data. The final model includes the actual therapy received and not ITT therapy. The collinearity between treatment received and risk classification was evaluated using an interaction term into the multivariate model (resulting nonsignificant). Cyto, cytogenetics/genetics; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio.