Development and validation of the HAS. (A) Heat map depicting the protein abundance of various hematopoietic lineage markers across the 8 proteomic subtypes. (B) Proteins included in the HAS. The final score is calculated as the weighted sum of abundance of the 19 proteins. Proteins in red and blue are predictive for a poor and good prognosis, respectively. Coefficients of these proteins in univariate Cox regression analysis are denoted. (C) Heat map showing the established HAS and the abundance of the 19 proteins, with patients arranged according to the magnitude of their scores. Age and prognosis of patients are annotated above the heat map. (D) Forest plot of multivariate Cox regression analysis for OS in patients with AML. Hazard ratios (HRs) and 95% CIs are listed next to variables. The dotted vertical line indicates an HR of 1. Red boxes denote the HR of each variable, and horizontal lines show 95% CIs. P values indicating the significance are shown on the right. (E) Heat map of markers incorporated into the HAS in an independent external data set. (F) The prognostic value of the HAS in univariate Cox regression in an independent external data set. (G) Kaplan-Meier curves for OS (left panel) and EFS (right panel) stratified by the HAS in an external data set, with the most obvious prognostic discrimination used as the cutoff value.