KRAS mutations Frequently Coexist with High-Risk MLL Fusions and Are Independent Adverse Prognostic Factors in MLL-Rearranged Acute Myeloid Leukemia

MLL(KMT2A) rearrangements are among the most frequent chromosomal abnormalities in acute myeloid leukemia (AML). MLL fusion patterns are associated with patient prognosis; however, little is known about the distribution and prognostic significance of driver mutations according to specific MLL fusion partner genes.

In this study, we conducted sequence analysis of 338 genes in pediatric patients with MLL-rearranged (MLL-r) AML (n = 56; JPLSG AML-05 study) alongside data from TARGET cohort pediatric MLL-r AML (n = 104) and non-MLL-r AML (n = 581). In addition, mutation data and clinical information from adult MLL-r AML patients (n=81), collected at the MLL Munich Leukemia Laboratory, were also analyzed.

The frequencies of driver mutations differed according to MLL fusion partner genes in pediatric MLL-r AML patients (JPLSG AML-05 study + TARGET cohort, n = 160). FLT3 mutations or internal tandem duplications were more frequent in patients with MLL-MLLT3 (27/63, 42.9%) and MLL-MLLT1 (4/11, 36.4%). RAS pathway genes were more frequently mutated in specific groups, such as KRAS in those with MLL-MLLT10 (17/37, 45.9%), MLL-MLLT4(AFDN) (5/12, 41.7%), MLL-MLLT1 (3/11, 27.3%), and other (4/14, 28.6%) fusions, and NRAS in those with MLL-ELL (10/23, 43.5%) and MLL-MLLT4 (4/12, 33.3%) fusions. Other pathway mutations also coexisted with specific partner genes; SETD2 mutations were frequent in patients with MLL-MLLT4 fusions (4/12, 33.3%), and STAG2 mutations were frequent in those with MLL-ELL (6/23, 26.1%).

Next, we examined the prognostic significance of each driver mutation. In pediatric MLL-r AML patients (n = 160), among eight frequently mutated genes (≥ 5%: ≥ 8 patients with mutations) and one copy number change (trisomy 8), only KRAS mutations were significantly associated with adverse prognoses, in terms of both event-free survival (EFS) and overall survival (OS). Compared with patients without KRAS mutations (n = 118, KRAS-WT), those with KRAS mutations (n = 42, KRAS-MT) had significantly inferior prognoses (KRAS-WT vs. KRAS-MT; 5 year (5y)-EFS: 51.8% vs 18.3%, P < 0.0001; 5y-OS: 67.3% vs 44.3%, P = 0.003). The adverse prognostic impact of KRAS mutation was confirmed in adult MLL-r AML. By contrast, there were no significant differences in EFS or OS (KRAS-WT vs. KRAS-MT; 5y-EFS: 50.6% vs 54.5%, P = 0.55; 5y-OS: 67.1% vs 64.9%, P = 0.84) between patients with KRAS-WT (n = 533) and KRAS-MT (n = 48) non-MLL-r AML.

We also analyzed the prognostic significance of KRAS mutations according to MLL fusion partner. First, the prognoses of patients were compared for each fusion partner gene. Consistent with a previous report, patients with MLL-MLLT10 (n = 37), MLL-MLLT4 (n = 12), and MLL-MLLT1 (n = 11) had poor prognoses compared to those with other fusion types. Therefore, we dichotomized patients into high- (MLLT10 + MLLT4 + MLLT1, n = 60) and intermediate/low- (MLLT3 + ELL + Others, n = 100) risk groups. The frequency of KRAS mutation was significantly elevated in the high-risk group (26/60, 43.3%) relative to the intermediate/low-risk group (16/100, 16.0%; P = 0.0002). Further, in the high-risk group, KRAS-MT patients (n = 26) had significantly adverse prognoses compared with KRAS-WT patients (n = 34; KRAS-WT vs. KRAS-MT; 5y-EFS: 31.5% vs 10.3%, P = 0.007; 5y-OS: 52.4% vs 40.5%, P = 0.16). Moreover, KRAS-MT patients in the intermediate/low-risk group (n = 16) had significantly adverse prognoses compared with KRAS-WT patients (n = 84; KRAS-WT vs. KRAS-MT; 5y-EFS: 60.3% vs 31.3%, P = 0.02; 5y-OS: 73.4% vs 50.0%, P = 0.04).

Finally, we investigated whether KRAS mutation is an independent prognostic factor in patients with pediatric MLL-r AML. Multivariate Cox regression analysis was performed including the following variables: age, WBC, MLL fusion gene, driver mutations, and trisomy 8. The results indicated that KRAS mutation was the only prognostic factor predicting both poor EFS (hazard ratio (HR), 2.21; 95% confidence interval (CI), 1.35-3.59; P = 0.002) and poor OS (HR, 1.85; 95% CI, 1.01-3.31; P = 0.045), whereas there were no statistically significant findings for MLL fusions.

These results suggest that mutations in MLL-rearranged AML are associated with MLL fusion partners, and KRAS mutations frequently coexist with high-risk MLL fusions. KRAS mutations are distinct adverse prognostic factors in MLL-r AML, regardless of risk subgroup, therefore, potentially useful for accurate treatment stratification.