P417: genomic landscape and prognosis in older acute myeloid leukemia patients not eligible for intensive chemotherapy

Background: The genomic landscape of acute myeloid leukemia (AML) has been previously mostly studied in younger patients who received intensive chemotherapy. Data in older patients receiving less intensive therapies are scarce. In addition, no genetic risk classification for older AML patients has so far been established. Aims: To characterize the genomic landscape and leukemogenic pathways of AML in older patients, and to study the clinical implications of these biological features. Methods: Targeted sequencing of 263 genes was performed in 604 patients enrolled in the randomized, multi-center phase 3 ‘ASTRAL-1’ trial (NCT02348489) evaluating the second-generation hypomethylating agent guadecitabine (SGI-110) in treatment-naïve AML pts not eligible for intensive chemotherapy in comparison to a treatment choice of decitabine, azacitidine, or low-dose cytarabine. Results: Recurrent mutations were found in ASXL1 (28%), TET2 (27%), SRSF2 (22%), DNMT3A (21%), RUNX1 (20%), TP53 (19%), NPM1 (15%) and FLT3 (13%; [internal tandem duplication (ITD) 8%, tyrosine kinase domain 6%]). DDX41mut were found in 5.5% of the patients, 61% of whom with suspected germline variant. Adverse cytogenetic features such as complex karyotype (29%), -7/del(7q) (22%), 5q/del(5q) (20%), -17/del(17p)/abn(17p) (12%) were common. Based on the 2022 ELN classification, patients were stratified into the adverse (73%), intermediate (14%) or favorable (13%) risk group. Based on the International Consensus Classification (ICC), most patients were classified as AML with myelodysplasia-related gene mutations (45%), AML with mutated TP53 (17%), AML with mutated NPM1 (16%), AML not otherwise specified (10%) and AML with myelodysplasia-related cytogenetic abnormalities (5%). A modelling algorithm yielded a stable oncogenetic tree (Figure 1A) identifying distinct leukemogenic trajectories with ASXL1, DDX41, DNMT3A, TET2 and TP53 mutations as leukemia-initiating events. ASXL1, DNMT3A, and TET2 gave rise to further clones, whereas DDX41 and TP53 terminated at the node, suggesting independence from further events that drive leukemogenesis. Supporting the ICC classification, the ASXL1 subtree contained 8 of 9 genes that define the new category of AML with myelodysplasia-related gene mutations. When assessing clinical impact, both the 2017 and 2022 ELN risk classifications failed to identify clinically meaningful prognostic groups. In Cox regression models, older age (HR 1.02 [1.01-1.04], p=.009), male sex (HR 1.32 [1.09-1.6], p=.004), ECOG score ≥2 (HR 1.55 [1.28-1.88], p<.001), higher white blood cell counts (HR 1.63 [1.34-1.97], p<.001), FLT3-ITD (HR 1.7 [1.2-2.4], p=.003), SRSF2 (HR 1.36 [1.06-1.76], p=.017), and TP53 mutations (HR 1.59 [1.24-2.05], p<.001) had an adverse impact on overall survival, whereas DDX41 mutations (HR 0.41 [0.24-0.69], p<.001) were exceptionally beneficial. To simplify the model, a subsequent backward elimination based on the Akaike information criterion led to delineation of 3 genetically defined risk groups (favorable: DDX41mut, adverse: TP53mut or FLT3-ITDpos, intermediate: all other) with predicted survival curves (Figure 1B). Summary/Conclusion: Using different modelling algorithms, our comprehensive analysis of the so far largest study in older, treatment naïve AML patients identified distinct trajectories of leukemia development, provided support for AML with mutated DDX41 as a new clinico-pathologic entity and a basis for the development of a risk stratification that may be applicable for the numerous older patients receiving less intensive therapies.Keywords: Tumorigenesis, Age, AML, Prognostic groups