Measurable Residual Disease (MRD) Monitoring in Acute Myeloid Leukemia (AML) with t(8;21)(q22;q22.1) RUNX1-RUNX1T1 Identifies Patients at High Risk of Relapse: Results of the AML Study Group (AMLSG)

Background: Acute myeloid leukemia (AML) with t(8;21)(q22;q22.1) resulting in the RUNX1-RUNX1T1 gene fusion is considered favorable in the 2017 genetic risk stratification by the European LeukemiaNet (ELN). After intensive chemotherapy most patients (pts) achieve complete remission (CR), but relapse occurs in about 50% and is associated with poor prognosis. In this AML subgroup monitoring of measurable residual disease (MRD) has been shown to identify pts at higher risk of relapse.

Aims: To assess the prognostic impact of MRD monitoring in bone marrow (BM) and peripheral blood (PB) in a large cohort of 155 clinically well-annotated t(8;21)-AML pts enrolled in one of six AMLSG treatment trials.

Methods: RT-qPCR was used to quantify RUNX1-RUNX1T1 transcript levels (TL) reported as normalized RUNX1-RUNX1T1 values per 10⁶ transcripts of the housekeeping gene B2M. Samples were analyzed in triplicate, the sensitivity was up to 10^-6.

Results: While pretreatment RUNX1-RUNX1T1 TL did not impact prognosis, both reduction of RUNX1-RUNX1T1 TL and achievement of MRD negativity (MRD^neg) at end of treatment (EOT) were of significant prognostic importance in BM as well as in PB: MR^2.5 (>2.5 log reduction) after treatment cycle 1 and MR^3.0 after cycle 2 were significantly associated with a reduced relapse risk (MR^2.5, BM: P=.034; PB: P=.008 and MR^3.0, BM: P=.028; PB: P=.036, respectively). After completion of therapy, MRD^neg was an independent favorable prognostic factor for cumulative incidence of relapse (CIR) (4-year CIR BM: 17% vs 36%, P=.021; PB: 23% vs 55%; P=.001) and overall survival (OS) (4-year OS rate BM: 93% vs 70%, P=.007; PB: 87% vs 47%; P<.0001). Moreover, maximally selected Gray´s statistic defined specific MRD cut-offs at EOT associated with a lower relapse risk: <83 RUNX1-RUNX1T1 TL in BM and <5 in PB predicted for superior 4-year CIR (BM: 18% vs 61%; P<.0001; PB: 23% vs 65%; P<.0001). During follow-up serial MRD analyses allowed prediction of relapse in 77% of pts exceeding an arbitrary cut-off of 150 RUNX1-RUNX1T1 TL in BM and in 84% of pts with >50 TL in PB, respectively. KIT mutation observed in 28% of pts predicted for lower CR rate and inferior outcome, but its prognostic impact was outweighed by RUNX1-RUNX1T1 TL during treatment.

To determine whether PB could provide similar prognostic information as BM, we compared 680 paired samples (diagnosis, n=125; after cycle 1, n=80; after cycle 2, n=86; at EOT, n=78; during follow-up, n=311). At diagnosis RUNX1-RUNX1T1 TL tended to be slightly higher in BM than in PB (P=.072), but were significantly higher after cycle 1 (P=.008), cycle 2 (P<.001), at EOT (P=.002), and during follow-up (P<.001). RUNX1-RUNX1T1 TL in BM and PB correlated well (r=.87; P<.0001) with on average 1-log lower values in PB. However, 2.5%, 26.7%, 26.9%, and 24.8% of all pairs were discrepant (BM^pos/PB^neg or BM^neg/PB^pos) after cycle 1, cycle 2, at EOT, and during follow-up. Of 104 PB^neg samples obtained during treatment, 46 (44%) were still BM^pos. In the post-treatment period, this fraction decreased to 28% (77 BM^pos/276 PB^neg pairs) (P=.003). Of note, RUNX1-RUNX1T1 TL in all but four of the 77 (5.2%) BM^pos samples were below the cut-off of 150 TL.

Virtually all relapses occurred within one year after EOT with a very short latency from molecular to morphologic relapse strongly suggesting to perform MRD assessment at short intervals during this period. Based on our data we refined the practical guidelines for MRD assessment in RUNX1-RUNX1T1-positive AML: i) along with the current ELN MRD recommendations, BM and PB should be analyzed after each treatment cycle; ii) during the follow-up period, in particular the first year after EOT, MRD monitoring in PB should be performed monthly; in pts with TL >50 in PB, increase of MRD TL >1-log, and/or conversion from MRD^neg to MRD^pos a complementary BM samples should be analyzed timely.

Summary:RUNX1-RUNX1T1 MRD monitoring allows for the discrimination of pts at high and low risk of relapse. MRD^neg in both BM and PB after completion of therapy was the most valuable independent favorable prognostic factor for relapse risk and OS. During follow-up, serial MRD analyses allowed the definition of cut-offs predicting relapse. Moreover, considering that virtually all relapses occurred within the first year after EOT with a very short latency from molecular to morphologic relapse MRD assessment in PB at shorter intervals during this period is indispensable.