Differential Impact of Domain Stratified Non-R882 DNMT3A Mutations in Myeloid Neoplasms

Introduction:

DNMT3A mutations are present in approximately 20% of acute myeloid leukemia (AML) patients and other myeloid neoplasms (MN) and are considered the earliest clonal event in leukemogenesis. In AML, 50-60% of DNMT3A mutations are heterozygous missense mutations affecting codon R882, which were shown to be dominant-negative in multiple studies. In contrast, there is still a complete lack of data related to the biological impact of a substantial number of widely dispersed non-R882 mutations and their clinical implications. It is possible that mutations in different DNMT3A domains are unique, leading to specific molecular characteristics and clinical phenotypes. Domain-based characterization and phenotypic effects of non-R882 mutations in more granularity is needed to understand the significance of DNMT3A mutations in MN.

Methods:

We conducted a large retrospective analysis of 16,565 MN patients by screening publicly available databases of AACR GENIE (v 13.1, n=4167), cBioPortal (Cerami et al., 2012, n=3669) along with Karmanos Cancer Institute (n=800) and a previously published metanalytic cohort (Awada et al., Blood 2021 and Kewan et al., Nature Communications, 2023). Clinical characteristics and genomic data were extracted for DNMT3A ^MT patients to study associated mutational signatures with respect to various biological characteristics. Chi square test was used for analysis of various parameters and KM curves were used to estimate overall survival (OS).

Results:

DNMT3A^MT was found in 17.5% (2903/16,565) of patients in the whole cohort. Of these, 1635 patients had one or more DNMT3A non-R882 mutations, with pAML in 920 (56.3%), sAML in 355 (21.7%), MDS in 259 (15.8%), MPN in 68 (4.2%) and MDS/MPN in 33(2.0%) patients.

Among 2088 non-R882 mutations, the most common type is missense (59%, n=1232), followed by frameshift (16.5%, n=345) and nonsense (12.3%, n=256) mutations. 78 (5%) patients had mutations in the N-terminal domain (NTD), 231 (14.1%) in Pro-Trp-Trp-Pro (PWWP) domain, 311 (19%) in ATRX-DNMT3-DNMT3L (ADD) domain and 777 (47.5%) in the methyltransferase (MTase) domain (p=0.0001). Patients with more than one domain or with splice site mutations were excluded. The median age of patients was 67.6 (59-74.8) ys. and male:female ratio showed no difference across domains (p=0.73 & p=0.64 resp.).

Patients with NTD ^MT were less frequently pAML when compared to the other three domains [38.5% vs. 56%, p=0.002]; however, there was no domain distribution pattern observed for sAML and MDS patients. NTD was enriched with abnormal cytogenetics [58% vs 37%, p=0.0034]. NTD ^MT patients had only numerically lower BM blast percentage, WBC, and platelet counts.

In domain-specific analysis, NTD ^MT had worse OS than non-NTD ^MT in pAML [8.2 vs. 15 mo., p=0.002], with no difference in sAML and MDS disease groups. PWWP ^MT had worse OS than non-PWWP ^MT in sAML alone [5.2 vs 10.6 mo., p=0.012]. ADD ^MT status was not related to prognostic outcome in any disease groups. Mtase ^MT had better OS than non-Mtase ^MT [11 vs 8 mo., p = 0.003] in sAML alone.

In NTD ^MT pAML patients with poor OS, frameshift ^MT and abnormal karyotype were enriched than in the non-NTD ^MT group [31% vs 17.9%, p=0.04 and 57.9% vs 29.1%, p = 0.001, resp.]. PWWP ^MT sAML patients with worse OS had lower missense [43.2 vs 75.5, p=0.000006], higher nonsense [34.1 vs 9.7, p=0.00006] and abnormal karyotype [68.4% vs 49.1%, p= 0.0086] than in non-PWWP ^MT. Mtase ^MT sAML patients with better OS had higher missense [80.2% vs 59.2%, p= 0.002] and lower nonsense mutations [7.3% vs 20.8%, p=0.01]. CEBPA [10% vs 1.7%, p=0.02] and CSF3R [10% vs 2.9%, p=0.03] were mostly co-mutated in NTD ^MT pAML patients and NRAS [17% vs 7%, p=0.016] in PWWP ^MT sAML patients. Mtase ^MT sAML patients had less BCOR [6% vs 13%, p=0.03] and WT1 [1.8% vs 7%, p=0.04] than in non-Mtase ^MT group.

Conclusion:

non-R882 DNMT3A^MT conveys differential prognosis based on the locus. It conferred poor prognosis in pAML, with more frameshift mutations and abnormal cytogenetics. PWWP ^MT is associated with worse and MTase ^MT with better outcomes in the sAML group. This is the first analysis to not only explore the non-R882 mutations at length but also reveal the significance of domain specific non-R882 DNMT3A ^MT in AML. Our study emphasizes the need to include the domain-based impact of DNMT3A mutational status in the prognostic schema for AML, which may help guide treatment plans and encourage further molecular analysis.