NPM1 Measurable Residual Disease: A Narrative Review

Abstract

Acute myeloid leukemia (AML) with mutated NPM1 (Nucleophosmin 1) is a distinct entity in the 5th edition of the World Health Organization (WHO) Classification as well as a favourable risk category in the 2022 European LeukemiaNet (ELN) Recommendations for AML. The development of myriad techniques for detection of NPM1 Measurable Residual Disease (MRD) has led to improved patient outcomes; however, there is no one winner. Each modality has its own nuances and caveats, and hence using a synoptic and integrated approach is recommended to avoid potential false negatives. Therapy guided by genomic MRD especially in decision for timing of transplant has been reported in literature, which has shown benefits of regular longitudinal monitoring of NPM1. This is a narrative review which discusses in detail the testing modalities for NPM1 MRD in AML. Keywords like ‘NPM1”, “MRD”, “NGS”, were searched in Pubmed, Embase and Scopus and a total of 40 articles were included. Since this is not a systematic review or meta-analysis no inclusion or exclusion criteria were laid and no formal statistics was done. This article was constructed in pursuit of raising awareness for this unmet need, as well as to amalgamate all old and new knowledge in this space. The future is artificial intelligence and machine learning which may pave the way for risk scores.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12288-025-01958-4.

Introduction

Acute myeloid leukemia (AML) with mutated NPM1 (Nucleophosmin 1) is a distinct entity in the 5th edition of the World Health Organization (WHO) Classification. It accounts for the largest molecular subgroup of AML accounting for 30–35% prevalence. The WHO has also abolished the blast percentage cut off for diagnosis of the same and hence all myeloid neoplasia irrespective of blast count, but presence of an NPM1 mutation qualifies as AML. [1, 2] However, in the International Consensus Classification a cut-off of 10% is retained. NPM1 mutation is a loss-of-function mutation, and is an AML specific alteration, which is independently capable of driving tumorigenesis [3]. It is known to co-occur with other genomic alterations like FLT3 (Fms like tyrosine kinase 3) and DNMT3A commonly, which could impact outcomes and upgrade risk stratification as per 2022 ELN classification [4].

Traditionally, after intensive chemotherapy, complete remission [5, 6](CR) is defined mainly on the basis of morphological evaluation of the bone marrow which was false negative in almost 50% cases owing to morphologically undetectable disease. The advent of multi parametric flow cytometry complemented by various molecular methods has augmented [7] the sensitivity remarkable to detect up to 1 copy of NPM1 mutant DNA (Deoxyribose nucleic acid). The International guidelines now mandate quantification of NPM1 mutation by one of the molecular methods which range from quantitative real time polymerase chain reaction (qPCR), digital droplet PCR (ddPCR) or the most recent next generation sequencing (NGS). NGS offers the advantage to detect residual co-mutations as well which may benefit in therapeutic decision making and transplant triage. Genetic risk scores, based on machine learning modeling in NPM1 mutated AML have also been reported, which in future shall also help identify the high risk among the good risk [8].

The development of myriad techniques for detection of NPM1 Measurable Residual Disease (MRD) has led to improved patient outcomes; however, there is no one winner. Each modality has its own nuances and caveats, and hence using a synoptic and integrated approach is recommended to avoid potential false negatives. Therapy guided by genomic MRD especially in decision for timing of transplant has been reported in literature, which has shown benefits of regular longitudinal monitoring of NPM1 [7–9].

This review discusses in detail the testing modalities for NPM1 MRD in AML. This will serve as a ready reckoner for treating hematologists and lab personnel aiding in correct estimation of MRD and shared decision making to render optimum care to the patients [2, 10].

Detection of NPM1 Mutation at Diagnosis

At diagnosis, qualitative detection of NPM1 mutation involves testing based on PCR which are usually followed by gel or capillary electrophoresis. NPM1 is usually a 4 bp insertion in exon 12 of the gene which leads to a frameshift truncating alteration leading to export of NPM1 [11]molecule out of the nucleus into the cytoplasm. This localization can now be detected by immunohistochemistry for NPM1 on bone marrow biopsy specimens, as well as by flow cytometry. More than 55 different mutations have been described in the exon 12 of the NPM1 gene based on the sequence inserted. This results in different transcripts denoted by the English alphabet namely “A”, for TCTG insertion, “B” for “CATG” insertion and “D” for “CCTG” insertion. These are the three most commonly encountered types, of which the most common is type A [3, 12, 13]. The methods for qualitative detection include IHC, MFC (either NPM1-cytoplasmic protein or unique immunophenotype of the myeloid blasts which have been described in various studies. NPM1 detection by flow cytometry as demonstrated in a few studies has been described in the supplementary file.

Molecular techniques include real time PCR based detection of NPM1 transcripts which are usually RNA based tests designed to detect the presence of 4 bp insertion in exon 12 of the gene. Multiplex assays designed to detect NPM1 at baseline usually cannot distinguish between the type of mutations owing to probe design [14, 15]. There are other off the shelf kits which can quantify when each transcript is used in single plex of locked nucleic acid probes. The sensitivity is 5% and it cannot quantify the mutation [16]. The other technique is restriction fragment length polymorphism (RFLP) where PCR is done followed by capillary electrophoresis, and the area under the curve serves as a surrogate quantitative assessment in comparison to the wildtype allele, however, this also suffers from user dependent variabilities, as well as interobserver variabilities and voltage differences [7, 17].

Standardization and optimization of capillary electrophoresis and fragment length analysis will involve stringent primer design, running control samples to establish background fluorescence cut offs and calibration of instrument to avoid voltage differences. The parallel running of controls is recommended to ensure quality checks.

The newest advancement is panel based NGS which incorporates NPM1 hotspot testing along with other genes implicated in myeloid malignancies. This carries the advantage of exact characterization of transcript, ascertainment of mutant allele burden as well the presence of other co-mutations [18] (See Fig. 1).

Fig. 1.

Depicts a representative case as evidence by IHC

NPM1 Molecular MRD [14]

As expected, since NPM1 is a specific disease driver in AML, tracking the same can help ascertain deep responses to ongoing therapy. It serves as one of the ideal leukemia specific targets for monitoring MRD [19]. The ELN working party Consensus document advocates the quantification of NPM1 mutant copies, analogous to other recurrently encountered genetic alterations like RUNX1/RUNX1T1, CBFB/MYH11 and BCR/ABL1. A desirable level of sensitivity for MRD detection is 10^− 3. Two approaches have been advocated for the same namely the PCR based approach and NGS method. This section will discuss both approaches in detail with their advantages and disadvantages especially in the Indian clinical setting. [20, 21]

PCR Approach

RNA based quantitative PCR approaches are deployed. The principle behind using RNA followed by complementary DNA (cDNA) is based on the premise that the expression of these genes is considerably high. This involves real time quantitative PCR and digital droplet PCR [21].

Sample Considerations

The sample can be either peripheral blood or bone marrow aspirate as per ELN recommendations as different disease timepoints. However, the sensitivity of peripheral blood will be five times lower than bone marrow [22–23]. EDTA samples are preferred over heparin, due to potential inhibitory effect of heparin on PCR process. If bone marrow aspirate is used as the sample, then the first 0.5mL i.e. the first pull should be sent for molecular MRD, as the subsequent sample is mainly hemodiluted with peripheral blood. Additionally, if an MFC MRD is requested which also requires a first pull, then the needle should be repositioned and then the sample for molecular MRD should be drawn. This reduces false negative rates of MRD and will help in ascertaining exact burden [24, 25]. In case MRD is requested on peripheral blood, then the sample requirement is minimum 10mL.

Real Time Quantitative PCR

This is based on the TaqMan principle using FRET (fluorescent resonant energy transfer) probes designed to detect mutant NPM1 against a housekeeping gene, which commonly is ABL1 (Abelson 1). The sample should be run in triplicate along with housekeeping gene, two positive controls (which are usually plasmid standards) and one no-template water control [17].

Similar to BCR/ABL1, a normalized copy number is calculated as:

[Normalized copy number: Number of copies of mutant NPM1/Number of ABL1 copies]

This is converted into percentage, and the result is reported. To ensure optimal quality of extracted RNA, the ABL1 copy number is recommended to be at least 10,000. The assay sensitivity is usually recommended to be at least 1 in 1000, although desirable is 10^− 4 to 10^− 6.

Digital Droplet PCR

This is based on the principle of emulsion PCR where each DNA fragment amplifies inside individual droplets. The droplets contain two different probes: one to detect the specific mutant DNA and the other to detect the wildtype probe [19, 26]. After PCR amplification, the fluorescence is detected and the different DNA molecules in each droplet are digitally counted. There are different commercially available platforms to do the same, however, some suffer from labor intensive and user dependent workflows, thus leading to discrepant results. The supreme sensitivity to be able to detect one mutant copy of DNA makes it a desirable modality. Hence fully automated workflow stations which do not require manual handling of droplets (since they are fragile) and which are not limited by high background fluorescence should be employed. The rules of sampling and controls are the same that apply to quantitative real time PCR. The disadvantage, other than technical, is the cost which is entailed herewith, owing to oil emulsion PCR.

Error Corrected NGS Approach

The use of error corrected NGS for detection of molecular MRD has now been included by ELN in 2022, which was initially not validated. Choosing the right tracking targets is quintessential for the assay to be valid and free of errors. However, there are some caveats to this technique. The nucleic acid of choice for detection is DNA, however, RNA based kits are also available. The prognostic impact of co-mutations has been demonstrated by some real-world studies, hence a panel-based approach is preferred, although many centers test only NPM1 alone using an error-corrected approach. However, the important attribute of a panel-based approach is that if a new alteration is detected which was not present at baseline, it should be reported only if it is a valid call against the background noise. The typical sensitivity is up to 0.1% and the sensitivity is calculated as lowest frequency ± 3SD [27]. Whole genome sequencing has been tried, however the same in the MRD setting is elusive owing to the depth required for MRD detection would exponentially increase the cost.

DTA co-mutations, IDH1/2 and FLT3 are known to occur with NPM1. Since DTA mutations may represent pre-leukemia clonal hematopoiesis they are not incorporated in MRD panels, as per ELN recommendations in 2021. However, appearance of new mutations in DTA are common at the time of CR assessment, leading to the clinician’s dilemma [6]. In a study on 150 NPM1 mutated AML patients, 15% patients had 1 or more DTA mutations, which either persisted or were acquired during the course of disease. Clonal hematopoiesis is known to occur in NPM1 mutated AML, and hence these new variations may be considered as part of disease process, rather than only CHIP. CHIP is also known to eventually lead to leukemia in many cases when seen longitudinally, and hence close clinical follow up is recommended [28].

FLT3 and IDH should be separately monitored if using targeted small molecular inhibitors. In the scenario of doublet therapy with VEN + HMA, it is likely to result in residual IDH mutations, which are considered by many researchers as part of clonal hematopoiesis. However, another explanation to this is that it is part of residual disease, and may need intervention [29].

Table 1 depicts the parameters which are to be reported by Labs in each technique.

Table 1.

The MRD assay parameters that need to be reported to the clinic

Parameter and MRD Technique	Flow-MRD	qPCR-MRD	NGS-MRD
MRD approach	Leukaemia-associated immunophenotypes (LAIP)/DfN (different from normal)	Quantitative reverse transcription polymerase chain reaction (qRT-PCR)	Amplicon/Panel-MRD
Analyzed tissue	Bone Marrow (BM)	Bone Marrow (BM) & Peripheral Blood (PB)	Bone Marrow (BM) & Peripheral Blood (PB)
Tissue sampling time Point relative to treatment	Consolidation Therapy (PC2), End of Treatment (EOT) etc.	Consolidation Therapy (PC2), End of Treatment (EOT) etc.	Consolidation Therapy (PC2), End of Treatment (EOT) etc.
Sample Quality And Analysis	Cell number >500,000(adequate) Inadequate	Housekeeping Gene >10,000 copies(adequate) Inadequate	Adequate (e.g. >100,000 reads or >10,000 read families) Inadequate
CD45 numbers + cells	Cell Number	-	-
Blast percentage (%)	Blast percentage (%)	-	-
Target phenotype 1	Markers of Leukaemia-associated immunophenotypes (LAIP) /DfN no.1	Mutated gene, e.g. Nucleophosmin (NPM1)	Mutated gene, e.g. Nucleophosmin (NPM1)
Quantifying the phenotype	Number of cells with phenotype 1 and % of CD45+ cells	Numbers of copy of gene*	% VAF (Variant allele frequency)
Control	--	Name of housekeeping gene	Name of positive control gene (e.g. Nucleophosmin 1)
Quantification of control	--	Copy number of housekeeping genes	%VAF of positive control
Qualitative Minimal Residual disease (MRD) result	Positive vs. negative MRD	Positive vs. negative MRD	Positive vs. negative MRD
Limit of detection (LOD) (especially in MRD is negative)	Limit of detection for each phenotype	Limit of detection for each marker	Limit of detection for each marker
Diagnosis	e.g. Complete Remission (CR) MRD negative	e.g. Complete Remission (CR) MRD negative	e.g. Complete Remission (CR) MRD negative

*Absolute copy numbers should be reported in addition to the fold increase to enable the clinician to make his/her own judgments

A detailed comparison and description of qPCR, dPCR and NGS MRD are depicted in Table 2.

Table 2.

A comparison of techniques to detect NPM1 measurable residual disease

Technique	Sensitivity	Specificity	Cost	Turnaround Time
IHC	20%	60%	2000 INR	24 h
Flow cytometry	37%	90%	20000INR	24 h
RT-PCR	BM: 91% PB: 92.2%	BM: 100% PB: 97%	5000 INR	24–48 h
D-PCR	BM: 100% PB: 70%	BM: 94.5% PB: 88%	15,000 INR	3-4 days
Deep sequencing	BM: 87% PB: 62%	BM: 100% PB: 89%	25,000 INR	7 days

IHC: immunohistochemistry, RT-PCR: real time polymerase chain reaction, D-PCR: digital polymerase chain reaction, BM: bone marrow, PB: peripheral blood, INR: Indian rupee

Definitions for Molecular Response

Timeline of Molecular MRD assessment- As per recommended ELN guidelines, the algorithmic approach to MRD assessment is depicted in Fig. 2.

Fig. 2.

(A) Current European Leukemia Net (ELN) NPM1 molecular MRD guidelines for patient receiving intensive chemotherapy. (B) Concerns about NPM1 molecular MRD in patients receiving non- intensive venetoclax- based regimens

Clinical Implications of MRD Results

MRD monitoring is a powerful prognostic tool to guide treatment decisions in NPM1 mutated AML patients [30]. While treatment decisions based on MRD monitoring is well established in patients receiving intensive chemotherapy, there are no established guidelines for MRD monitoring in patients receiving less intensive treatment regimens. Persistence of NPM1 transcripts in peripheral blood(PB) after 2 cycles of intensive induction chemotherapy was associated with a higher 3 year relapse risk [31]. Conversely, patients who achieved PB MRD negativity after 2 cycles of intensive induction chemotherapy had a 3 year overall survival of 70% and there was no additional benefit of allogeneic stem cell transplant in CR1 in this cohort of patients, regardless of the co-mutation status [32]. ELN guidelines recommend assessing MRD by NPM1 PCR from bone marrow at end of treatment. Patients who are MRD positive (≥ 2%) at end of treatment or have a lesser than 3 to 4 log_{10 reduction} in NPM1 transcripts from baseline should be considered for allogeneic transplant. Patients who have low level of MRD (< 2%) at end of consolidation have a low risk of relapse [33]. Allogeneic transplant helps to improve outcomes in patients who have MRD relapse (conversion from MRD negativity to MRD positivity or MRD increase > 1 log_{10 between} 2 positive samples). Salvage chemotherapy to eradicate MRD should be considered prior to allogeneic transplant as pre-transplant MRD level is predictive of post-transplant outcomes [34]. Patients with negative, low level (< 200 copies in the peripheral blood and < 500 copies in the bone marrow aspirate) and high level MRD had an estimated 2-year OS of 83%, 63% and 13% respectively (p < 0.0001) at median follow up of 4.9 years [35].

Clinical utility of MRD monitoring in AML patients undergoing low intensity treatment is not well established. Patients who achieved MRD negativity (< 10^− 3) with Azacytidine and Venetoclax treatment had better EFS and OS in the VIALE-A study [36]. Recent studies have revealed the utility of NPM1 molecular MRD monitoring in patients receiving venetoclax based low intensity therapies. Patients achieving MRD negativity after 4 cycles of treatment had a 2-year overall survival of 84% compared to 46% if they were MRD positive. In patients who electively stopped treatment after BM MRD negative status after a median of 8 cycles, 2-year treatment free remission rate was 88% [37]. MRD based treatment de-escalation in unfit NPM1 mutated patients is being explored in several clinical trials.

MRD Testing in Indian Perspective: Challenges and Limitations

In the Indian setting, the concept of MRD is currently in nascent stages. It is restricted to specialized tertiary cancer centers and high-end laboratories. This is attributable to the availability of facilities in small facilities and towns. The obvious other reason is the economic burden involved for the patient to do repeated monitoring. In this regard a PCR based MRD test scores over NGS based MRD which in India is currently expensive. However, many centers have developed lab developed validated approaches and launched cost-effective solutions for the same.

Future Perspectives and Conclusion

Despite remarkable strides in the genomics space to measure residual disease, there are still grey zones in definitions as to what is CHIP and what is disease. There is still a lack of harmonization in terms of assay technique, content, sensitivity and reporting guidelines. Despite ELN working party recommendations, not all labs are equipped due to cost restraints and hence there is an unmet need for regional recommendations which may be nation-specific, in terms of cost, personnel and technology which may be available. This article was constructed in pursuit of raising awareness for this unmet need, as well as to amalgamate all old and new knowledge in this space. The future is artificial intelligence and machine learning which may pave the way for risk scores [38].

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Declarations

Conflicts of Interest

None.