Residual mutations in epigenetic modifiers are associated with poor prognosis and high relapse rate in Acute Myeloid Leukemia

Abstract Background Genetic mutations play a crucial role in the development and progression of Acute Myeloid Leukemia (AML), impacting treatment efficacy and influencing a patient's prognosis. Residual mutations, often at low levels and identified through Next-generation sequencing (NGS), indicate the persistence of specific genetic abnormalities in the leukemic cells even after achieving a complete remission (CR), and can significantly increase the risk of relapse (Ikoma Y, Leukemia 2025). Patients with residual mutations could achieve CR after receiving therapy, while half of these CR patients, even with age ＜60 years, still experience relapse (Aitken MJ, J Hematol Oncol 2021). The study aimed to explore the clinical impact of residual mutations on prognosis and relapse of AML patients. Methods A total of 122 consecutive AML patients (including 115 de novo AML and 7 secondary AML) were enrolled from May 28, 2019, to May 27, 2025, at Zhongda Hospital, Institute of Hematology, Southeast University. The bone marrow samples from 97 patients were sequenced at the two time points of diagnosis and CR, and 25 patients were sequenced at the three time points of diagnosis, CR, and relapse, respectively, and a total of 265 samples were included in the analysis. The next-generation sequencing for mutation profiling was performed with our targeted exome-seq panel, which covered the whole exomes of 52 leukemia driver genes (Jun L, Blood Cancer J. 2022). Results Being compared between at CR and diagnosis, the mutation clearance rates were higher for FLT3 (22/26, 84.6%), NRAS (11/13, 84.6%), NPM1 (19/24, 79.2%), CEBPA (13/17, 76.5%), KRAS (8/11, 72.7%), and IDH2 (11/17, 64.7%); while it is significantly lower for SRSF2 (0/5, 0%), PHF6 (0/3, 0%), SF3B1 (0/3, 0%), DNMT3A (5/31, 16.1%) and TET2 (7/27, 25.9%). All detected mutations were cleared in 45.9% (56/122) patients with CR, while 54.1% (66/122) still had residual ones. The patients still harboring residual mutations after CR vs those with full clearance had a higher relapse rate (17.9% vs. 45.5%, P = 0.001), worse relapse-free survival (RFS) median RFS not reached vs. 15 (95% CI 8.42-21.58) months, P 0.001 and overall survival (OS) median OS not reached vs. 21 (95% CI 16.70-25.30) months, P 0.001. Moreover, we analyzed the functional impact of the residual mutations and found that the residual mutations are mainly from the genes in epigenetic modifiers, which include TET2, ASXL1, DNMT3A, EZH2, WT1, BCORL1, IDH1, IDH2, STAG2, and BCOR. Patients with the residual mutations had significantly worse OS median OS 21 (95% CI 13.02-28.98) months vs. not reached, P 0.001 and RFS median RFS 15 (95% CI 11.15-18.85) months vs. not reached, P =0.003 than those without the mutations. Furthermore, we delved into the clonal evolution from the point of diagnosis to relapse in 25 25-patient cohort, and four distinct patterns of dynamic genetic mutation changes were observed from diagnosis to relapse in the cohort. Pattern 1 has four patients who lost all the mutations detected at diagnosis after CR and experienced a relapse, either with or without newly acquired mutations. Pattern 2 includes 10 patients who had residual mutations at CR, expanded during relapse, but without the emergence of any additional new mutations. Pattern 3 encompasses 9 patients who had residual mutations at CR, expanded during relapse, with newly acquired mutations undetected at diagnosis. Pattern 4 involves 2 patients who had residual mutations at CR, expanded during relapse, with comeback mutations detected at diagnosis. These results showed that residual mutations appeared in the pattern 2-4, which occupied 21 of 25 relapsed patients, indicating the critical role of residual mutations in the relapse of AML. Finally, we compared the VAF between relapse and CR in the genes of different signaling pathways. Results showed that the median VAF was significantly increased at relapse than in CR for the gene mutations in TET2 (45.23% vs.5.8%, P=0.011), epigenetic modifiers (42.06% vs.7.03%, P0.001). These data further indicated that clone expansion of residual mutations in epigenetic modifiers may be responsible for the relapse. Conclusion The residual mutations were associated with adverse prognosis and relapse, and residual mutations in the genes involving epigenetic modifiers may be the critical adverse prognostic factors for RFS and OS.