In rare acute myeloid leukemia patients harboring both RUNX1 and NPM1 mutations, RUNX1 mutations are unusual in structure and present in the germline

It has become clear that there are distinct interrelationships between recurring genetic mutations in acute myeloid leukemia (AML). Whereas certain mutations frequently co-exist in AML, others are mutually exclusive. Mutations in NPM1 and RUNX1, two of the most commonly mutated genes in AML, are almost entirely mutually exclusive. In six recent studies that evaluated the frequency of both NPM1 and RUNX1 mutations in AML,1–6 these two mutations co-existed in just 0.6% (13 of 2348) of the cases. In the two studies conducted by our group1,4 which included large cohorts of patients with primary AML and a normal karyotype (n=392)4 and patients with sole trisomy 8 (n=80),1NPM1 and RUNX1 mutations coexisted in 0.8% (4 of 472) of the patients. The goal of the study reported herein was to more closely examine these few AML patients carrying both mutations in order to better understand the relationship between NPM1 and RUNX1 mutations in AML.

Clinical and molecular characteristics of the 4 patients with co-existing RUNX1 and NPM1 mutations are provided in Table 1. Patients provided written informed consent, and study protocols were in accordance with the Declaration of Helsinki. All NPM1 mutations were type A, consisting of duplication of TCTG at position c.860_863 (NCBI Accession NM_002520.5). In contrast, each of the 4 patients had a different RUNX1 mutation (numbering based on the RUNX1c isoform; NCBI Accession NM_001754.4). Patient 1 harbored c.952T>G, resulting in a missense change, p.S318A, in the first amino acid of the transactivation domain (TAD). Patient 2 harbored c.1098_1103delCGGCAT, resulting in an in-frame deletion of two amino acids, p.I366_G367del, in the TAD. Patient 3 harbored c.620G>A, resulting in a missense change, p.R207Q, just downstream of the Runt domain (RD), the domain responsible for DNA binding and heterodimerization with the beta subunit of core binding factor. Patient 4 harbored c.155T>A, resulting in a missense change, p.M52K, upstream of the RD. All RUNX1 mutations (4 of 4) in NPM1-mutated patients were in-frame and located outside of the RD (Figure 1A). In contrast, only 4.5% (4 of 88) of the RUNX1 mutations in NPM1-wild-type patients were in-frame and located outside of the RD (P<0.001).

To determine whether RUNX1 and NPM1 mutations were somatically acquired in AML patients harboring both mutations, pre-treatment germline material was screened in those RUNX1-mutated/NPM1-mutated patients for whom buccal cells were available (UPN 1-3). Surprisingly, in all 3 cases, the same RUNX1 mutation identified in leukemic blasts was also present in the germline (Figure 1B). In contrast, NPM1 mutations were not detected in the germline (Figure 1C), confirming that germline material was not contaminated with leukemic cells. To determine the specificity of this finding to NPM1-mutated cases, we next tested the frequency of germline RUNX1 mutations in NPM1-wild-type patients known to have RUNX1 mutations in their leukemic blasts. Of the RUNX1-mutated/NPM1-wild-type patients with buccal cells available (n=56), none harbored germline RUNX1 mutations (P<0.001). Thus, in our cohort, AML patients with coexisting RUNX1 and NPM1 mutations were not found to acquire both of these mutations somatically; rather, our data suggest that RUNX1 mutations pre-existed and NPM1 mutations were a later, acquired event.

Table 1.Clinical and molecular characteristics of primary AML patients with co-existing NPM1 and RUNX1 mutations.

We considered the possibility that these germline RUNX1 mutations may have been inherited and caused a predisposition to the development of AML (familial platelet disorder with associated myeloid malignancy: Online Mendelian Inheritance in Man #601399). To address this issue, we sought to identify whether any of these patients or their family members had a history of hematologic problems, including myelodysplastic syndrome and/or AML. Direct contact with patients and/or families was not possible since all patients are now deceased and had not consented to their families being contacted at the time of study enrollment. Thus, medical and family histories taken at the time of diagnosis were obtained from chart review. This limited analysis revealed that none of the germline RUNX1-mutated patients had a known history of bleeding or thrombocytopenia pre-dating their diagnosis of AML, or a family history of hematologic problems. However, since the recorded family histories included information on first-degree relatives only, and RUNX1 mutation-associated syndromic disease is variably expressed,11 this analysis was insufficient for formal evaluation of familial phenotype. To attempt to address the disease potential of the RUNX1 mutations discovered here, we subjected them to MutationTaster analysis, which is a free, web-based application for the evaluation of disease-causing potential of DNA sequence variations.12 This analysis predicted that c.952T>G and c.1098_1103delCGGCAT were likely to be harmless polymorphisms; in contrast, c.620G>A and c.155T>A were predicted to be disease-predisposing. Supporting the likelihood that c.620G>A plays a causal role in AML pathogenesis is the fact that it was somatically acquired in another patient within our cytogenetically normal AML cohort.4 Despite these predictions, the extent to which the germline RUNX1 mutations discovered here contributed to the pathogenesis of AML remains unclear. Although several previous studies have highlighted the near mutual exclusivity of RUNX1 and NPM1 mutations,1–6 to our knowledge, no prior study has closely examined the rare AML cases harboring both mutations. We found that the RUNX1 mutations in these cases are structurally unusual relative to the majority of those found in AML; namely, they are in-frame, located outside the RD, and present in the germline. The functional significance of these types of RUNX1 mutations has not been established13 and future studies should be aimed at determining whether they are innocent polymorphisms or disease-predisposing alleles. The complete exclusivity between RUNX1 mutations known to be disease-predisposing and NPM1 mutations in our AML cohort suggests that they may have redundant roles in promoting leukemogenesis. This role may be one of causing expansion of the myeloid precursor cell compartment, as recently demonstrated in mouse models expressing either RUNX114 or NPM115 mutations in primitive hematopoietic progenitor cells. In summary, our work suggests that NPM1 mutations do not co-exist with classical RUNX1 mutations (i.e. those RUNX1 mutations known to be disease-predisposing) in primary AML. In rare patients harboring both NPM1 and RUNX1 mutations, NPM1 mutations seem to be acquired in the context of pre-existing RUNX1 mutations of unclear functional significance. Thus, our work adds to the growing body of knowledge defining key mutational interrelationships underlying the pathogenesis of AML.

Figure 1.(A) Schematic diagram of the RUNX1 protein demonstrating the location of RUNX1 mutations in AML patients with co-existing RUNX1 and NPM1 mutations. (B) Sequencing chromatograms demonstrating RUNX1 mutational status in leukemic blasts and buccal cells from AML patients with co-existing RUNX1 and NPM1 mutations. (C) Sequencing chromatograms demonstrating NPM1 mutational status in leukemic blasts and buccal cells from AML patients with co-existing RUNX1 and NPM1 mutations.