Molecular MRD Assessment in Acute Myeloid Leukemias

Abstract Detection of measurable residual disease (MRD) is of significant value in the management of acute myeloid leukemia (AML) patients. Along with multicolor flowcytometry (MFC), molecular techniques form an integral tool in AML MRD detection. Multiple studies have reiterated the role of molecular MRD evaluation in AML at defined timepoints during the course of therapy, helping in risk stratification, prediction of relapse, and as guide for pre-emptive therapy. The latest World Health Organization (WHO) classification (WHO-HEME5) has refined the classification of AML bringing forth newer entities defined by molecular abnormalities, especially fusions. AML is a clonally heterogeneous disease characterized by a spectrum of multiple molecular abnormalities including gene mutations and fusions. Accordingly, the molecular methods employed are also diverse and need robust technical standardization in clinical laboratories. Real-time quantitative polymerase chain reaction (PCR), digital PCR, and next-generation sequencing (NGS) are the major molecular platforms for AML MRD. The European LeukemiaNet (ELN) MRD Working Party consensus document recently updated in 2021 for the first time has reflected on the technical recommendations for NGS MRD in AML and stressed the value of an integrated approach. It is, therefore, desirable for physicians, scientists, and pathologists alike to thoroughly understand these molecular methods for appropriate utilization and interpretation. In this article, we discuss the various facets of molecular methods for MRD detection in AML including technical requirements, advantages, drawbacks, and applications.


Introduction
Acute myeloid leukemia (AML) constitutes approximately 1% of all cancer cases as per the Surveillance, Epidemiology, and End Results Program (SEER) Program in the United States and considered a disease with dismal prognosis.It is one of the most common leukemias in adults and not a very uncommon leukemia in children.Over the last two decades, our understanding of disease biology and pathogenic mechanisms has improved and has resulted in newer targeted therapies in AML.Detection of the residual disease burden beyond conventional morphological techniques known as measurable residual disease (MRD) is a standard of care in leukemia patients and plays a crucial role in disease prognostication, predicting and monitoring of response.MRD positivity during complete morphological remission is known to be associated with increased relapse rates and poor survival outcomes in AML. 1,2In addition, MRD status is also explored for treatment decisions.
The European LeukemiaNet (ELN) MRD Working Party recommendations are widely followed for AML MRD, which were last updated in 2021. 3The two major modalities for MRD detection in AML are multiparametric flow cytometry (MFC)  and molecular MRD testing.Each of these two techniques has its merits and demerits.MFC is widely available and can be applied in almost all cases of AML.Assessment of CD34 þ/ CD38-leukemic stem cell burden in combination with aberrant markers by MFC has also been used for MRD detection.However, the detection of true minimal residual disease against a background of regenerative hematopoiesis and chemotherapy effects (such as immunophenotypic shifts), along with the technical and analytical expertise required, together pose practical challenge for AML MRD detection by MFC.This is further complicated by the fact that unlike in precursor B lineage acute lymphoblastic leukemia, the frequencies of leukemia-associated immunophenotypes in AML do not always permit a clear detection of the leukemic clone.
AML is a heterogeneous disease marked by the presence of numerous clones and subclones of leukemic cells harboring multiple molecular changes that undergo fluctuations during the disease course and therapy.Tracking of the molecular aberrations detected at diagnostic timepoint forms the basis of molecular MRD testing.The list of newer AML entities defined by molecular methods continues to expand, as evident in the latest World Health Organization (WHO) classification of hematolymphoid neoplasms (WHO-HEME5). 4The implementation of molecular techniques for MRD monitoring is therefore need of the hour for hematopathology laboratories.We review the molecular methods for AML MRD considering the recent ELN AML MRD recommendations 3 and also discuss their clinical utility in this paper.

Overview of the Molecular Aberrations and Methods of Detection
Molecular studies performed at diagnostic timepoint help in the identification of molecular aberrations in most AML patients, each of which can be potentially tracked for MRD using molecular techniques.7][8][9][10][11] Given the genomic heterogeneity of AML, a variety of molecular techniques are employed for monitoring MRD.These range from real-time quantitative polymerase chain reaction (RT-qPCR) and digital (dPCR) to next-generation sequencing (NGS)-based approaches.The latter include conventional "bulk" NGS (for single genes as well as multigene panels) and error corrected NGS.Other methods that have been reported in literature but not widely used for the purpose of MRD detection include chimerism studies posttransplant 12 and immunoglobulin heavy chain and T cell receptor gene rearrangements. 13Exploratory methods for AML MRD include sequencing of cell free DNA, single cell genomics, and metabolomic profiling. 14,15►Fig. 1 provides an overview of the common molecular aberrations and the molecular methods used to detect AML MRD.
Needless to say, technical standardization and stringent validation in the laboratory are utmost challenging and irreplaceably required for implementation and reporting in a clinical setting.

Technical Considerations: Preanalytical Sample Considerations
The recommended sample for molecular MRD in AML as per ELN 2021 recommendations is either bone marrow (BM) or peripheral blood (PB).The volume of draw for BM aspirate should be up to 5 mL, while a volume of more than or equal to 10 mL can be required when using PB. 3 In addition, the white cell count and the assay to be performed also need to be considered while determining the required volume of sample drawn.First pull of marrow aspirate should be preferred to avoid hemodilution.
Although ELN recommends both ethylenediaminetetraacetic acid and heparin as the anticoagulants for sample collection, heparin is better avoided given its inhibitory effect on PCR reactions.Cell isolation techniques to reduce hemodilution by granulocytes can be used (e.g., Ficoll separation), provided that the method followed in laboratory is consistent, as leukemic cell percentage can be altered based on it.

Peripheral Blood versus Bone Marrow
The use of BM yields up to one log higher sensitivity compared to PB. 16 Multiple studies have also shown concordance between the two sample types.Recently, Skou et al reviewed literature concerning the utility of PB versus BM for molecular AML MRD and suggested that PB may be sufficient for predicting impending AML relapses. 17The ELN 2021 MRD document, however, currently recommends the use of both samples (PB and BM) at diagnostic timepoint (as comparators) if using log reduction for MRD calculation.PB sample is preferred at post-induction (PI) timepoint in NPM1 mutated and core binding factor (CBF) AMLs, while BM sample is recommended for post-consolidation (PC) timepoint.Subsequent follow-up can either be done on PB or BM. 3

Other Considerations
Once the sample is received by the laboratory and sample requirements are met, it is essential to confirm timepoint of testing, the diagnostic molecular results, and the current disease status on morphology.These factors can influence the choice of assay to be performed and the interpretation.detection of CBF-AML, acute promyelocytic leukemia (APL), and NPM1 mut AML. 3 For patients with no molecular abnormality identified, the recommended method remains MFC. 3

RT-qPCR: General Considerations
RT-qPCR has been the most widely adopted method for molecular MRD assessment and involves quantitative detection using fluorescent probes designed specifically to target the regions of interest.The molecular abnormalities detected by RT-qPCR broadly include (a) fusion transcripts, (b) single gene variants and indels, and (c) genes with altered expression levels in AML.
][20][21][22][23][24] The technical requirements established by Europe Against Cancer (EAC) in 2003 remain the resort for the implementation of molecular MRD in AML by RT-qPCR. 3,25,26Along with the region of interest, a house keeping gene (e.g., ABL1) is also amplified (minimum 10,000 copies) and detected using specifically designed primers and probes.The obtained copy number of target gene/ fusion transcript is then normalized against that of a housekeeping gene or the wild-type counterpart of the respective mutated gene (e.g., NPM1).
A positive RT-qPCR result is defined as amplification with cycle threshold (Ct) values below 40 in at least two of the three (triplicates) results of the sample tested, when the value of cycling threshold is 0.1. 3,25,26Assay sensitivity needs to be determined especially while reporting a negative result.The sensitivity achieved by RT-qPCR ranges between 10 À4 and 10 À5 .►Table 1 highlights the major studies where RT-qPCR has been used for the detection of molecular MRD in AML in major molecularly defined entities.

RT-qPCR for Uncommon Fusion Transcripts
While the RT-qPCR for CBF AMLs and APL is performed using absolute quantification using the respective standards, the reporting of uncommon fusion transcripts (e.g., KMT2A:: MLLT3, NUP98::NSD1) may require the use of relative quantification method.As these transcripts are rare, it is not practically feasible for clinical laboratories to procure and sustain standards for absolute quantitation of all of them.Relative quantification involves comparison of baseline and follow-up samples calculating delta delta Ct (DD Ct). 25

RT-qPCR for Common Gene Mutations
Mutations in NPM1 gene constitute over 30 to 40% of all AMLs and are the "driver mutations."Being stable over course of disease, NPM1 mutations form a suitable target for MRD detection.There are more than 50 types of NPM1 mutations reported, depending on which 4 bp insertion is present in the exon 11 (NM_002520) of NPM1 gene.The type A NPM1 mutation (TCTG) is the most common ($75-80%). 41,42Interpretation of RT-qPCR requires prior knowledge of mutation type, as type-specific probes would be required.Failure to use specific type of primer-probes can lead to a false negative MRD result.Currently, commercial plasmid standards are available only for the three most common subtypes (A, B, and  D) of NPM1 mutated AML. 43Also, the possible cross reactivity of probes between wild-type and mutated NPM1 alleles is also an important consideration while using RT-qPCR for NPM1 MRD. 41,42Similarly, RT-qPCR studied for FLT3-ITD MRD 44 also carries drawbacks especially ascribed to the variable length of ITD in every case.
Although fragment length analysis (GeneScan) is conventionally used for the detection of NPM1 mutation and FLT3-ITD at diagnostic timepoint, its utility in MRD detection is limited.In context of FLT3-ITD, however, fragment length analysis does offer an advantage of detecting multiple mutant peaks and calculation of allelic ratio (calculated as Area Under Curve of mutant peaks divided by that of wild-type peak).

Limitations of RT-qPCR
Although RT-qPCR is considered "gold standard" for molecular MRD and is highly sensitive, cost-effective, and widely available, there are limitations to the method as well.Drawbacks of RT-qPCR include limited applicability, requirement of high-quality validated standards, prior knowledge of the molecular abnormality, need for specific primer-probes, separate assays for every molecular abnormality, and practical difficulty for application in tracking insertions deletions (indels).Moreover, application of RT-qPCR can be extended to only around half of all AML patients, limiting its utility in substantial proportion of cases. 45

Digital PCR
Digital PCR forms a lucrative alternative to RT-qPCR with potentially higher sensitivity and specificity.PCR is conducted in numerous sub partitions in the form of solid chambers known as chips or water-in-oil droplets (digital droplet PCR).Absolute quantitation of target gene copy number is obtained by this technique.Unlike RT-qPCR, the need for standards is bypassed.However, the requirement to set up individual assays for different molecular targets still remains.Guidelines for performing quantitative digital PCR have been published in 2013 and updated in 2020 by the Minimum Information for Publication of Quantitative Digital PCR Experiments (dMIQE) group. 46,47][50][51][52][53][54] Though the appropriate threshold remains to be established in larger studies, provisionally, a variant allele fraction (VAF) of more than or equal to 0.2% is defined as dPCR positivity for genomic DNA.There is no suggested optimal threshold positivity using cDNA currently.Drawbacks of dPCR include subsampling errors, partitioning errors, and the need to develop individual assays for each genetic alteration. 3,43

Limit of Detection
The limit of detection (LOD) is the measure of lowest concentration of the analyte that can be reliably measured by an assay, with a specific degree of confidence.A LOD of 10 À3 or lower is recommended for molecular MRD assessment. 3It is required that LOD be established for every marker targeted by the MRD assay, both RT-qPCR and NGS.Along with this, limit of blank (LOB) also needs to be determined using healthy/ negative controls. 3As numerous factors can affect the calculation of LOD of PCR assays, it is recommended to follow the EAC guidelines for establishment of the same.

Next-Generation Sequencing
High-throughput sequencing or NGS makes the detection of multiple mutations (panel based) possible for multiple

Conventional NGS versus Error Corrected NGS
MRD detection being a rare event analysis, distinguishing a true call from false call, is of utmost significance.Conventional NGS technology is prone to errors (0.1-1%), owing either to the sequencing technique itself or can be PCR generated. 65Hence, application of error correction methods is important to ensure appropriate results while using NGS for AML MRD.Error correction methods are broadly physical (involving changes in library preparation and processing steps) or computational. 65-67►Fig. 2 illustrates error cor-rection in brief (2A) and summarizes the major error correction methods (2B).

Library Preparation
Library preparation involves sample processing steps to make the sample ready for sequencing, and broadly includes fragmentation, end repair, A-tailing, and adaptor ligation.
Modifications in library preparation are required for MRD detection in order to achieve higher sensitivity and correct variant calls.Incorporation of unique molecular identifiers (UMI) is the most widely used technique that can be performed using commercial or laboratory developed methods (►Fig.2B).Broadly, the methods for UMI incorporation are either single molecular inversion probe system (smMIPS) based or PCR based.Further, the UMIs can be incorporated on  single strand or both strands (duplex), with duplex method providing higher error correction. 65,66We have demonstrated the utility of error-corrected NGS using in house laboratory developed 34-gene panel using smMIPS and single gene assays for NPM1 and FLT3-ITD. 64

LOD for NGS Assays
Results of NGS are expressed in terms of VAF, defined by proportion of reads containing the mutant allele out of the total reads obtained for that particular locus.The LOD for conventional NGS methods applied at diagnostic timepoints ranges from 2 to 5%, which lowers down to the desirable 0.1 to 0.5% when error correction is applied. 66

Sequencing Depth
Achievement of high sensitivity in MRD requires a higher sequencing depth (>20,000x) while using non-error corrected sequencing. 3,66A read depth high enough to discriminate true call from background noise should be targeted. 3eeper approach is usually possible using a limited set of genes and tracking mutations detected earlier in the patient.Use of smaller panels with deeper sequencing leads to compromise on "breadth" leading to chance of missing mutations: (a) present at diagnostic timepoint but not targeted in the MRD panel and (b) newer mutations that could potentially lead to relapse.

Markers for Panel-Based NGS-MRD
Potentially every gene mutation detected at diagnostic timepoint can be targeted for MRD detection by targeted NGS.A panel of 23 commonly mutated genes in AML has been suggested by ELN. 3 Four important considerations while choosing markers for analysis of NGS MRD are: (i) DTA mutations (DNMT3A, TET2, ASXL1) associated with clonal hematopoiesis of indeterminate potential (CHIP): The mutations associated with CHIP should not be considered for MRD analysis as they are known to persist post-remission and also may not be part of leukemic clone. 57,64,68If DTA mutations are the only detected mutations at the diagnostic time point, MFC based MRD or PCR-based MRD should be applied. 3ii) Germline mutations: Mutations in genes known to be associated with germline predisposition to myeloid malignancies example, CEBPA, DDX41, are detected at near heterozygous VAF ($50%) at baseline and persist at further timepoints.These variants are noninformative and should not be used as MRD markers.3 A potential exception to this situation could be post allogenic hematopoietic stem cell transplant (allo-HCT) 65 (iii) Signaling pathway genes: Mutations in the signaling pathway, for example, FLT3-ITD, FLT3-TKD, KIT, KRAS, NRAS, are likely secondary events in leukemogenesis and are mostly subclonal.When detected at MRD timepoint, they represent true MRD positivity; however, their absence in isolation should not be taken as evidence of MRD negativity.3 (iv) Targeted therapy: When performing MRD in a patient who has received targeted therapy, for example FLT3 inhibitors or IDH1/2 inhibitors, a marker apart from that targeted should also be included in the MRD analysis.3 Single Gene NGS: NPM1 and FLT3-ITD The limitations of RT-qPCR (as discussed in previous sections) are addressed to a large extent with the use of NGS for NPM1 MRD detection as patient-specific primers are not required. 41,42,69The FLT3-ITD mutations in AML involve insertions in the exon 14 (NM_004119) of variable length affecting the juxta membrane domain of the FLT3 gene.With the widespread use of targeted therapy in the treatment of FLT3-ITD AML, monitoring forms an essential part of patient management.Sensitive PCR-based assays would require patient specific FLT3-ITD primers limiting its practical utility.Additionally, the variable length of ITDs also limits the use of conventional NGS algorithms. 701][72] As emphasized in a previous section, ELN recommends integration of an additional marker in MRD analysis while interpretating a negative FLT3-ITD MRD, owing to the unstable and possible subclonal nature of the FLT3-ITD mutations. 73,74fining MRD Positivity by NGS Presence of a mutation at VAF more than or equal to 0.1% provisionally defines NGS-MRD positivity.As a VAF lower than 0.1% might also be possibly associated with adverse outcome, it is recommended to be reported as MRD-LL (low level). 3

Informatics Considerations
Bioinformatics involves computational methods to process sequencing raw data and derive results for analysis.Calculation of error rate, reducing false positive calls, and calculation of sensitivity are crucial elements in bioinformatics in context of AML MRD.Derivation of background error rate involves calculation of the largest VAF of nucleotides that flank the target (excluding primer sequence).Sensitivity for a sample is calculated by dividing the mean background error by number of read families (or reads) expressed as percentage. 3here are currently no uniform recommendations on the use of specific bioinformatics pipeline, and this area still requires harmonized efforts for standardization.It is important to note that tools for detection of indels are required to be different from the regular alignment tools for SNVs.
ELN recommends that for a sample to be evaluable when error-corrected sequencing approach is used, at least 10,000 read families and more than 10 mutant reads should be present.At least three reads should be present in each read family.For a sample to be evaluable for non-error-corrected sequencing approaches, the total reads and mutant reads recommended are more than or equal to 60,000 and more than 60, respectively.When background error correction is applied, MRD positivity is defined by a VAF greater than the sum of mean background error and 3x standard deviation of the background error.There can be other methods to define MRD positivity. 3

Targeted RNA Sequencing
As previously discussed, RT-qPCR forms the gold standard for molecular MRD detection of common fusion transcripts, while the follow-up of uncommon fusions relies on relative quantification.Targeted RNA sequencing using UMI correction forms a promising alternative for molecular MRD of fusion transcripts possible in a single assay for multiple fusion transcripts. 75Dillon et al demonstrated a sensitivity of 1 in 100,000 for detection of molecular MRD in AML using UMIbased multiplexed RNA sequencing assay. 76The utility of RNA sequencing has also been explored by Kim et al in cases of CBF-AML, where they demonstrated that the reduction in disease burden was comparable between RNA sequencing and RT-qPCR. 77However, unlike DNA-based NGS MRD, the literature on RNA sequencing based MRD for follow-up of fusion transcripts is still sparse and requires further exploration.

Limitations of NGS
While high sensitivity, throughput, and wider applicability give NGS an edge over other techniques, lack of standardization across laboratories and establishment of clinically relevant cutoffs are some issues which need addressal.Other factors influencing interpretation of NGS MRD include possibility of finding certain gene mutations in healthy individuals, and the fact that some genetic abnormalities can persist at low levels while patient is clinically stable can pose interpretation problems while reporting. 45,78Consequently, NGS as a standalone technique is not currently recommended by ELN. 3

Definitions of Response
The recommended definitions of response and relapse based on MRD are summarized in ►Table 3. 3 Timepoints for Testing MRD Specific timepoints for MRD testing have been recommended by ELN for established AML entities including NPM1 mutated AML, CBF AML, and APL. 3 Disease burden in PB at diagnosis can be used as a baseline comparator if blast percentage is more than or equal to 20%.If log reduction is being used as a method to calculate MRD response, both PB and BM samples are required to be processed at baseline.PB MRD testing after two cycles of induction is the recommended first follow-up timepoint for NPM1 and CBF-AML, BM MRD testing at the end of treatment or PC as second timepoint, followed by a total of 2 years' follow-up (every 3 months if using BM or 4-6 weekly if using PB).
For APL, the PC timepoint directly forms the MRD timepoint.Following this, no further follow-up is recommended for low-risk APL if PC MRD is negative.A further follow-up of 2 years is recommended for high-risk APL similar to NPM1 mutated and CBF AML.

Clinical Impact
Numerous studies have highlighted the value of MRD detection in AML in the prognostication of patients using intensive as well as less toxic chemotherapeutic regimes and at various treatment timepoints.MRD positivity has been shown to be associated with higher relapse rates and inferior survival outcomes.A meta-analysis of 81 studies on AML MRD reported by Short et al emphasizes the significance of MRD as a prognostic and surrogate marker for 5-year disease-free survival and overall survival in AML. 80Important literature concerning therapeutic implications of AML MRD has been recently reviewed by Aitken et al. 81 The MRD status also guides in recategorization of a patient at PI timepoint, for example, a PI MRD positivity in a favorable risk patient would newly categorize the patient into intermediate risk, especially seen in NPM1 mutated AML. 79t is crucial to note that relapse can occur in cases who are MRD negative.Conversely, a consistent low level MRD positivity can be associated with a stable clinical course, of note in NPM1 and RUNX1::RUNX1T1. 79The implication of MRD-LL, which would be technically reported as MRD negative, is currently unclear in view of insufficient clinical evidence. 3,79mportantly, the literature pertaining to NGS MRD is widely diverse in context of technical aspects like methods of error correction, DNA inputs, genes included in panel and cutoffs used.Taking all these factors into account, the broad directions concerning clinical intervention in the form of preemptive therapy based on molecular MRD are difficult to be generalized and would require individualized decisions.

Future Directions
In addition to larger studies utilizing the discussed molecular methods for MRD detection, newer methods are also being widely studied and could find a place in clinical practice.The role of circulating cell free DNA by targeted NGS using PB samples has been explored by Short et al 82 and others. 83,84echniques like whole exome and whole genome sequencing can detect new alterations implicated in MRD negative relapse; however, lower coverage obtained in them currently hampers the implementation for deeper sequencing.More recently, single-cell mutational profiling has also found its way to the armamentarium of AML MRD techniques, 15 thus giving a glimpse of rapid upgrades in near future in this field.

Fig. 1
Fig. 1 Overview of molecular aberrations in acute myeloid leukemia (AML) tracked by molecular methods and their major methods of detection along with recommended methods.BM, bone marrow; dPCR, digital PCR; MRD, measurable residual disease; NGS, next-generation sequencing; PB, peripheral blood; PC, post-consolidation; RT-qPCR, real-time quantitative polymerase chain reaction.

Fig. 2
Fig.2Illustration of error correction in next-generation sequencing (A) and overview of error correction techniques applied for measurable residual disease detection in NGS (B).[65][66][67]PCR, polymerase chain reaction; smMIPS, single molecular inversion probe system.

Table 1
Major studies that utilized RT-qPCR for AML MRD detection in CBF-AML, APL, and NPM1 mutated AMLs -PI BM MRD >3 log, BM copies >500 and PB >1000 informative for RUNX1:: RUNX1T1 -PC BM MRD !4 log reduction associated with lower CIR, and presence of >500 RUNX1::RUNX1T1 copies predicted relapse -Presence of >10 copies of CBFB::MYH11 in PB associated with higher CIR Zhu HH 2013 31 As per QinY-ZLiJLZhuHH et al Not mentioned -Positive MRD after 2 nd consolidation discriminated patients with high relapse risk Jourdan E 2013 32 As per EAC criteria Not mentioned -PC MRD (after 2 nd consolidation) could best discriminate patients with high relapse risk Wang Y 2014 33 As per EAC criteria Not mentioned -Post-HSCT monitoring of RUNX1:: RUNX1T1 at 1, 2, 3 months

Table 1 (
3ontinued) in a single assay, unlike the PCR-based methods.This is especially relevant for AML, a clonally heterogeneous disease showing dynamic fluctuations in a patient over the course of therapy.Apart from panel-based sequencing, NGS assays have also been developed for single genes, for example, NPM1 and FLT3-ITD.For the first time, ELN MRD docu-ment in 2021 has reflected upon technical requirements for NGS MRD in AML.3Targeted NGS performed at the diagnostic timepoint can detect somatic mutations (SNVs, indels) in nearly all patients Abbreviations: BM, bone marrow; CBF-AML, core binding factor-acute myeloid leukemia; DFS, disease-free survival; EAC, Europe Against Cancer; HSCT, hematopoietic stem cell transplant; MRD, measurable residual disease; NPM1, nucleophosmin 1; OS, overall survival; PB, peripheral blood; PC, post-consolidation; PFS, progression-free survival; PI, post-induction; RT-qPCR, real-time quantitative polymerase chain reaction.samples

Table 2
Summary of major panel-based NGS-based molecular MRD studies in AML Indian Journal of Medical and Paediatric Oncology Vol.44 No. 6/2023 © 2023.The Author(s).Molecular MRD in AML Harankhedkar, Patkar 571 of AML 43 and hence can be potentially applied for MRD detection in most patients like MFC-based MRD.►Table 2 summarizes major studies utilizing panel-based NGS for AML MRD detection.

Table 3
3efinitions of response based on AML MRD (ELN 2021 recommendations)3Conversion of a negative MRD result to positive by any MRD method or 2. increase in MRD copy numbers !1 log 10 between any 2 positive samples in the cases with CR-MRD-LL monitored by qPCR.3. Above results should be repeated on a second sample and confirmed, BM preferably.