Abstract
Background: Although European Leukemia Net (ELN) 2022 risk classification is a standard prognostic model in adult acute myeloid leukemia (AML) patients, utility of this classification specifically for allogeneic hematopoietic stem cell transplant (allo-HSCT) recipients remains unclear. Furthermore, given the high incidence of non-relapse mortality (NRM) after allo-HSCT, identifying populations with low survival benefit from allo-HSCT is a clinically important issue.
Patients and methods: The 198 adult AML patients, who underwent their first allo-HSCT at our institution between April 2017 and February 2025, and were classifiable according to ELN 2022 risk classification based on cytogenetic and genetic data at diagnosis, were included in this retrospective study. Gene mutation data, obtained by target sequencing for 70 genes recurrently mutated in myeloid malignancies, was available for the 134 patients (the sequenced cohort). Overall survival (OS) was measured from transplant to death. Fisher's exact test and Kruskal-Wallis test were used to compare binary and continuous variables, respectively. OS was estimated with the Kaplan-Meier method and compared using the log-rank test. The cumulative incidences (CI) of relapse and NRM were evaluated with Gray's test, with the other considered as a competing risk. Uni- and multi-variate analysis for OS were performed using the Cox proportional hazard model. Genetic abnormalities with a p < 0.20 in univariate analysis were selected as the candidates for a novel prognostic model. The level of statistical significance was set at p < 0.05.
Results: Of the 198 patients in the entire cohort, 81 (41%) were female, and the median age at transplant was 52 years (range, 16–71). The median follow-up period of survivors was 771 days (range, 71–2853 days). According to ELN 2022 risk classification, 41 (21%), 49 (25%), and 108 patients (55%) were categorized into the Favorable (Fav), Intermediate (Int), and Adverse (Adv) group, respectively. There was a significant difference among the 3 groups in age (median: 43 vs. 48 vs. 58 years; p < 0.01) and disease status at transplant (complete remission [CR]: 88% vs. 78% vs. 60%; p < 0.01). The OS rates differed significantly among the 3 groups (73% vs. 58% vs. 44% at 2 years after transplant; p = 0.01). The CI of NRM were similar across groups (11% vs. 14% vs. 19%; p = 0.28), whereas the CI of relapse showed a non-significant trend toward difference (32% vs. 41% vs. 53%; p = 0.05). Multivariate analysis for OS identified ELN 2022 risk classification (Fav vs. Adv: hazard ratio [HR] = 2.3; p = 0.023) as an independent risk factor, besides disease status (CR vs. non-CR: HR = 3.2; p < 0.01). When stratified with disease status at transplant, the OS rates were significantly different among patients in CR (79% vs. 65% vs. 53%; p = 0.014), but not among non-CR (not available vs. 30% vs. 29%; p = 0.94).
To further refine this risk classification, the impact of individual gene abnormalities on OS were explored in the sequenced cohort. Of the 134 patients, 26 (19%), 42 (31%), and 66 patients (50%) were categorized into the Fav, Int, and Adv group, respectively. In univariate analysis for OS, TP53 (HR = 4.3, p < 0.01), EZH2 (HR = 2.1, p = 0.12), and RUNX1 (HR = 1.9, p = 0.12) were identified as candidates incorporating into a novel prognostic model. In the Adv group, patients with mutations in any of these 3 genes (Adv-TER: n = 35) showed significantly worse OS than those without these mutations (Adv-others: n = 31) (10% vs. 76%; p < 0.01). Notably, the OS rate in the Adv-others group was similar to the Fav group, and OS rates were clearly stratified among these 4 groups (75% vs. 42% vs. 76% vs. 10% in Fav, Int, Adv-others, and Adv-TER, respectively; p < 0.01). The prognostic predictive ability of this novel model was confirmed even when limited to patients in CR (n = 93) (85% vs. 66% vs. 81% vs. 10%; p < 0.01). In multivariate analysis for OS among patients in CR, Adv-TER risk was identified as a sole independent risk factor (HR = 10.24; p < 0.01).
Conclusion: These findings suggest that ELN 2022 risk classification is applicable to adult AML patients undergoing allo-HSCT, and mutation profiles of TP53, EZH2, and RUNX1 further stratify patients in Adv risk into two groups with significantly different prognoses. Incorporating these mutation profiles into ELN 2022 framework might contribute to further optimization of transplant eligibility assessment.
