Risk Stratification in Older Intensively Treated Patients With AML

Abstract

Purpose: AML is a genetically heterogeneous disease, particularly in older patients. In patients older than 60 years, survival rates are variable after the most important curative approach, intensive chemotherapy followed by allogeneic hematopoietic cell transplantation (allo-HCT). Thus, there is an urgent need in clinical practice for a prognostic model to identify older patients with AML who benefit from curative treatment.

Methods: We studied 1,910 intensively treated patients older than 60 years with AML and high-risk myelodysplastic syndrome (HR-MDS) from two cohorts (NCRI-AML18 and HOVON-SAKK). The median patient age was 67 years. Using a random survival forest, clinical, molecular, and cytogenetic variables were evaluated in an AML development cohort (n = 1,204) for association with overall survival (OS). Relative weights of selected variables determined the prognostic model, which was validated in AML (n = 491) and HR-MDS cohorts (n = 215).

Results: The complete cohort had a high frequency of poor-risk features, including 2022 European LeukemiaNet adverse-risk (57.3%), mutated TP53 (14.4%), and myelodysplasia-related genetic features (65.1%). Nine variables were used to construct four groups with highly distinct 4-year OS in the (1) AML development, (2) AML validation, and (3) HR-MDS test cohorts ([1] favorable: 54% ± 4%, intermediate: 38% ± 2%, poor: 21% ± 2%, very poor: 4% ± 1%; [2] 54% ± 9%, 43% ± 4%, 27% ± 4%, 4% ± 3%; and [3] 54% ± 10%, 33% ± 6%, 14% ± 5%, 0% ± 3%, respectively). This new AML60+ classification improves current prognostic classifications. Importantly, patients within the AML60+ intermediate- and very poor-risk group significantly benefited from allo-HCT, whereas the poor-risk patients showed an indication, albeit nonsignificant, for improved outcome after allo-HCT.

Conclusion: The new AML60+ classification provides prognostic information for intensively treated patients 60 years and older with AML and HR-MDS and identifies patients who benefit from intensive chemotherapy and allo-HCT.

FIG 1.

CONSORT diagram. MDS, myelodysplastic syndrome.

FIG 2.

Mutational landscape of the patient cohort. Driver events found in 1,910 elderly patients with newly diagnosed AML. Each bar represents a driver lesion, including gene mutations and chromosomal abnormalities. The colors (green, blue, red) in each bar represent the ELN 2022 risk class (favorable, intermediate, adverse) among each driver lesion. abn, abnormality in; ck, complex karyotype (annotated according to the 2022 ELN risk classification); ELN, European LeukemiaNet; mk, monosomal karyotype.

FIG 3.

AML60+ classification model development. (A) Cox regression analysis for OS using the top predictors identified by the random survival forest. Patients are assigned points on the basis of the presence of predictors; (B) shows the distribution of patients across the total sum of points and four risk groups in the development cohort. OS by risk categories of the novel risk score in the (C) AML development cohort, (D) AML validation cohort, and (E) HR-MDS test cohort. Time is measured from the date of study entry. F, number of failures (ie, death); HR, hazard ratio; HR-MDS, high-risk myelodysplastic syndrome; N, number of patients at risk; OS, overall survival.

FIG 4.

Restratification and OS of patients from ELN 2022 to AML60+ classification. (A) Restratification of patients from ELN 2022 risk classification to the new AML60+ classification. OS by the new AML60+ classification in the ELN 2022 (B) favorable-, (C) intermediate-, and (D) adverse-risk groups. The corresponding outcome curve of each ELN 2022 group is shown in gray in (B), (C), and (D). Time is measured from the date of study entry. ELN, European LeukemiaNet; F, number of failures (ie, death); N, number of patients at risk; OS, overall survival.

FIG 5.

OS of allo-HCT versus no allo-HCT in first CRi in AML60+ risk groups. OS by the (A) favorable, (B) intermediate, (C) poor, and very poor (D) groups of the AML60+ classification in which allo-HCT is considered as a time-dependent covariate according to a Simon-Makuch plot. Time is measured from the date of CRI, and patients switch from the no allo-HCT to the allo-HCT at the time of allo-HCT if they received allo-HCT. allo-HCT, allogeneic hematopoietic cell transplantation; CRi, complete remission with incomplete count recovery; F, number of failures (ie, death); N, number of patients at risk; OS, overall survival.