Gene- and subtype-dependent prognostic impact of ras pathway mutations in acute myeloid leukemia: a cohort study of 2,500 patients

Blood Cancer Journal CORRESPONDENCE www.nature.com/bcj OPEN Gene- and subtype-dependent prognostic impact of ras pathway mutations in acute myeloid leukemia: a cohort study of 2,500 patients © The Author(s) 2026 1234567890();,: Blood Cancer Journal (2026)16:82 ; https://doi.org/ 10.1038/s41408-026-01532-9 To the Editor: Acute myeloid leukemia (AML) is a hematologic malignancy driven by diverse molecular aberrations [1, 2]. Constitutive activation of signal transduction pathways is a hallmark of leukemogenesis [1, 3]. Among these, the RAS/MAPK signaling pathway is a central regulator of critical cellular processes, including proliferation, differentiation, and apoptosis [4]. RAS pathway mutations occur in approximately 25–30% of AML patients, but their prognostic significance remains controversial [5–7]. In AML, excessive activation of this pathway is predominantly driven by somatic mutations in RAS family members (NRAS and KRAS) or their upstream regulators and downstream effectors, such as PTPN11, CBL, and NF1 [8, 9]. Most prior studies were limited by relatively small sample sizes or the use of outdated risk stratification criteria [10–12]. Furthermore, the clinical impact of less common effectors like NF1 and CBL remains undercharacterized. Therefore, we conducted a comprehensive analysis of NRAS, KRAS, PTPN11, CBL, and NF1 mutations in a large cohort of 2500 AML patients based on the latest European LeukemiaNet AML risk classification guidelines (ELN 2022). Between January 2015 and July 2025, a total of 2917 consecutive patients diagnosed with AML were screened at the Institute of Hematology and Blood Diseases Hospital of the Chinese Academy of Medical Sciences. After excluding patients aged > 65 years or those who did not receive intensive induction therapy, 2500 patients were included in the final analysis. The targeted next-generation sequencing (NGS) panels used at our center evolved over time: a 114-gene panel was used from 2015 to 2017, followed by a 175-gene panel (2018–2020), a 267-gene panel (2021–2023), and a 365-gene panel (2023–present). Detailed methods, treatment protocols, and statistical analyses are provided in the supplemental materials. The cohort (n = 2500) comprised 1,311 (52.4%) males and 1189 (47.6%) females, with a median age of 41.0 years (interquartile range [IQR], 31.0–51.0 years). A total of 899 (36.0%) patients harbored at least one RAS pathway mutation (RASmut). Patients were classified as wild-type (WT, n = 1601), co-mutation group (≥ 2 RAS pathway mutations, n = 253), NRAS-mutated (n = 400), KRAS-mutated (n = 106), PTPN11-mutated (n = 77), CBL-mutated (n = 35), or NF1-mutated (n = 28). Patients with NRAS, KRAS, and PTPN11 mutations had significantly higher white blood cell counts at diagnosis than WT patients, consistent with the proliferative role of RAS/MAPK signaling. ELN 2022 risk distribution differed by mutation: KRAS and NF1 mutations were enriched in adverse-risk AML, whereas NRAS mutations were more frequent in favorable- risk AML. Detailed baseline characteristics are presented in Table 1. Among RASmut patients, the most frequent alterations were FLT3 (27%), NPM1 (19%), KIT (14%), DNMT3A (13%), and WT1 (10%, Fig. S1A). Co-mutation analysis demonstrated that NRAS mutations were mutually exclusive with other RAS pathway mutations (all p < 0.001, Fig. S1B), suggesting functional redundancy within the RAS pathway. PTPN11 mutations were associated with NPM1 and DNMT3A mutations, whereas KRAS mutations were mutually exclusive with NPM1. Co-mutation analysis results are shown in Fig. S1B. In our cohort, 44 (1.8%) patients harbored RAS pathway mutations without recurrent co-mutations. No significant differences in OS or EFS were observed between patients with isolated and non-isolated RAS pathway mutations (5-year OS: 55.3% [95% CI, 40.8–75.1%] vs. 63.4% [95% CI, 59.4–67.8%], p = 0.400; 5-year EFS: 53.8% [95% CI, 38.5–75.3%] vs. 51.1% [95% CI, 46.6–56.0%], p = 0.950, Fig. S2A, B). Notably, all 44 patients with isolated RAS pathway mutations were classified as intermediate-risk according to the 2022 ELN guidelines. There were no significant differences in OS or EFS between patients with isolated and non-isolated RAS pathway mutations (5-year OS: 55.3% [95% CI, 40.8–75.1%] vs. 51.9% [95% CI, 42.9–62.9%], p = 0.690; 5-year EFS: 53.8% [95% CI, 38.5–75.3%] vs. 38.4% [95% CI, 29.2–50.5%], p = 0.310, Fig. S2C, D). Then RAS pathway mutations were considered as a whole. There was no significant difference in OS or EFS between the RASmut and WT group (5-year OS: 62.9% [95% CI, 58.9–67.1%] vs. 58.8% [95% CI, 55.6–62.1%], p = 0.670; 5-year EFS: 51.4% [95% CI, 47.1–56.2%] vs. 48.5% [95% CI, 45.5–51.8%], p = 0.510; Fig. 1A, B). When patients were stratified by the 2022 ELN guidelines, RAS pathway mutations were associated with superior OS (5-year OS: 79.0% [95% CI, 74.2–84.1%] vs. 69.7% [95% CI, 65.5–74.2%], p = 0.036; Fig. S3A) in the favorable-risk subgroup, whereas no significant difference was observed in EFS (5-year EFS: 64.1% [95% CI, 58.1–70.8%] vs. 59.3% [95% CI, 55.1–63.9%], p = 0.120; Fig. S3B). No significant associations with either OS or EFS were observed in the intermediate- or adverse-risk subgroups (Fig. S3C–F). To identify which molecular subtype drove this effect, we performed subtype analyses. We found that the superior OS and EFS associated with RAS mutations were restricted to patients harboring the RUNX1::RUNX1T1 fusion (5-year OS: 84.0% [95% CI, 75.8–93.1%] vs. 66.0% [95% CI, 59.6–73.1%], p = 0.016; 5-year EFS: 67.4% [95% CI, 55.8–81.4%] vs. 57.0% [95% CI, 50.8–64.0%], p = 0.039; Fig. S4A, B). No such effect was observed in CBFB::MYH11, NPM1-mutated, or CEBPA bZIP in-frame mutated (CEBPAbZIP-inf) AML (Fig. S4C–H). We further evaluated the prognostic impact of individual RAS pathway mutations. NRAS mutations were associated with favorable outcomes, showing significantly superior OS and EFS (5-year OS: 67.3% [95% CI, 62.7–72.2%] vs. 57.8% [95% CI, Received: 2 February 2026 Revised: 18 April 2026 Accepted: 22 May 2026 12.2 (3.71, 38.6) 84.0 (71.0, 100.0) 41.0 (31.0, 51.0) 14.4 (4.3, 42.4) 83.0 (70.0, 99.0) 41.0 (25.0, 74.0) Age at diagnosis (median, IQR) WBC, ×109/L (median, IQR) Hb, g/L (median, IQR) Plt, ×109/L (median, IQR) 469 (22.8) 192 (9.3) 204 (9.9) 39 (1.9) 14 (0.7) 17 (0.8) 10 (0.5) AML with KMT2A rearrangement AML with CBFB::MYH11 fusion AML with NUP98 rearrangement AML with DEK::NUP214 fusion AML with MECOM rearrangement AML with BCR::ABL1 fusion 438 (21.3) AML with RUNX1::RUNX1T1 fusion AML with NPM1 mutation 327 (15.9) AML with CEBPAbzip-inf mutation WHO 2022(n, %) 86 (3.4) ASXL1 CSF3R 118 (4.7) IDH1 91 (3.6) 125 (5.0) TET2 U2AF1 161 (6.4) IDH2 112 (4.5) 178 (7.1) WT1 103 (4.1) 225 (9.0) DNMT3A RUNX1 313 (12.5) KIT GATA2 681 (27.2) 335 (13.4) FLT3 Mutation (n, %) 42.0 (31.0, 51.0) 1189 (47.6) Female Blood Cancer Journal (2026)16:82 8 (0.6) 5 (0.4) 10 (0.8) (...truncated)