NPM1 mutation subtype switch in acute myeloid leukemia

Mutations in exon 11 of the Nucleophosmin (NPM1) gene are detected in approximately 40% of adult acute myeloid leukemias (AML) with a normal karyotype.¹ Since their discovery, more than 70 different NPM1 mutations have been described in exon 11.² The most common mutation, representing about 70% of NPM1 mutations, is a TCTG tetranucleotide duplication (type A). Other common mutations, accounting for 10% and 5% of cases respectively, involve the insertion of CATG (type B) or CCTG (type D) at the same nucleotide position. Rare mutations or translocations involving exons 5, 9, and 10 have also been observed in AML.³

NPM1 mutations were first included as a specific AML subtype-defining abnormality in the 4^th classification of the World Health Organization (WHO) in 2008, with the blast cut-off being finally lowered in more recent classifications (both WHO and International Consensus Classification).^4,5 Numerous studies have established the favorable prognosis of NPM1-mutated AML, especially in the absence of FLT3-internal tandem duplication (ITD).⁶

Studies of paired diagnosis and relapse samples have shown that NPM1 mutations remain highly stable throughout the disease, making them an effective marker for measurable residual disease (MRD) monitoring.^7,8

Here, we present 3 cases of NPM1-mutated AML who developed a second leukemic episode with a different NPM1 mutation, accompanied by phenotypic changes. All patients were initially treated with intensive chemotherapy alone without allogeneic hematopoietic stem cell transplantation (ASCT) in first complete remission or targeted therapy or immunotherapy and none received maintenance therapy before the disease recurrence. In all patients, mutation change was identified by high-throughput sequencing (HTS) and subsequently quantified using reverse transcription (RT)-quantitative PCR (qPCR) for type A/B/D variants or RT-digital droplet PCR (ddPCR) for other NPM1 variants. IDH2 mutations were retrospectively monitored using genomic DNA-based ddPCR (threshold 0.1% variant allele frequency [VAF]) in all patients. Additionally, a single-cell proteogenomic analysis was conducted in one patient. This study was approved by an Institutional Review Board and conducted in accordance with the Declaration of Helsinki. The first patient (UPN1) was a 29-year-old adult with AML. Immunophenotypic analysis of bone marrow (BM) at diagnosis revealed 56% of CD34-negative blasts with a myeloid phenotype (CD33⁺). Cytogenetic analysis showed a normal karyotype. HTS identified mutations in NPM1 (type A variant), IDH2, FLT3-ITD (detected by fragment analysis) and KRAS (Table 1, Figure 1A). The patient received intensive chemotherapy and achieved undetectable BM-NPM1 MRD by RT-qPCR at the end of treatment. At this point, IDH2 mutation was also undetectable by ddPCR. Two years after diagnosis, the patient experienced a second leukemic episode with the same IDH2-mutated founding clone but displayed an NPM1 mutation switch to a new mutation (i.e., c.863_864insTCGG) along with a new FLT3-TKD1 subclone. MRD monitoring of the type A variant by RT-qPCR remained negative. Additionally, immunophenotyping showed the loss of CD33 expression and aberrant expression of CD56 compared to the first leukemic episode. The disease recurrence was treated with azacytidine and venetoclax, followed by ASCT in complete remission. To date the patient is still in remission.

The second patient (UPN2) was a 60-year-old adult with AML. Morphological examination of BM at diagnosis revealed infiltration of 82% of myeloid blasts. Karyotype was normal and HTS revealed mutations in IDH2, NPM1 (type A variant), SRSF2, RUNX1 and FLT3-JMD. The patient received intensive chemotherapy resulting in undetectable BM-NPM1 MRD by RT-qPCR while the IDH2 mutation was still detectable by ddPCR with a VAF of 5.27%. A second leukemic episode occurred five years after diagnosis and molecular analysis identified the same IDH2 and SRSF2 mutations with an NPM1 mutation switch (i.e., 863_864insCTCG) along with the acquisition of new mutations in ASXL1 and FLT3-TKD1/2 (Figure 1B). The patient underwent an ASCT, allowing a durable complete remission.

The third patient (UPN3) was a 57-year-old man. Morphological examination of BM at diagnosis showed infiltration of 87% of myeloid blasts. Karyotype was normal and HTS revealed mutations in IDH2, SRSF2 and NPM1 (type A variant). The patient received intensive chemotherapy resulting in undetectable BM-NPM1-MRD by RT-qPCR while the IDH2 was still detectable by ddPCR with a VAF of 11.2%. A new leukemic episode occurred four years after initial diagnosis. At this time, blasts showed a monoblastic phenotype: CD117^-, CD11b⁺, HLA-DR⁺, CD36⁺, CD4⁺. The IDH2 and SRSF2 mutations were still present, along with a NPM1 mutation switch (i.e., c.863_864insTCAG) and emerging mutations in ASXL1, KRAS and SMC1A (Figure 1C). NPM1 mutation type A was quantified at 1.31% by RT-qPCR while the newly acquired NPM1 mutation (i.e., c.863_864insTCAG) was quantified by RT-ddPCR at 113%. We were unable to determine whether the low type A mutation signal was due to cross-reactivity with the new NPM1 variant or to persistence of the initial mutation at very low levels (the type A variant was not detected in the HTS raw data). The patient died after two weeks, following a rapid deterioration of his general condition.

To explore the clonal hierarchy in greater depth, we conducted a single-cell analysis on cryopreserved cells of UPN1, at both leukemic episodes. Due to low cell viability at diagnosis, surface marker analysis could not be performed. The single cell genomic and proteogenomic assays were carried out on the Tapestri platform (MissionBio^®, USA). This enabled us to analyze 3,991 cells and 1,814 cells at each episode, respectively. The study of genetic phylogeny at initial diagnosis by single-cell sequencing confirmed the results inferred from HTS bulk sequencing (Figure 1), except from the FLT3-ITD mutation which was undetectable by single-cell sequencing due to bioinformatics issues. The study of the second episode showed that none of the 1,814 analyzed cells carried the NPM1 type A mutation, and, similarly, the NPM1 mutation associated with disease recurrence was absent in the 3,991 cells analyzed at initial diagnosis. The proteogenomic analysis revealed that the NPM1 mutation was present both in the granulo-monocytic (CD33⁺ CD117⁺) and erythroblastic (CD36⁺ CD71⁺) compartments. Moreover, single cell analyses enabled us to define the stem cell compartment (CD34⁺ CD38^-) as being composed of hematopoietic stem cells (HSC, wild-type) and leukemic stem cells (LSC, mutated) (Figure 2A, B).

Table 1.Mutational analysis by high throughput sequencing performed at diagnosis and disease recurrence in the 3 patients.

Figure 1.Clonal evolution of NPM1-switch cases. (A) Clonal evolution in Patient 1 (UPN1) determined by single-cell analysis. (B and C) Clonal evolution inferred from bulk high throughput sequencing analysis in Patients 2 (UPN2) and 3 (UPN3.) (D) NPM1 exon 11 sequence at diagnosis and disease recurrence. HSC: hematopoietic stem cells; LSC: leukemic stem cells; WT: wild-type.

Figure 2.Single-cell analysis at disease recurrence in Patient 1. (A) Single cell proteomic analysis and definition of cell populations in Patient 1 (UPN1). (B) Single cell proteogenomic analysis. HSC: hematopoietic stem cells; LSC: leukemic stem cells; WT: wild-type.

To our knowledge, 2 other AML cases with an NPM1 mutation subtype switch have been reported in the literature.^9,10 Interestingly, in line with our 3 observations, these 2 cases displayed immunophenotypic changes between the 2 episodes. Both patients experienced disease recurrence eight and six years following diagnosis, respectively.^9,10

NPM1-mutated AML is one of the most common AML entities in adults and it carries a generally favorable prognosis. However, these diseases may relapse, in which case they mostly involve a stable NPM1-mutated clone, allowing MRD monitoring through sensitive techniques like RT-qPCR or RT-ddPCR. Here, we present 3 cases demonstrating that NPM1 mutation switches can occur. All 3 cases harbored an IDH2 (R140Q) founder clone, raising questions about the promoting effect of such mutation (or subsequent epigenetic changes) in the emergence of NPM1-mutated clones. This also raises the question of whether this new leukemic episode should be considered as a relapse or as the emergence of a second ‘novo-like’ leukemia (in line with the delayed onset of the second episode, suggesting a long period of underlying IDH2-mutated clonal hematopoiesis), which could confer a chemosensitivity profile and prognosis distinct from true relapses.

Together with cases from the literature, these observations illustrate pitfalls in MRD monitoring. While NPM1-specific PCR assays are recommended by international guidelines, these observations argue for extensive genetic screening at the time of relapse (also allowing the detection of new actionable targets) and the combination of other methods or markers (i.e., flow cytometry, WT1 expression). The use of single cell genomics and HTS-based MRD approaches to track both the founder clone and the emergence of new mutations (including new NPM1 mutations) is a current topic of interest.^11,12 This is all the more relevant in the context of developing targeted therapies, such as IDH inhibitors or menin inhibitors, which have shown activity against NPM1-mutated clones.

Footnotes

• Received November 6, 2024
• Accepted March 12, 2025

Correspondence

B. PODVIN - benjamin.podvin@chu-lille.fr

Disclosures

No conflicts of interest to disclose.

Contributions

Funding

ND was supported by grants from the Ligue Contre le Cancer (AAP 2021 du Septentrion, Comité du Pas-de-Calais) and the CHU of Lille (AAP 2021 du Fonds Hospitalier d’Aide à l’Emergence).

Acknowledgments

We thank Christophe Roumier (Tumour Bank, certification NF 96900-2014/65453-1, Centre Hospitalier Universitaire de Lille, Lille, France) for handling, conditioning, and storing patient samples.