Next-Generation Sequencing (NGS)-Based Measurable Residual Disease (MRD) Monitoring in Acute Myeloid Leukemia with FLT3 Internal Tandem Duplication (FLT3-ITD+ AML) Treated with Additional Midostaurin

Background: FLT3-ITD occurs in ~25% of adult AML patients (pts) and is associated with poor prognosis. MRD monitoring is of high prognostic relevance, but restricted to certain AML subtypes. FLT3-ITD represents an attractive target for MRD monitoring in particular in pts treated with a tyrosine kinase inhibitor. FLT3-ITD MRD monitoring is hampered by the broad heterogeneity of ITD length and insertion site (IS). NGS may overcome these limitations offering the opportunity for MRD monitoring in FLT3-ITD⁺ AML.

Aims: To validate our recently established NGS-based FLT3-ITD MRD assay in a defined cohort of FLT3-ITD⁺ AML pts treated within the AMLSG16-10 trial (NCT01477606) combining intensive chemotherapy with midostaurin followed by midostaurin maintenance and to evaluate the prognostic impact of FLT3-ITD MRD monitoring.

Methods: Using FLT3-ITD paired-end NGS (Illumina MiSeq) with a variant allele frequency (VAF) sensitivity of 10^-4-10^-5 (Blätte et al., Leukemia 2019), 227 bone marrow (BM) and 17 peripheral blood samples from 61 FLT3-ITD⁺ AML pts were analyzed at diagnosis (Dx), after two cycles of chemotherapy (Cy2), at the end of treatment (EOT), and during 3-6 months follow-up (FU). All pts achieved complete remission (CR) after Cy2. Allogeneic hematopoietic cell transplantation in first CR was performed in 40 (66%) pts. Mutational status for NPM1 and DNMT3A was available for all pts (NPM1^mut, n=48; DNMT3A^mut, n=33; NPM1^mut/DNMT3A^mut, n=31), and NPM1^mut MRD data for 41 pts.

Results: At Dx we identified 191 ITDs; median length was 45 nucleotides (range, 9-194) and median VAF 0.279% (range, 0.006-90.21). Of the 191 ITDs, 133 (70%) located in the juxtamembrane domain (JMD) and 58 (30%) in the tyrosine kinase domain-1 (TKD1). There was no correlation of VAF with length or IS, whereas ITD size correlated with IS: the more C-terminal the IS, the longer the ITD (Rho=0.51; p<.001). Total ITD VAF per pt was in median 34.3% (range, 0.007-90.21) and correlated positively with white blood cell count (WBC, Rho=0.314; p=.021) and lactate dehydrogenase serum level (LDH, Rho=0.274; p=.04), and inversely with the number of ITDs (Rho=-0.265; p=.04). Most pts (67%) exhibited >1 ITD at Dx (median 2; range, 1-16). Categorizing pts according to IS as JMDsole (46%), JMD/TKD1 (34%), and TKD1sole (20%) revealed that JMD/TKD1 pts exhibited more ITD subclones (p<.001) and a lower total VAF at Dx (p=.03). There were no correlations with any other clinical or genetic features.
Pts’ total ITD VAF significantly decreased after Cy2 and at EOT (median log₁₀ reduction: 4.4 and 4.7; p<.001, each), and MRD negativity (MRD^-) was achieved in 67% and 87% of pts, respectively. According to subgroups, pts with JMDsole or TKD1sole showed deeper MRD reduction compared to JMD/TKD1 pts after Cy2 (4.6 vs 4.7 vs 3.7 log₁₀; p=.06) and at EOT (4.8 vs 4.8 vs 4.0 log₁₀; p=.02) but this did not result in a significant difference in achievement of MRD^-. Concurrent NPM1^mut was of favorable impact for log₁₀ VAF reduction (median, 4.7 for DNMT3A^mut/NPM1^mut vs 4.6 for NPM1^mut vs 2.8 others; p=.003) and MRD^- (77 vs 76 vs 31%; p=.01) after Cy2, but exerted no impact at EOT. Correlating NPM1^mut and FLT3-ITD MRD course revealed a positive correlation after Cy2 (Rho=0.327; p=.03), but not at EOT (Rho=0.250; p=.10), likely due to the higher sensitivity of the real-time quantitative PCR-based NPM1^mut MRD assay.
Median follow-up was 3.4 years (95% CI, 2.6-4.6). Survival analyses with respect to cumulative incidence of relapse (CIR; n=60) and overall survival (OS; n=61) revealed significantly lower CIR for total VAF at Dx >34.3% (p=.03), a VAF reduction >4.7 log₁₀ (MR^4.7) at EOT (p=.001), and for MRD^- pts at EOT (p=.001). There was no impact on OS. In preliminary exploratory Cox regression (n=48), including BM blasts, WBC, LDH, age, and NPM1^mut as covariables, MRD^- at EOT was the only consistent favorable variable for CIR (HR, 0.1; p=.001) and OS (HR, 0.27; p=.03). During FU, 5/8 (63%) MRD⁺ pts at EOT became MRD^- and 4/53 (8%) MRD^- pts converted to MRD⁺ resulting in consecutive relapse in 2 pts.

Conclusion: In this first cohort of FLT3-ITD⁺ AML pts treated with intensive chemotherapy and midostaurin in the prospective AMLSG16-10 trial we could demonstrate that FLT3-ITD NGS-based MRD monitoring is feasible and represents a promising tool to evaluate therapy response and identification of pts at a higher risk of relapse. Further analysis of the study cohort is ongoing.