NPM1 mutations (NPM1m), seen in ~30% of adults with acute myeloid leukemia (AML), often co-occur with FLT3 internal tandem duplication (FLT3-ITD) [1, 2]. Measurable residual disease (MRD) testing in first complete remission (CR) identifies patients at high risk of relapse after allogeneic hematopoietic cell transplantation (alloHCT) [3, 4]. Quantitative RT-PCR assays are currently recommended for NPM1 MRD testing, while FLT3-ITD MRD testing is DNA-based using next generation sequencing (NGS) [5,6,7]. DNA-based NGS assays for NPM1 MRD testing may offer advantages over RNA, including improved template stability, accurate quantification, and high throughput capabilities, but require validation. We performed NPM1 MRD detection in blood collected from patients with AML during first CR (CR1) prior to alloHCT using a highly sensitive, commercially available DNA-based NPM1 NGS MRD assay.

A randomly selected subset of 190 patients aged 18 or older from the Pre-MEASURE study [3] who received an alloHCT for NPM1 mutated AML during CR1 in the USA between 2013-2019 were included in this analysis (Supplementary Table 1; Supplementary Figs. 1 and 2). Patients provided written informed consent for participation in the NMDP institutional review board-approved CIBMTR database (NCT01166009) and biorepository (NCT04920474) protocols. Research was performed in compliance with all applicable federal regulations pertaining to the protection of human research participants and with approval of the CIBMTR observational research group.

A commercially available research testing kit (IVS, Invivoscribe, San Diego, CA), which can detect DNA-based NPM1 MRD, was used to establish a workflow following clinical testing standards and validated to detect NPM1 insertion variants down to a variant allele fraction (VAF) of at least 0.005% (Supplementary Fig. 3). Results were compared to those previously reported using an anchored multiplex PCR-based (AMP) targeted NGS assay (detection limit 0.01%) and, where applicable, FLT3-ITD MRD results. Additional details are provided in the Supplementary Methods.

NPM1 NGS MRD testing was successful for 186 of 190 eligible (98%) patients, of which 48 (26%) relapsed (median: 4.0 months; range: 0.7-52.5) and 64 (34%) died (median: 9.7 months; range: 0.4-58.9) post-alloHCT. The IVS assay detected NPM1 insertion variants in 71 (38%) patients with a median VAF of 0.0026% (range:0.0002-2.1%) (Supplementary Table 2).

Using the previously reported VAF threshold of 0.01%, the concordance between the IVS and the AMP assays was 96% (n = 21 as both positive and n = 158 as both negative) (Supplementary Fig. 4). Using this threshold, 24 patients tested positive for NPM1 MRD pre-alloHCT by the IVS test, which was associated with significantly increased rates of relapse (52% vs. 20% at 3 yrs; HR = 4.3; P < 0.001) and decreased OS (34% vs. 71% at 3 yrs; HR = 3.6; P < 0.001) compared to those testing negative, in close alignment with the clinical outcomes previously reported using the AMP assay (Supplementary Fig. 5).

The IVS test, however, has an increased sensitivity compared with the previously reported AMP test, allowing confident detection of NPM1m below a VAF of 0.01%. Removing this threshold (ie: any NPM1m detection considered to be MRD positivity) dropped the concordance between the IVS and the AMP assays to 76% (109 cases reported as both negative and 33 as both positive) and resulted in additional patients being classified as NPM1 MRD positive by the IVS test. Patients identified as NPM1m positive in CR1 blood prior to alloHCT by the IVS test (n = 71, 38%) had increased rates of relapse (40% vs. 15% at 3 yrs; HR = 3.4; P < 0.001) and decreased OS (50% vs. 75% at 3 yrs; HR = 2.9; P < 0.001) after transplant compared with those testing negative (Supplementary Fig. 5). When considering additional VAF thresholds, patients in the highest VAF group (≥0.1%) had the worst clinical outcomes compared to patients in other groups (relapse 60%, OS 27%, at 3 yrs) (Fig. 1a). In multivariable analysis, NPM1m MRD burden in CR1 blood prior to alloHCT was also associated with increased risk of relapse and death in a dose-dependent manner, with a VAF of 0.01% or greater associated with the highest risk (Supplementary Fig. 6).

Fig. 1: Association of NPM1 and/or FLT3-ITD measurable residual disease (MRD) and clinical outcomes in patients with acute myeloid leukemia (AML) prior to allogeneic hematopoietic cell transplant.
figure 1

Cumulative incidence of relapse (left) and overall survival (OS, right) are shown for patients based on the presence (MRDpos) or absence (MRDneg) of NPM1 and/or FLT3-ITD variants detected by the Invivoscribe MRD next generation sequencing assays. a All patients with NPM1-mutated AML grouped by NPM1 MRD thresholds. b Patients co-mutated for NPM1 and FLT3-ITD at baseline grouped by the NPM1 MRD status versus NPM1 and/or FLT3-ITD MRD status.

Full size image

Patients were further divided based on their reported FLT3-ITD mutational status at baseline. A total of 123 patients (66%) were co-mutated for FLT3-ITD at baseline. Increased risk of relapse was seen for NPM1m MRD positive (NPM1m VAF > 0) patients regardless of their baseline FLT3-ITD status. For those without FLT3-ITD at baseline, patients testing MRD positive for NPM1 by IVS had increased CIR (29% vs. 10% at 3 yrs; HR = 4.4; P = 0.01) compared to those testing negative; for those with FLT3-ITD at baseline, the same trend was observed (45% vs. 18% at 3 yrs; HR = 2.8; P = 0.003) (Supplementary Fig. 7).

For patients with mutations in both NPM1 and FLT3-ITD reported at baseline, FLT3-ITD MRD was previously performed using the IVS assay [4]. Defining MRD based on the presence of NPM1 vs. NPM1 and/or FLT3-ITD revealed equivalent CIR (45% vs. 44%) and OS (45% vs. 46%) at 3 years. 54 were MRD positive for NPM1 and/or FLT3-ITD (22 NPM1 MRD positive only, and 4 FLT3-ITD MRD positive only). (Fig. 1b). The MRD negative groups also had similar CIR (18% vs. 17%) and OS (75% vs. 76%) at 3 years, and the MRD positive group by either definition had inferior clinical outcomes (CIR and OS) compared to those in the corresponding MRD negative group (P < 0.001 for both outcomes under two definitions). We next evaluated the prognostic value of each test alone. A total of 6 relapse events were identified only by NPM1 MRD testing and 1 relapse event only by FLT3-ITD MRD testing; with equivalent CIR at 3 years for the two tests (45% vs. 64%, NPM1 vs. FLT3-ITD, P = 0.237), (Supplementary Fig. 8). These results were confirmed in the full Pre-MEASURE cohort for co-mutated patients (n = 317) using the AMP assay, where 13 and 1 additional relapse events were identified using NPM1 and FLT3-ITD MRD testing, respectively, and equivalent CIR at 3 years (70% vs. 73%, NPM1 vs. FLT3-ITD, P = 0.499).

In this study evaluating patients with NPM1 mutated AML from the Pre-MEASURE cohort, we show that the detection of NPM1m in pre-transplant blood during first complete remission using a highly sensitive DNA-based assay is associated, in a dose-dependent manner, with a significantly increased risk of relapse and death after alloHCT. Some of this risk could be mitigated by conditioning regimen, with higher intensity regimens being associated with decreased relapse and improved survival in NPM1 MRD positive patients (Supplementary Fig. 9).

Similar to FLT3-ITD [4], an NPM1 MRD VAF threshold of ≥0.01% pre-transplant identified patients at greatest risk of relapse and death. Current guidelines for NPM1 MRD testing have been based on measuring mutant RNA transcript levels [5], additional analytical and clinical utility validation work is needed to compare DNA and RNA based NPM1m MRD testing thresholds [8, 9].

NPM1 and FLT3-ITD mutations often co-occur, presenting multiple potential targets for MRD tracking. In patients co-mutated for both FLT3-ITD and NPM1 at diagnosis, we found that tracking NPM1m showed equivalent ability to identify relapsing patients comparing to tracking mutations in both NPM1 and FLT3-ITD or FLT3-ITD alone pre-alloHCT. Our observation that very few co-mutated patients test MRD positive for FLT3-ITD but not NPM1m is supported by data from other studies [9,10,11]. Therefore, while there may be additional important predictive use cases for FLT3-ITD NGS-MRD testing, our results support findings that validate the prognostic value of pre-transplant NPM1 MRD and support prioritizing it over FLT3-ITD when only one test is available [12].