MLL (KMT2A) rearrangements are among the most frequent chromosomal abnormalities that occur in acute myeloid leukemia (AML). Mutational landscapes in KMT2A-rearranged AML have been reported; however, most studies are missing data at relapse. Therefore, matched diagnostic and relapse samples were analyzed in this study, and the clonal evolution pattern in KMT2A-rearranged AML was examined. Further, the prognostic significance of the clonal architecture was investigated.Sixty-two diagnostic and 16 relapse samples obtained from pediatric patients with KMT2A-rearranged AML enrolled in the Japan Children's Cancer Group (JCCG) AML-05/AML-99 study were analyzed for 338 genes using targeted sequencing. The data were analyzed with the published data of 105 diagnostic and 9 relapse samples with KMT2A-rearranged AML. Additionally, as a control, the mutation data of matched diagnostic and relapse samples of 107 patients with non-KMT2A-rearranged AML were collected.Among 25 patients with KMT2A-rearranged AML with matched data at diagnosis and relapse, mutations of signaling pathway genes (FLT3, KRAS, NRAS, PTPN11, CBL, and BRAF) were frequently detected in diagnostic samples (25 mutations/25 patients). However, 21 of 25 (84.0%) mutations were lost at relapse. In contrast, 7 of 19 (36.8%) mutations of other pathway genes were lost at relapse, and the percentage was significantly lower than that of mutations in the signaling pathway genes (P = 0.002). Six mutations in the signaling pathway genes and 11 mutations in other pathway genes were acquired at relapse. Particularly, mutations of transcription factor genes (WT1, SPI1, GATA2, and RUNX1) were acquired at relapse: 7 of 8 (87.5%) mutations were detected only at relapse. These results suggest that mutations of signaling pathway genes are unstable in the clonal evolution of KMT2A-rearranged AML. Mutations of other pathway genes, especially those of transcription factor genes, may contribute to relapse in patients with KMT2A-rearranged AML.Next, attention was turned to the KRAS mutations (KRAS-MT) because we have previously shown that KRAS-MT are independent adverse prognostic factors in KMT2A-rearranged AML (Blood Adv. 2020). Among 25 patients with KMT2A-rearranged AML with matched data at diagnosis and relapse, 10 (40.0%) patients harbored KRAS-MT at diagnosis. Interestingly, KRAS-MT were lost at relapse in 9 of 10 (90.0%) patients. Among 107 patients with non-KMT2A-rearranged AML with matched data at diagnosis and relapse, 10 (9.3%) patients harbored KRAS-MT at diagnosis. The frequency of KRAS-MT was significantly higher in KMT2A-rearranged AML (40.0% vs. 9.3%, P = 0.0006). This may be explained on the basis of the fact that KRAS-MT is associated with a high relapse rate in KMT2A-rearranged AML, but not in non-KMT2A-rearranged AML. KRAS-MT was lost at relapse in 5 of 10 (50.0%) patients with non-KMT2A-rearranged AML. The percentage of KRAS-MT loss at relapse was higher in KMT2A-rearranged AML. However, it was not statistically significant (90.0% vs. 50.0%, P = 0.14). Therefore, KRAS-MT may be unstable in clonal evolution regardless of disease subtypes in AML. The underlying mechanisms of the paradox between the high relapse rate in patients with KRAS-MT and frequent loss of KRAS-MT at relapse in patients with KMT2A-rearranged AML should be examined in future studies.The loss of KRAS-MT at relapse suggests that the mutations were in subclones at diagnosis. Therefore, we finally examined the prognosis of 167 patients according to the clonality of KRAS-MT at diagnosis. In patients with KMT2A-MLLT3 (n = 67), those with subclonal KRAS-MT (n = 6) had adverse 5-y event-free survival compared with both patients with wild-type KRAS (KRAS-WT) (n = 56) (KRAS-WT vs. subclonal KRAS-MT: 58.7% vs. 16.7%, P = 0.04) and patients with clonal KRAS-MT (n = 5) (clonal KRAS-MT vs. subclonal KRAS-MT: 80.0% vs. 16.7%, P = 0.07). However, 5-y overall survival (OS) was similar among the three groups. In contrast, among patients with KMT2A-MLLT10 (n = 37), those with clonal KRAS-MT (n = 5) had adverse 5-y OS compared with both patients with KRAS-WT (n = 20) (KRAS-WT vs. clonal KRAS-MT: 59.7% vs. 0.0%, P = 0.006) and patients with subclonal KRAS-MT (n = 12) (subclonal KRAS-MT vs. clonal KRAS-MT: 58.3% vs. 0.0%, P = 0.04). According to these results, the effects of the clonality of KRAS-MT on prognosis may depend on which KMT2A fusion is present. DisclosuresNannya: Otsuka Pharmaceutical Co., Ltd.: Consultancy, Speakers Bureau; Astellas: Speakers Bureau. Saito: Toshiba corporation: Research Funding. Ogawa: Kan Research Laboratory, Inc.: Consultancy, Research Funding; Otsuka Pharmaceutical Co., Ltd.: Research Funding; Dainippon-Sumitomo Pharmaceutical, Inc.: Research Funding; Eisai Co., Ltd.: Research Funding; Ashahi Genomics: Current holder of individual stocks in a privately-held company; ChordiaTherapeutics, Inc.: Consultancy, Research Funding.