S138: validation of the revised 2022 european leukemianet (eln) risk stratification in adult patients with acute myeloid leukemia

Background: In 2022, the ELN risk stratification for patients (pts) with acute myeloid leukemia (AML) has been updated. The key changes of the revised ELN recommendations are as follows: i) bZIP in-frame CEBPA mutations are now considered favorable-risk (ELN22fav), irrespective of monoallelic/biallelic mutation status; ii) pts with FLT3-ITD are now considered intermediate-risk (ELN22int), irrespective of allelic ratio or NPM1 status; and iii) pts with alterations in myelodysplasia-related (MR) genes (BCOR, EZH2, SF3B1, SRSF2, STAG2, U2AF1 and ZRSR2) are categorized as adverse-risk (ELN22adv) AML. Aims: We aimed to validate the prognostic value of the ELN22 recommendations in a large, intensively treated patient cohort. Methods: We evaluated 1,570 pts aged ≥ 18 years with newly diagnosed AML intensively treated on Study Alliance Leukemia (SAL) protocols (AML96, AML60+, AML2003 or SORAML trial, respectively). Pretreatment genetic alterations were assessed using G banding, FISH, PCR, fragment length analysis and targeted next generation sequencing. Results: Pts’ median age was 55.5 years (range, 18−89 years), 47.3% were female. Most pts (84.1%) were diagnosed with de-novo AML. ELN22 risk was favorable-, intermediate-, and adverse-risk in 575 (36.6%), 410 (26.1%), and 585 (37.3 %) pts, respectively. As compared to the ELN17 classification, risk group allocation was revised in 340 pts (21.7%). Most pts were re-classified into ELN22int (45.6%, mainly due to FLT3-ITD) or ELN22adv risk group (32.9%, mostly because of evidence of MR gene alterations), respectively. However, there were only slight net effects on the percentage of pts classified as favorable-, intermediate- and adverse-risk (-0.7%, -3.4%, and +4.1%, respectively) (Fig. 1A). The CR rates in pts assigned to the ELN22fav, ELN22int and ELN22adv-risk groups were 87.3%, 76.6% and 49.2%, as compared to 86.4%, 70.5% and 53.2% in pts stratified according ELN17, respectively. The allocation of pts according to the ELN22 risk categories resulted in a significantly distinct EFS, RFS and OS. The 5-y OS for pts in the ELN22fav, ELN22int and ELN22adv-risk category was 53%, 32%, and 13%, while the corresponding 5-y OS in pts classified according ELN17 was 49%, 30% and 16%, respectively; indicating a more accurate discrimination of favorable- and adverse-risk pts in ELN22 (Fig. 1B). Within the ELN22fav-risk group, pts with CBF-AML or CEBPA bZIP in-frame mutations showed superior OS as compared to patients with NPM1 mutations (5-y OS 59% and 58% vs. 41%). Within the ELN22adv-risk group, we observed marked survival differences across mutational groups with TP53-mutated pts having the worst prognosis (5-y OS 3.3%), followed by pts with ASXL1 and/or RUNX1 mutations (5-y OS 14%), and pts with MR gene mutations (5-y OS 22%) (Fig 1C). In pts harboring MR gene mutations, EHZ2, STAG2 and ZRSR2 mutated pts showed an intermediate-like OS (5-y OS ~30%), while patients with U2AF1 mutations had significantly worse OS (5-y OS 4.6%). As compared to ELN17, reclassification according to the ELN22 recommendations resulted in an improved prognostic discrimination for OS (2-y AUC 0.67 vs. 0.70, p=.003), as evaluated by inverse-probability-of-censoring weighting area under time-dependent receiver-operating-characteristic curves. Summary/Conclusion: In conclusion, the novel ELN22 risk stratification improves prognostic discrimination in a large cohort of intensively treated AML pts. Given the heterogeneous outcome in pts with MR gene mutations, ranging between those of intermediate- and adverse-risk pts, we suggest reevaluation of risk allocation in these patients.Keywords: AML