In Depth Molecular Characterization of AML with RUNX1-RUNX1T1 at Diagnosis and Relapse

Introduction: Minimal residual disease (MRD) monitoring by qPCR is feasible in AML with RUNX1-RUNX1T1 fusion. The absence of RUNX1-RUNX1T1 transcripts is considered as complete molecular remission (CMR). Risk stratification according to MRD is possible and initial studies allocating MRD positive patients to allogeneic stem cell transplantation have been undertaken (Zhu et al., Blood 2013). However, despite CMR about 10% to 30% of patients relapse. In our cohort 90/130 AML patients with RUNX1-RUNX1T1 fusion reached CMR. We analyzed all 17/90 that relapsed following CMR.

Aims: To analyze the frequency, the clinical parameters and the mutational landscape of RUNX1-RUNX1T1 positive AML relapsing after CMR and to gain insights into the clonal evolution.

Patients and Methods: Between 2005 und 2017 we investigated a total of 130 intensively treated AML patients with RUNX1 - RUNX1T1 fusion at diagnosis and at least two follow up samples. We applied quantitative real-time PCR to detect RUNX1-RUNX1T1 / ABL1 ratios with a sensitivity of 0.001%. We identified 17 patients with relapse following CMR and used unique molecular identifier (UMI) based gene panel sequencing to analyze the clonal architecture and evolution of this cohort in comparison to paired results at diagnosis. UMIs were used for digital error suppression to detect minimal variant allele fractions (VAFs) of up to 0.2%.

Results: CMR was reached in 90/130 pts (69%) after a median of 5 months (range 5-45 months). Median event free survival (EFS) of patients with CMR was not reached (EFS at 2 years 82%). However, 17/90 (19%) relapsed during follow up with a median time to relapse of 13 months after CMR (range 4 to 38 months). The ratio of RUNX1-RUNX1T1 / ABL1 was 2.1 fold higher at diagnosis than at relapse with a median of 80 (range 4-508) versus 40 (range 2-501) respectively (n.s.).

For all 17 cases who relapsed following CMR the diagnostic and relapse samples were subject to sequencing of 63 genes known to be mutated in hematologic neoplasms. At diagnosis we identified a total of 27 somatic mutations in 11 different genes co-occurring with RUNX1-RUNX1T1 in 14/17 patients: ASXL1 (n=6 ), ASXL2 (5), KIT (5), NRAS (2), CSF3R (2), FLT3- TKD (2), FLT3- ITD (1), IDH2 (1), KRAS (1), TET2 (1), RAD21 (1), SRSF2 (1). The median number of mutations per patient was 2 (range 0-5). At relapse we identified a total of 13 mutations in 7 different genes in 8/17 patients. 18 mutations in 9 genes (ASXL2, ASXL1, KIT, NRAS, CSF3R, FLT3-TKD, KRAS, RAD21, SRSF2) were lost at relapse and 4 mutations in 3 genes (KIT, PHF6, TET2)were gained at relapse. The median number of mutations per patient at relapse was significantly reduced: Fourteen patients (82%) had at least one mutation at diagnosis and only 9 patients (53%) at relapse (p=0.026). The median variant allele fraction (VAF) of co-occurring mutations at diagnosis was 24% (range 1-24) and 42% (range 3-67) at relapse (1.7 fold, n.s.). Changes in VAF did not reveal an evident general principle, however in some cases the gene with the highest VAF at diagnosis disappeared at relapse suggestive of the selection of a subclone.

When analyzed for clonal evolution 13/17 patients (76%) showed a clonal change at relapse: nine of 17 patients (53%) relapsed with an ancestral clone, 3 (17%) with an ancestral clone with evolution and one (6%) with the initial clone with evolution. Four of 17 patients (24%) relapsed with the initial clone.

As control we sequenced the diagnostic sample in 42/90 patients who did not relapse following CMR. A total of 30 mutations were found in 9 genes in 21/42 patients (median=0.5): ASXL2 (25%), TET2 (11%), KIT (10%), NRAS (10%), FLT3- ITD (6%), CBL (5%), ASXL1 (5%), FLT3- TKD (2%), IDH2 (2%). When compared to the 17 patients with relapse following CMR, the presence of an ASXL1 (p=0.0172) mutation at diagnosis and the total number of mutations at diagnosis (median=2 versus median=0.5; p=0.0024) were significantly associated with relapse.

Conclusion: We analyzed the clonal evolution of RUNX1-RUNX1T1 positive AML following CMR. We observed losses and gains of mutations at relapse, but the majority of cases showed a reduction of numbers of mutations. This underscores that RUNX1-RUNX1T1 fusion is the key driver aberration and stable over time. However mutations in ASXL1 and the number of co-mutations at diagnosis is significantly associated with relapse following CMR. This should be considered when tailoring treatment strategies.