Background: Prognosis and relapse risk in acute myeloid leukemia (AML) depend on factors like age, comorbidities, and notably cytogenetic and molecular alterations which are incorporated into ELN2022 risk classification. High-relapse-risk patients often undergo allogeneic hematopoietic stem cell transplantation (allo-HSCT), however post-transplant relapse remains the leading cause of mortality. Variant allele frequency (VAF) is an emerging parameter in AML that reflects the proportion of cells carrying specific mutations. Although VAF is not yet an established prognostic marker nor integrated into standard risk scores, it may provide additional information to optimize risk stratification and patient classification after allo-HSCT. This study explores the potential utility of VAF for improving relapse risk assessment in AML patients.

Methods: This retrospective, multicenter study by the PETHEMA working group analyzed data from the PETHEMA AML Registry, including adults (≥18 years) with AML in first complete remission who underwent allogeneic HSCT between November 2017 and February 2024. Only patients with complete molecular and outcome data were included, following ethical approval and consent. All received standard induction and consolidation therapy. Cytogenetic analyses were local, while mutational profiles were assessed via next-generation sequencing at seven Spanish reference labs using harmonized AML gene panels. Primary endpoint was relapse-free survival (RFS) and we explored the optimal VAF cut-off point for each gene using 2-year RFS. An aggregate category of myelodysplasia-related genes (ASXL1, BCOR, EZH2, RUNX1, SF3B1, SRSF2, STAG2, U2AF1, ZRSR2) was defined, for which an optimal VAF cutoff was also identified, considering the maximum VAF of each gene.

Results: A total of 717 adult patients AML were analyzed, with a median age of 56.5 years; 53.8% were male, 21.7% had secondary AML. According to the ELN 2022 classification, 14.1% had favorable risk, 31.2% intermediate risk, and 54.7% adverse risk. The most frequent mutations were DNMT3A (27.2%), FLT3-ITD (24.3%), and NPM1 (22.5%).

Mutations in FLT3-ITD, NPM1, and IDH2 were associated with significantly improved OS (overall survival) and RFS. Specifically, these mutations demonstrated lower hazard ratios and many cases have not reached median survival times, reflecting better outcomes compared to wild-type. In contrast, TP53 mutations confer a poor prognosis, with markedly shorter median OS and RFS and higher risk of relapse and death.

The highest median VAF among the studied genes were MPL (50%), CSF3R (47.6%) and CALR (46.7%) while the lowest median VAF were FLT3-TKD (7.1%), PTPN11 (11.9%) and NRAS (15.2%). Using the previously calculated cutoff significant RFS differences were found in DNMT3A (cutoff 49.7%, median NR vs. 12.1 months; HR 2.84 with 95% CI 1.41-5.72, P=0.0022), FLT3-TKD (cutoff 40.6%, median NR vs. 11.0 months; HR 6.83 with 95% CI 1.48-31.53, P=0.0045), U2AF1 (cutoff 43.7%, median NR vs. 18.0 months; HR 6.08 with 95% CI 1.22-30.4, P=0.0022), and WT1 (cutoff 47.5%, median NR vs. 13.6 months; HR 3.2 with 95% CI 1.25-8.19, P=0.011). No statistically significant differences were observed in OS using these cutoff points.

Regarding the category of myelodysplasia-related genes, using a VAF cutoff of 45.1%, significant differences were found in RFS (median not reached, HR 1.61 [95% CI 1.07-2.44], p=0.022); no significant differences were observed in OS using this cutoff.

Conclusion: Our findings indicate that not only the dichotomous presence or absence of mutations—such as FLT3-ITD—but also the VAF of these mutations plays a crucial role in determining transplant outcomes. Patients with higher or lower VAFs may experience different prognoses within the same mutational category. These results underscore the clinical value of comprehensive molecular profiling that includes VAF and highlight the potential of refined scoring models incorporating both mutation status and VAF to optimize pre- and post-transplant risk assessment and management in adult AML.

This content is only available as a PDF.
2025
Sign in via your Institution