FLT3-ITD NGS-MRD refines risk stratification in NPM1 and FLT3 co-mutated AML beyond NPM1 PCR-MRD: Analysis of the UK NCRI AML17 and AML19 trials

Abstract

Abstract Introduction In NPM1 mutant AML, detection of measurable residual disease by reverse transcription quantitative PCR (hereafter, NPM1 MRD) in peripheral blood (PB) after two cycles of intensive chemotherapy (PC2) identifies patients at high risk of relapse who benefit from allograft in 1st complete remission (CR1-allo). However, 30-40% of MRD negative patients still relapse. Emerging data supports the prognostic impact of ultra-sensitive FLT3 internal tandem duplication (FLT3-ITD) detection by PCR-NGS (hereafter, FLT3 MRD). In this study, we investigated whether NPM1 and FLT3 MRD technologies have a complementary role in patients co-mutated in both genes. Methods The UK NCRI AML17 and 19 trials enrolled patients with newly diagnosed AML generally aged < 60 years who were fit for intensive chemotherapy. Most patients with NPM1 mutantAML underwent molecular MRD monitoring, which was performed in a central laboratory according to ELN recommendations. We retrospectively performed FLT3 MRD for all patients with NPM1 and FLT3-ITD co-mutations who had stored PB and/or bone marrow (BM) samples. The method used was based on Blatte et al (getITD, Leukemia 2019) using 500 ng gDNA input. Assay limit of quantification and detection were validated to VAF 0.001% and 0.0001%, respectively. The outcomes of interest were overall survival (OS) assessed by the Kaplan-Meier method and log-rank test and cumulative incidence of relapse (CIR) assessed with death in CR as competing risk by Gray's method. The impact of CR1-allo was assessed using time-dependent Cox regression and the Mantel-Byar/Simon-Makuch method. Results Of 525 patients with NPM1 FLT3-ITD co-mutated AML in the trials, 273 (52%) with sufficient gDNA and NPM1 MRD data available were identified. Their median age was 50 years (IQR: 42-57), and 28% underwent CR1-allo. We analyzed FLT3 MRD in 475 samples: 103 PB post-cycle 1 (PC1), 108 BM PC1, 147 PB PC2, 117 BM PC2. NPM1 MRD results were available in 152 PB PC1, 96 BM PC1, 170 PB PC2, 128 BM PC2. The median FLT3 MRD coverage was 1,891,149 reads (IQR: 433,838) with VAF in MRD positive samples ranging from 0.0002% to 15.3120%. Across all time points, more patients were NPM1 MRD positive than FLT3 MRD positive (PB PC1 59% vs 52%, BM PC1 72% vs 66%, PB PC2 29% vs 14%, BM PC2 55% vs 30%). Overall, very few samples (6%) were FLT3 MRD positive and NPM1 MRD negative in the same tissue and timepoint; in half of such cases RNA quality was suboptimal (CtABL > 26.5). For NPM1 MRD, the most prognostic timepoint for OS and CIR was PB PC2 as previously reported. FLT3 MRD was not prognostic for either outcome at PC1 in either PB or BM. However, at PC2 both PB and BM FLT3 MRD status (MRD positive vs MRD negative) was strongly prognostic for OS (PB 35% vs 71% at 4-year; HR 3.1, 95CI: 1.7-5.9 and BM 33% vs 75%; HR 3.5, 95CI: 1.9-6.3) and CIR (PB 75% vs 45% at 4-year; HR 2.7, 95CI: 1.4-5.0 and BM 70% vs 38%; HR 3.1, 95CI: 1.8-5.5). As FLT3 MRD-positivity at PC2 was higher in BM (30%) vs PB (14%), further analysis was focused on patients with PC2 BM FLT3 MRD results available. 94 patients were evaluable for both PC2 PB NPM1 MRD and PC2 BM FLT3 MRD, where 69/94 (73%) tested PB NPM1 MRD negative. Of these, 14/69 (20%) tested BM FLT3 MRD positive, with outcomes notably inferior compared to those testing MRD negative for both PB NPM1 and BM FLT3 (4-year OS 50% vs 82%; HR 3.0, 95CI: 1.2-7.8 and 4-year CIR 57% vs 36%; HR 2.0, 95CI: 0.9-4.7). These findings were consistent on sensitivity analysis limited to 62 patients with optimal NPM1 MRD samples (CtABL ≤ 26.5) (OS HR 5.6, 95CI: 1.6-19.3 and CIR HR 3.5, 95CI: 1.4-8.9). Patients who were PC2 BM FLT3 MRD or PC2 PB NPM1 MRD positive who received CR1-allo had improved 4-year OS (52% vs 24%; HR 0.5, 95CI: 0.2-0.9, p = 0.028). In contrast, there was no benefit from CR1-allo in patients MRD negative for both PB NPM1 and BM FLT3-ITD at the PC2 timepoint (4-year OS 78% vs 84%; HR 1.8, 95CI: 0.4-9.3, p = 0.400). Conclusions FLT3 MRD testing was most prognostic when applied to the BM after the second cycle of chemotherapy and added prognostic information to NPM1 MRD testing, revealing amongst patients with a favorable NPM1 MRD response an additional 20% of patients with poor outcomes who may benefit from additional therapy, such as allograft in CR1. In contrast, patients testing MRD negative for both NPM1 in the PB and FLT3 in the BM at this timepoint had favorable outcomes and did not appear to benefit from allograft in CR1.