RUNX1 Mutations in MDS, s-AML and De Novo AML: Differences in Accompanying Cytogenetic and Molecular Genetic Alterations, Clonal Architecture and Outcome

Background: RUNX1 mutations are frequent in de novo AML, s-AML and MDS cases. AML with mutated RUNX1 constitutes a new provisional entity, was reported to be associated with a distinct pattern of cytogenetic and molecular genetic abnormalities and with an adverse prognosis. However, differences in accompanying aberrations and prognosis between the respective entities have not been analyzed to date.

Aims: Evaluation and comparison of associated genetic alterations, prognosis and presence of RUNX1 in the main clone in (1) de novo AML, (2) s-AML and (3) MDS cases with RUNX1 mutations.

Patient cohorts and methods: The total cohort comprised 546 cases, 379 with de novo AML, 73 with s-AML and 94 MDS cases with RUNX1 mutations. The total cohort included 348 male and 198 female patients, median age was 72 years (range: 18-91 years). All patients were investigated using chromosome banding analysis (CBA) and amplicon sequencing of RUNX1 . All cases with s-AML and MDS and 141 cases with de novo AML were subject to further mutation analyses of ASXL1, BCOR, CBL, DNMT3A, ETV6, EZH2, FLT3 -ITD , GATA2, IDH1, IDH2, JAK2, KRAS, MPL, SF3B1, SRSF2, TET2, TP53, U2AF1 and ZRSR2 . Variants of unknown significance were excluded from statistical analysis.

Results: In the total cohort, CBA analysis revealed a normal karyotype (NK) in 56% of cases, 18% harbored trisomies, 2% a complex karyotype (>3 abnormalities), 23% other aberrations. 56% of cases with trisomies harbored +8, in 26% of patients +13 was found. In the subgroups, the proportion of cases with NK was largest in cases with MDS (73%), followed by s-AML (58%) and de novo AML (52%) (p<0.001 for MDS vs. s-AML/ de novo AML). Thus, cases with other aberrations were less frequent in MDS (12% vs. 26% in de novo AML and 23% in s-AML; p=0.003), whereas the frequency of trisomies was comparable between the subgroups (15%, 18% and 19% for MDS, s-AML and de novo AML, respectively). However, differences were detected regarding the trisomy types: in de novo AML, +8 and +13 were frequently found (49% and 35% of all cases with trisomies, respectively), in contrast to s-AML and MDS, which almost exclusively harbored +8 (92% and 86%, respectively), whereas +13 was absent in these entities (p=0.032 and p=0.018, respectively).

Mutation analyses of 308 cases (141 de novo AML, 73 s-AML, 94 MDS) revealed SRSF2 (43%), ASXL1 (40%), TET2 (26%), SF3B1 (16%), DNMT3A (14%), EZH2 (12%) and IDH2 (12%) as most frequently mutated genes. In MDS, a higher frequency of ASXL1, TET2 and EZH2 mutations was observed compared to the other cases (ASXL1 : 57%, 44% and 27% in MDS, s-AML and de novo AML, respectively, p=0.001; TET2 : 38%, 33% and 14%, p=0.011; EZH2 : 22%, 7%, 7%, p<0.001). On the other hand, DNMT3A and IDH2 mutations were more abundant in cases with de novo AML (DNMT3A : 7%, 12%, 18%, p=0.006; IDH2 : 7%, 7%, 17%, p=0.004).

To analyze if the RUNX1 mutation was present in the main clone or in a subclone of the respective patient, the RUNX1 mutation load (ML) was compared to the ML of accompanying mutations. RUNX1 mutation was considered to constitute the main clone, if the difference between the MLs was ±10%. In the majority of cases with de novo AML and s-AML, RUNX1 was present in the main clone (80% and 82%, respectively), whereas this was the case in merely 64% of MDS patients.

Median overall survival (OS) in the total cohort was 14 months. OS was found to be significantly shorter in s-AML (8 months) cases compared to de novo AML (14 months, p=0.035) and MDS (15 months, p=0.017). Mutations in EZH2 and SF3B1 negatively impacted OS in the total cohort (8 vs. 14 months, p=0.024; 8 vs. 14 months, p=0.036), while SRSF2 mutations showed a trend towards a better outcome (16 vs. 13 months), which was especially pronounced in s-AML (16 vs. 6 months). ASXL1 and DNMT3A mutations depicted a negative impact on OS in cases with de novo AML only (9 vs. 20 months, p= 0.033, 11 vs. 16 months, p=0.049), whereas EZH2 and TET2 mutations caused a shorter OS in MDS (14 vs. 17 months, p=0.049; 9 vs. 18 months, p=0.039).

Conclusion: 1) MDS cases with RUNX1 mutations differ from RUNX1 -mutated AML by accompanying cytogenetic aberrations (high frequency of NK, lack of +13) and additional mutations (high frequency of ASXL1, TET2 and EZH2 mutations). 2) In the majority of de novo AML and s-AML cases, RUNX1 mutations constitute the main clone. 3) Patients with RUNX1 mutated s-AML had a poorer OS than patients with de novo AML and MDS, although the presence of SRSF2 mutations counterbalances this effect.