The Natural History of NPM1MUT Measurable Residual Disease (MRD) Positivity after Completion of Chemotherapy in Acute Myeloid Leukemia (AML)

Introduction

Molecular MRD assays targeting NPM1 mutant (mut) transcripts have an established role for monitoring treatment efficacy in patients with NPM1^mut AML. Approximately 25% of NPM1^mut patients show persistent MRD level in the bone marrow (BM) at the end of treatment (EOT), which is associated with a higher risk of relapse (Ivey, NEJM 2016; Kronke, JCO 2011). Molecular persistence at low copy number (MP-LCN) is defined by the European LeukemiaNet (ELN) as MRD positivity in patients in morphological complete remission (CR) with <1000–2000 transcripts per 10⁵ ABL and a relative increase of <1 log between any two positive samples collected after EOT (Schuurhuis, Blood 2018). The UK NCRI working group recently reported the impact of NPM1^mut MRD burden and FLT3-ITD status on risk of relapse after allogeneic stem cell transplantation (Dillon, Blood 2020), however the clinical relevance of NPM1^mut MP-LCN in patients who are not transplanted is unknown. We aimed to characterize the natural history of persistent NPM1^mut MRD after chemotherapy and to identify factors associated with subsequent disease progression.

Methods

Consecutive patients with newly diagnosed NPM1^mut AML receiving at least 2 cycles of intensive chemotherapy were included if MRD remained detectable at EOT. UK patients were all enrolled in the NCRI AML17 and AML19 studies. Australian patients were retrospectively identified. BM samples were analyzed by either RT-qPCR (n=75) (Ivey, NEJM 2016) or ultra-deep next-generation sequencing (UD-NGS, n=5, excluded from Cox model). We applied the ELN definitions for molecular progression or relapse (considered together as molecular failure [MF]), MP-LCN and complete molecular remission (CR_MRD-). Kaplan Meier survival estimates were calculated from EOT to death (OS), and/or morphologic relapse (RFS), and/or MF (EFS). The optimal NPM1^mut expression threshold was pre-determined by recursive partitioning. Variables with p<0.1 from univariate analysis were included in the final multivariate Cox model and confirmed by backward selection method. Copy numbers are expressed per 10⁵ ABL copies (for RT-qPCR) or as percentage of total reads (for UD-NGS).

Results

A total of 80 patients had NPM1^mut detectable at a median of 37 days (range 19–90; 12 [15%] were >60 days) from last course of chemotherapy; characteristics summarized in Table 1. The majority of patients (89%) had received 3–4 cycles of chemotherapy. At EOT, the median NPM1^mut transcript level by RT-qPCR was 11 copies (range 0.3–8699; 3 were >2000) and median reduction from baseline was 4.7 log (range 2.1–6.0). A total of 34 patients (45%; 4 not monitored) fulfilled the criteria for MP-LCN.

With a median follow-up time of 24 mo (range 8–77), 27 (34%) spontaneously achieved a sustained CR_MRD- for >6 months, 31 (39%) had MF and 13 (16%) had morphologic relapse without preceding MF (due to rapid relapse [n=8], lack of monitoring [n=3], extramedullary or NPM1 wild-type relapse [n=1 each]) (Figure 1). Nine patients (11%) had persisting MP-LCN (included transient MRD negativity): median interval between EOT to the last known positive sample was 12 mo (range 3–32), and all patients currently remain alive and without morphologic relapse by the data cut-off date July 13, 2020.

Next, we examined the variables associated with MF and/or morphologic relapse. In univariate analysis, factors associated (p<0.1) with disease progression were white cell count, FLT3-ITD, <4 cycles of chemotherapy received, EOT NPM1^mut ≥15 copies, and <4 log reduction from baseline (Table 2). In multivariate analysis (and confirmed by backward selection), only baseline FLT3-ITD (HR 2.5, 95% CI 1.3–4.6) and EOT NPM1^mut MRD <4 log reduction from baseline (HR 3.2, 95% CI 1.7–6.2) retained significance. Patients with none, either, or both of these features had a median EFS of not reached, 5.8 and 1.4 mo, respectively (Figure 2). The corresponding 2-year RFS was 73%, 54% and 0% (median 3.5 mo).

Conclusions

We describe the natural history of patients with NPM1 mutated AML who remain MRD positive after completing intensive chemotherapy. Almost half either spontaneously achieve MRD-negativity or have stable low-level expression without relapse with minimum follow up of 8 mo. Patients with both FLT3-ITD and <4 log reduction in transcript levels at the end of therapy are at very high risk of disease progression and warrant consideration of pre-emptive treatment.