Clinical use of measurable residual disease in adult ALL: recommendations from a panel of US experts

Acute lymphoblastic leukemia (ALL) in adults has a heterogeneous prognosis that is influenced by both patient-related and disease-related factors.¹ Although these baseline features affect long-term clinical outcomes, response-based factors such as measurable residual disease (MRD) may be the most substantial contributor to the overall risk of relapse. The strong prognostic impact of MRD in ALL was shown in a large meta-analysis, in which the 10-year disease-free survival of adults who achieved MRD negativity was 64% vs only 21% in those who were MRD positive after initial induction and/or consolidation therapy.² In subgroup analyses, the prognostic impact of MRD was consistent across clinical contexts, including ALL subtypes, time points of MRD assessment, and the specific MRD assay used. Although the achievement of MRD negativity is clearly an important clinical end point that predicts for superior long-term outcomes in ALL, when and how to assess MRD and, most importantly, how to use this information to guide treatment decisions are less clear. Several assays are available for the assessment of MRD in ALL, each of which differ in their sensitivity and specificity, as well as in their clinical implications.³ Baseline cytomolecular features and therapeutic context also interact with MRD information to influence prognosis, which further complicates decisions on how to use MRD for treatment decisions. In this review, we discuss the literature on MRD in ALL and provide recommendations on how to use MRD to guide prognostication and therapeutic decision-making in ALL.

A panel of 10 experts in the management of ALL from the United States met on 11 July 2024 to discuss the optimal use of MRD in adults with ALL. Before meeting, 1 author (N.J.S.) performed a systematic literature review of PubMed (1 January 2000 to 1 June 2024) for retrospective and prospective published studies in English on “measurable/minimal residual disease” and “acute lymphoblastic leukemia” and developed 7 clinical questions related to MRD assessment in ALL. The experts reviewed and critically evaluated the clinical data on the prognostic and therapeutic implications of MRD assessment in adults with ALL, with an emphasis on these 7 clinical questions. Recommendations related to MRD monitoring and interpretation and its role in therapeutic decision-making were developed through in-person discussions at the meeting and were refined based on extensive postmeeting feedback until all authors agreed with the final recommendations. Formal grading of the evidence was not performed.

Several MRD assays are available for assessment of ALL MRD, each of which has advantages and disadvantages (Table 1). Multiparameter flow cytometry (MFC) detects MRD through the assessment of leukemia-associated immunophenotypes. MFC can detect MRD with a sensitivity of ∼1 leukemic cell per 10 000 nucleated cells (ie, 1 × 10⁻⁴), although not all clinical laboratories can achieve this level of sensitivity. Patient-specific polymerase chain reaction (PCR) for immunoglobulin (IG) or T-cell receptor (TR) gene rearrangements is commonly used in some countries in which this approach has been sufficiently standardized, but it is not routinely available in the United States because of the requirement for development of patient-specific reagents with this approach.⁴ This assay identifies the unique sequence(s) of IG and/or TR rearrangements arising from V(D)J (variable-diversity-joining) recombination in the diagnostic ALL specimen and then uses patient-specific primers and quantitative PCR probe sets to track these sequences in remission samples, with a sensitivity of ∼1 × 10⁻⁵. Newer, next-generation sequencing (NGS)–based assays have now been developed that track these IG/TR gene rearrangements but can achieve an analytic sensitivity of 1 × 10⁻⁶. In the United Status, the clonoSEQ assay (Adaptive Biotechnologies Inc, Seattle, WA) is an NGS-based MRD assay for detecting and quantifying IG/TR rearrangements and is the only US Food and Drug Administration–cleared MRD assay for lymphoid malignancies, including B-cell ALL, chronic lymphocytic leukemia, and multiple myeloma. Driven largely by the superior sensitivity of this assay (ie, 1 × 10⁻⁶), several studies of patients with B-cell ALL have shown better discrimination for relapse and survival with NGS MRD assessment of leukemia-specific IG/TR sequences (also referred to as “clonotypes”) as compared with other methods, including after frontline therapy and in the peritransplant setting.^5-9 NGS-based MRD is less well-validated in T-cell ALL, and studies assessing its utility in this ALL subtype are needed.

In patients with Philadelphia chromosome (Ph)-positive ALL, reverse transcription (RT) PCR for BCR::ABL1 transcripts has historically been the most common method of MRD assessment. RT-PCR for BCR::ABL1 is superior to MFC in patients with Ph-positive ALL.¹⁰ Nevertheless, RT-PCR for BCR::ABL1 may not be appropriate for clinical decision-making for all patients with Ph-positive ALL, because ∼20% to 25% of patients may have multilineage involvement of BCR::ABL1, with BCR::ABL1 transcripts detected in the nonlymphoblast compartment (eg, myeloid and/or nonmalignant lymphoid cells).^9,11,12 Because of the potential for false-positive MRD results when using RT-PCR for BCR::ABL1 in patients with Ph-positive ALL with multilineage involvement, MRD assays that assess IG/TR gene rearrangements, which are more specific for the leukemic clone and thus more representative of clinically significant MRD, may be more prognostic than MRD assessed by RT-PCR for BCR::ABL1.¹³ In 1 analysis of patients who achieved MRD negativity by NGS for IG/TR, relapse-free survival (RFS) and overall survival (OS) were similar regardless of the presence or absence of detectable BCR::ABL1 transcripts by RT-PCR, suggesting that RT-PCR for BCR::ABL1 did not provide additional prognostic information.⁹ The clinical utility of serial monitoring of RT-PCR for BCR::ABL1 in patients with multilineage Ph-positive ALL in whom monitoring of NGS MRD is also being performed is uncertain.

• In patients with B-cell ALL, NGS-based MRD quantification of leukemia-specific IG/TR clonotypes is preferred for clinical decision-making over MFC, PCR for IG/TR, or RT-PCR for BCR::ABL1 for MRD assessment, because NGS for IG/TR has superior sensitivity and discrimination for relapse than these other assays.

• In Ph-positive ALL, NGS for IG/TR and RT-PCR for BCR::ABL1 are complementary methods of MRD assessment, although NGS for IG/TR may be more suitable for prognostication and most therapeutic decision-making, because this assay has greater specificity for the leukemic clone than does RT-PCR for BCR::ABL1.

• In patients with T-cell ALL, NGS for IG/TR and other MRD assays (eg, MFC) are complementary methods of MRD assessment, as NGS for IG/TR provides superior sensitivity to other MRD assays but is not well-validated in T-cell ALL.

• MFC or PCR-based assays may be helpful in select cases, such as when NGS-based MRD for IG/TR is unavailable (eg, towing to the lack of a trackable clonotype) or is too financially costly, or when a very rapid turnaround time is needed for urgent decision-making. MFC may also be complementary to other MRD assays when assessment of antigen expression is needed for selection of antigen-directed immunotherapy. When used, MFC-based MRD should be performed in a validated laboratory that can achieve a sensitivity of 1 × 10⁻⁴.

• In patients with Ph-positive ALL for whom NGS-based MRD for IG/TR is unavailable, PCR for IG/TR is preferred over MFC or RT-PCR for BCR::ABL1. In cases in which NGS or PCR assays for IG/TR are both unavailable, RT-PCR for BCR::ABL1 is preferred over MFC.

Peripheral blood (PB) monitoring of MRD in ALL offers the advantage of avoiding the need for invasive bone marrow (BM) sampling. However, the disease burden may be 1- to 2-log-fold lower in the PB than in the BM.^14-16 PB monitoring of MRD using PCR for BCR::ABL1 is commonly used in patients with Ph-positive ALL, although few studies have compared MRD testing in the PB and BM. In Ph-positive ALL, BCR::ABL1 transcript levels in the PB are generally at least 1-log-fold lower than those obtained in the BM.¹⁷ 

Several studies have compared PB and BM assessment for NGS MRD for IG/TR.^8,14,18,19 In a retrospective study of 69 patients, a strong correlation in NGS MRD quantification was observed between these 2 compartments (r = 0.87; P < .001).¹⁸ Compared with MRD in the BM, the positive and negative predictive values of PB MRD by NGS for IG/TR were 87% and 90%, respectively. In a large study of 808 children with standard-risk ALL evaluated with NGS MRD, 84% of patients with detectable BM MRD also had detectable PB MRD, with a strong correlation between these 2 sources observed across cytogenetic risk groups.¹⁹ These studies suggest that PB NGS MRD may be a reasonable alternative to BM MRD monitoring. However, because of the lower overall disease burden in the PB, PB monitoring may fail to detect very low levels of MRD and therefore may be more appropriate for patients with a low pretest probability of MRD positivity, particularly those patients in whom NGS MRD negativity in the BM has already been documented.

Some patients with ALL may have extramedullary involvement, which may be more amenable PB MRD monitoring. Although the clinical utility of this approach in extramedullary ALL has not been systematically evaluated, PB MRD monitoring has been shown to have prognostic importance in lymphomas and may be similarly useful in cases of ALL with extramedullary involvement.²⁰ 

• For patients with BM involvement, BM MRD monitoring is recommended at least until MRD negativity is documented. After achievement of MRD negativity using a high-quality BM sample, MRD monitoring of the PB (either alone or alternating with periodic BM assessments), ideally using a high sensitivity NGS-based MRD assay, can be considered to reduce the frequency of invasive BM examinations.

• For patients with extramedullary-only disease (eg, B-cell or T-cell lymphoblastic lymphoma), PB MRD monitoring may be clinically useful, although few data are available supporting its use in this setting.

In patients undergoing frontline ALL therapy, the timing at which MRD negativity is achieved plays an important role in risk stratification. In Ph-negative ALL, achievement of MRD negativity after induction is associated with superior outcomes, regardless of the MRD assay used.^5,21-24 In 1 study of patients treated predominantly with hyper-CVAD (hyperfractionated cyclophosphamide, vincristine, adriamycin, and dexamethasone), patients who achieved MRD negativity by MFC after induction had a 3-year OS of 76%, whereas those who achieved MRD negativity later in consolidation had a 3-year OS of 58% (P = .001).²¹ In the CALBG 10403 study, patients who achieved MRD negativity by PCR for IG/TR after induction had a 3-year disease-free survival rate of 85% (vs 54% for those who were MRD positive), highlighting the favorable prognostic impact of early MRD negativity in ALL.²⁴ Using more sensitive MRD assays may better identify patients who have very low risk of relapse. In a study of adult patients who received a frontline hyper-CVAD or mini hyper-CVD–based regimen who underwent NGS MRD for IG/TR with a sensitivity of 10⁻⁶, none of the patients who achieved MRD negativity after induction relapsed, and their 5-year OS rate was 90%. Achievement of MRD negativity at later time points was less prognostic.⁵ 

In patients with Ph-positive ALL, achievement of complete molecular response (CMR) by PCR for BCR::ABL1 (ie, transcript level of <0.01) after ∼3 months of therapy with chemotherapy plus a BCR::ABL1 tyrosine kinase inhibitor is prognostic for relapse and survival outcomes.^25,26 In 1 study of patients with Ph-positive ALL undergoing frontline therapy who did not undergo allogeneic hematopoietic stem cell transplant (HSCT), the 4-year OS rate for patients who achieved CMR after 3 months was 66% vs 36% for those with inferior responses (P = .009).²⁵ CMR at 3 months was the only variable prognostic for OS on multivariate analysis. Achievement of 3-month CMR may also identify patients with Ph-positive ALL who may not benefit from allogeneic HSCT (discussed in more detail later).^26,27 

The appropriate duration for monitoring MRD in patients with ALL undergoing frontline therapy is not well established. Most hematologic relapses occur within 3 years of diagnosis, although later relapses may be observed in some patients, particularly those with Ph-positive ALL.²⁸ In patients who achieve initial MRD negativity after frontline therapy, the likelihood of detecting recurrent MRD during routine surveillance decreases with the duration of remission.

For patients undergoing allogeneic HSCT, both pretransplant and posttransplant MRD have important prognostic implications.^7,29-31 In a meta-analysis of 21 studies, detectable MRD by MFC or PCR before HSCT was associated with significantly worse RFS and OS.³² The quantity of detectable MRD before HSCT also has prognostic implications. In 1 study evaluating the impact of pre-HSCT NGS MRD levels, the cumulative incidence of relapse ranged from ∼10% in patients with undetectable pre-HSCT MRD, to 30% with MRD between 1 × 10⁻⁶ and 1 × 10⁻³, to >50% with MRD of >1 × 10⁻³.⁷ Detectable MRD at any level after HSCT was also associated with a dramatic increase in risk of relapse (hazard ration, 6.31; 95% confidence interval, 4.56-8.74; P < .0001). Although data are scant to support the optimal duration of MRD monitoring after HSCT, >75% of posttransplant relapses may occur within the first year after allogeneic HSCT, making this period the most critical for MRD monitoring.³³ 

MRD is also prognostic after chimeric antigen receptor (CAR) T-cell therapy.^29,34-38 In 1 study of children and young adults who received tisagenlecleucel, detectable MRD at any level by NGS for IG/TR independently predicted for worse event-free survival (EFS) and OS.²⁹ The median EFS for patients with detectable MRD 3 months after CAR T-cell therapy was only 5.8 months vs not estimable in those who were NGS MRD negative (P < .0001). In patients with detectable MRD by MFC, MRD detectable by NGS at a level ≥10⁻⁶, and MRD detectable by NGS at a level <10⁻⁶, the median times to morphological relapse were 52, 70, and 168 days, respectively, suggesting that close monitoring with high-sensitivity NGS-based MRD may provide sufficient time for clinical interventions before overt relapse.

Achievement of MRD negativity is associated with superior outcomes in patients with relapsed/refractory ALL undergoing other salvage therapies.^39-44 In an analysis of patients with relapsed/refractory ALL receiving blinatumomab or inotuzumab ozogamicin–based therapies, MRD assessed by MFC was prognostic for patients in first salvage but not for those in second salvage, for whom outcomes were dismal regardless of MRD response (2-year EFS of <10%).⁴³ Outcomes were best for patients who achieved MRD negativity with first salvage therapy and who underwent subsequent allogeneic HSCT, for which the 2-year OS rate was 65%.

• Patients undergoing frontline ALL therapy should undergo MRD assessment after induction and then at least every 3 to 4 months for at least the first 2 to 3 years. More frequent MRD assessment should be considered for patients who remain MRD-positive after induction (eg, every 1-2 months until MRD negativity is documented) and for patients with adverse-risk disease features who are at relatively higher risk of relapse.

• For patients undergoing allogeneic HSCT or CAR T-cell therapy, MRD assessment should be performed before HSCT or CAR T-cell therapy (for patients in hematologic remission) and approximately every 2 to 3 months for at least the first year after HSCT or CAR T-cell therapy.

• When determining the duration of MRD monitoring for an individual patient, the overall risk of relapse, the availability of a high-sensitivity MRD assay for PB monitoring as a potential alternative to BM assessment, and the cost of MRD monitoring should all be considered.

Allogeneic HSCT may at least partially overcome the poor prognosis associated with high-risk cytomolecular features or MRD-positive disease in patients undergoing frontline ALL therapy.⁴⁵ Although the role of allogeneic HSCT may be diminishing with the development of more effective immunotherapy-based approaches in ALL, HSCT may still be beneficial for some patients at very high risk of relapse with standard therapy.^45,46 

HSCT decisions in patients with Ph-negative ALL may be guided by both baseline cytomolecular risk and by MRD information. Although published studies have varied widely in their definition of “high-risk ALL,” consensus recommendations such as those from the National Comprehensive Cancer Network provide guidance, which may be helpful in clinical decision-making.⁴⁷ For patients with standard-risk ALL, routine allogeneic HSCT does not improve outcomes.^45,48,49 However, allogeneic HSCT may be appropriate for some patients without a high-risk baseline cytomolecular feature but with persistent MRD or MRD relapse during frontline therapy. In patients with MRD-positive B-cell ALL, blinatumomab achieves high rates of MRD negativity.⁵⁰ Whether consolidative allogeneic HSCT is beneficial in patients with standard-risk ALL (based on cytomolecular features) who achieve MRD negativity after blinatumomab is unclear, because some patients who achieved MRD negativity after blinatumomab can have durable remissions without HSCT.⁵¹ 

In patients with Ph-positive ALL, 3-month CMR is associated with superior OS and may identify patients who do not benefit from allogeneic HSCT in first remission.^25,26 In a retrospective study of 230 patients treated across 5 transplant centers and who achieved 3-month CMR with frontline therapy, allogeneic HSCT was not associated with improved RFS or OS on multivariate analysis, suggesting that HSCT may be safely deferred in patients who achieve this MRD end point.²⁶ Although HSCT reduced the cumulative incidence of relapse, HSCT was associated with significant nonrelapse mortality, which negated any potential benefit on survival outcomes.

An important clinical question is whether favorable MRD kinetics can overcome the poor prognosis associated with high-risk cytomolecular features. Early achievement of deep MRD negativity of <10⁻⁶ in patients undergoing frontline therapy is associated with a very low risk of relapse in several studies.^5,6,52-54 It is worth noting, however, that the number of patients in these studies with high-risk cytomolecular features (eg, Ph-like ALL, low hypodiploidy/near triploidy, complex cytogenetics, KMT2A rearranged, TP53 mutated, etc) who achieved early MRD negativity at this cutoff is small. Larger studies are needed to confidently determine whether allogeneic HSCT can be safely deferred in these patients.

• For patients with standard-risk ALL (including those with Ph-positive ALL without an IKZF1-plus genotype) who achieve MRD negativity by NGS-based MRD for IG/TR and/or by PCR for BCR::ABL1 (for Ph-positive ALL) within ∼3 months of initiation of frontline therapy (with or without the use of blinatumomab), routine allogeneic HSCT is not recommended.

• For patients with high-risk ALL who achieve MRD negativity at 1 × 10⁻⁶, there are insufficient data to recommend either for or against allogeneic HSCT, and this decision should be individualized, based on patient preferences, availability of alternative therapies, and risk of transplant-related mortality. For high-risk patients who do not undergo allogeneic HSCT, serial monitoring of MRD should be performed.

• Patients who remain MRD positive after a frontline therapy that incorporates blinatumomab have poor outcomes and may benefit from HSCT in first remission. When feasible, strategies to eradicate MRD before HSCT should be considered, as post-HSCT outcomes in patients with MRD-positive ALL are poor.

Achievement of MRD negativity is also prognostic in the relapsed/refractory setting, as has been shown in several analyses in the contemporary era using immune-based approaches.^39-44 Using MFC-based MRD, the benefit of MRD negativity is greatest for patients in first salvage, compared with those undergoing later lines of therapy.⁴³ Given the poor outcomes of patients who undergo allogeneic HSCT with detectable MRD, assessment of MRD may identify patients with relapsed/refractory ALL who may benefit from alternative therapies (eg, CAR T-cell therapy) rather than immediate HSCT.³² This approach may be more feasible in patients with B-cell ALL than in T-cell ALL, for whom there are fewer effective salvage therapies capable of achieving MRD negativity.⁵⁵ 

Historically, all fit patients with relapsed/refractory ALL who achieved remission with salvage chemotherapy were recommended to undergo allogeneic HSCT as the only chance of cure; however, this calculus may be changing with the development of more effective immune-based salvage regimens.^45,46 For example, in an ongoing study of mini-hyper-CVD, inotuzumab ozogamicin, and blinatumomab in relapsed/refractory B-cell ALL, long-term survival was identical for patients who underwent allogeneic HSCT consolidation vs those who did not (3-year OS rate 54% for both groups).⁴⁴ These data suggest that some patients with relapsed/refractory ALL can have durable remissions even without HSCT consolidation.

For patients undergoing CD19 CAR T-cell therapy, the role of consolidative HSCT is debated.⁵⁶ NGS MRD for IG/TR has been shown to be highly prognostic for post-CAR outcomes and may help to guide this decision. In an analysis of children and young adults with relapsed/refractory B-cell ALL (median of 3 prior lines of therapy, and 53% of whom had undergone prior HSCT) who received tisagenlecleucel, MRD detected at any level by 3 months after CAR T-cell infusion was associated a hazard ratio for relapse of 12 (95% confidence interval, 2.87-50; P < .001) on multivariate analysis.⁸ Importantly, no patient with detectable MRD at 3 months after CAR T-cell therapy had long-term survival without subsequent intervention. In contrast, the median EFS and OS was not reached for patients who achieved NGS MRD negativity at a sensitivity of 1 × 10⁻⁶ at 3 months. Given the poor outcomes of patients with persistent MRD after CAR T-cells, allogeneic HSCT (or other MRD-directed therapies) should be considered for these patients, whereas the role of consolidative HSCT for MRD-negative patients is less clear.

• Patients with relapsed/refractory ALL should be consolidated with allogeneic HSCT or CAR T-cell therapy (preferably in the context of a clinical trial) regardless of the MRD response to salvage therapy. Eradication of MRD before these consolidative approaches is associated with superior outcomes, albeit not feasible in all patients.

• Patients in whom MRD is detected after CAR T-cell therapy should be referred for allogeneic HSCT or investigational MRD-directed therapies, if eligible.

Several effective nontransplant therapies are currently available for the treatment of MRD-positive B-cell ALL. Blinatumomab was studied in a phase 2 trial in patients with B-cell ALL and persistent or recurrent MRD at a level of 1 × 10⁻³.⁵⁰ Overall, 78% of patients achieved a complete MRD response (defined as MRD <1 × 10⁻⁴). Complete MRD responders had superior RFS and OS compared with MRD nonresponders. Other studies have also suggested favorable long-term outcomes with blinatumomab in patients with lower levels of MRD (eg, 1 × 10⁻³ to 1 × 10⁻⁴).⁵⁷ Notably, the use of blinatumomab for MRD-positive ALL may be less relevant in the modern era, because blinatumomab is now indicated for all patients with newly diagnosed Ph-negative ALL, regardless of MRD response, based on positive OS data from the randomized phase 3 ECOG1910 study showing benefit of upfront blinatumomab in patients achieving MRD negativity at a threshold of 1 × 10⁻⁴.⁵⁸ 

Although blinatumomab has the highest quality data as an MRD-directed therapy, inotuzumab ozogamicin and CD19 CAR T-cells may also be effective in this setting. In a phase 2 study, low-dose, fractionated inotuzumab ozogamicin was evaluated in 26 patients with B-cell ALL and detectable MRD of ≥0.01%.⁵⁹ The MRD negativity rate (defined as undetectable MRD by MFC and PCR for BCR::ABL1 with a sensitivity of 1 × 10⁻⁴) was 69%, which translated 2-year RFS and OS rates of 54% and 60%, respectively. Although prospective studies of CAR T-cell therapy for MRD-positive B-cell ALL are lacking, in a retrospective study of commercial tisagenlecleucel in children or younger adults with relapsed/refractory B-cell ALL, 40 of 41 patients (98%) with low-disease burden (defined as MRD-positive disease and/or central nervous system 1-2 disease) responded.⁶⁰ Similarly high rates of MRD-negativity in patients with MRD-only disease have also been observed in other studies with investigational CD19 CAR T-cell products.³⁸ 

• Patients with persistently detectable MRD despite appropriate therapy or who experience MRD relapse should receive MRD-directed therapy. Intervention is recommended regardless of the MRD level, if there is high confidence that reported finding represents true MRD.

• Blinatumomab is recommended for most patients with B-cell ALL who have MRD persistence or relapse and do not have prior exposure to a CD19-targeted therapy.

• Inotuzumab ozogamicin, CD19 CAR T-cell therapy, or investigational MRD-directed therapies may also be considered for suitable patients, ideally in the context of a clinical trial.

When using NGS MRD for IG/TR to make treatment decisions around therapeutic interventions, it is imperative that the MRD results are interpreted correctly. Common diagnostic or interpretative challenges with NGS-based MRD assays are shown in Table 2. For MRD-negative samples, the cellular input will affect the confidence in this result. For example, for the clonoSEQ assay, an input of at least 1.9 million nucleated cells is required to provide 95% confidence that a sample is MRD negative with a sensitivity of 1 × 10⁻⁶. A negative MRD assessment in a suboptimal sample should be interpreted with caution and a shorter interval before the next assessment should be considered, particularly in higher-risk patients.

A common diagnostic conundrum is the interpretation of reports with MRD reported below the limit of detection (LOD), which could represent either very-low-level clinically significant MRD or “background” nonmalignant sequences without clinical implications. Reports with low-level MRD positivity should be interpreted with caution to avoid unnecessarily intervening on a clinically insignificant background sequence, particularly when the results do not resonate with the clinical picture (eg, a patient in long-term remission with a 1-time result showing possible MRD below the LOD). In interpreting these reports, the uniqueness of the detectable sequence should be considered, which can help to make this distinction. Immunoglobulin heavy-chain (IGH) sequences are (on average) more unique than immunoglobulin κ light chain (IGK)/immunoglobulin λ light chain (IGL) sequences and have been found to be more prognostic than IGK/IGL in some analyses.^7,61,62 However, it is important to note that relatively unique IGK/IGL sequences or relatively nonunique IGH sequences may also be observed. To determine the uniqueness of a given trackable sequence, the LOD and limit of quantification (LOQ) are provided in NGS MRD reports. Sequences with lower LOD/LOQ are generally more unique and, therefore, when present, are more likely to represent clinically meaningful MRD than are sequences with higher LOD/LOQ. When multiple sequences are trackable, more weight should be placed on results from sequences with the lowest LOD/LOQ. However, when only 1 sequence is trackable, it may be impossible to make this determination with high confidence. In these cases, very close sequential MRD monitoring to evaluate for trends in MRD may be helpful in distinguishing true MRD, which is more likely to rise or fall (depending on the clinical setting), from background nonmalignant sequences, which are more likely to be relatively static over time.

For patients with T-cell ALL for whom only TR clonotypes are being tracked, distinguishing clinically significant MRD from “background” can be more challenging, because the LOD/LOQ of these sequences were historically not reported. TR-β clonotypes are generally more reliable MRD markers than TR-γ clonotypes, which should be considered when interpreting these reports.⁶² Recently, the clonoSEQ T-cell assay was updated to combine analyses of TR-β and TR-γ rearrangements into a single report and to also provide a uniqueness score for each sequence. Although clinical validation of the T-cell NGS MRD assay is still needed, these changes are anticipated to help with MRD interpretation in patients with trackable TR clonotypes.

• When using an NGS-based MRD assay for IG/TR, distinguishing clinically relevant MRD from low-level, nonmalignant background sequences is imperative before initiating MRD-directed therapy. For low-level MRD that is below the LOD/LOQ, careful interpretation of the results is needed to confirm that this represents true MRD, and expert consultation should be requested in uncertain cases. Repeat testing to evaluate dynamics of the clonotype may be helpful in distinguishing true MRD from clinically insignificant background sequences.

• Caution should be used in interpreting low-level MRD in T-cell ALL being tracked by TR sequences, because it is particularly challenging to distinguish clinically significant MRD from background in this population.

• In equivocal cases, the risks and benefits of MRD-directed therapy should be weighed before initiating treatment. For cases in which observation is recommended because of lack of confidence that the low-level detected sequence represents true MRD, very close serial monitoring of the sequence dynamics should be performed, and MRD-directed therapy should be initiated in the case of a clear trend of rising MRD levels.

When making treatment decisions based on MRD results, it is imperative to remember that the prognostic implications of MRD are context dependent, influenced by the assay used and its sensitivity/specificity, the timing of assessment, and disease-related features (eg, cytomolecular risk), among other variables. These factors, as well as the availability of effective drugs, which is greater for B-cell ALL than for T-cell ALL, all influence the therapeutic implications of MRD. NGS-based MRD assays that can achieve a sensitivity of 1 × 10⁻⁶ are also changing our calculation of relapse risk in ALL, allowing us to identify patients with early and deep MRD responses who can have excellent outcomes without allogeneic HSCT. Conversely, the presence of even very low levels of MRD (eg, 10⁻⁴ to 10⁻⁶) confers an unacceptably high risk of relapse, and these patients may benefit from early MRD-directed therapeutic intervention. MRD assessment is therefore a fundamental component of ALL treatment algorithms, and its role will likely further expand as more data are generated using newer, high-sensitivity MRD assays across clinical contexts. Studies evaluating the dynamics of MRD with novel ALL therapies in both the frontline and relapsed/refractory settings, the impact of these dynamics on long-term clinical outcomes, and the role of MRD-guided therapeutic interventions even for very low levels of MRD should be major research priorities and hopefully will lead to higher rates of cure for patients with ALL.

This research is supported, in part, by the MD Anderson Cancer Center Leukemia SPORE (Specialized Program of Research Excellence) CA100632, and the National Institutes of Health/National Cancer Institute Cancer Center support grant P30 CA016672. Adaptive Biotechnologies funded an advisory board to discuss measurable residual disease in acute lymphoblastic leukemia, out of which a decision was made by the participants to generate these recommendations. Adaptive Biotechnologies provided funding for the time spent drafting the manuscript, but had no role in the decision to develop these recommendations nor in final recommendations contained herein.

Contribution: N.J.S. wrote the first draft of the manuscript; and all other authors critically reviewed and edited the manuscript and approved the final version of the manuscript.

Conflict-of-interest disclosure: N.J.S. has received consulting fees from Pfizer Inc, GlaxoSmithKline, Nkarta, Autolus, Adaptive Biotechnologies, and Sanofi; received research funding from Takeda Oncology, Astellas Pharma Inc, Xencor, GlaxoSmithKline, NextCure, Ascentage, and Novartis; and has received honoraria from Adaptive Biotechnologies, Amgen, Takeda, Pfizer Inc, Astellas Pharma Inc, and Sanofi. I.A. has received consultant fees from Kite, Jazz, Syndax, Wugen, AbbVie, Takeda, Amgen, Pfizer, and Adaptive; and has received research support from MacrGenics, AbbVie, Jazz. M.K. has received consulting fees from Syndax, Novartis, Servier, AbbVie, Menarini-Stemline Therapeutics, Adaptive, Dark Blue Therapeutics, MEI Pharma, Legend Biotech, Sanofi Aventis, Auxenion GmbH, Vincerx, Curis, Intellisphere, Janssen, and Servier; has received research funding from Klondike, AbbVie, and Janssen. J.L. has received consulting fees from Adaptive Biotechnologies, Autolus, Amgen, Kite, Pfizer, and Takeda; and has received research funding from AbbVie, and honoraria from Adaptive Biotechnologies. A.C.L. has received consulting fees from Amgen, Actinium, Bristol Meyers Squibb, Pfizer, Sanofi, and Takeda; and has received research funding from Amgen, Astellas, Autolus, Sanofi, Kite/Gilead, and Talaris. W.S. has received consultant fees from Adaptive, Newave, and Servier; and has received research support from Kura. E.J. has received research grants and consultancy fees from AbbVie, Adaptive Biotechnologies, Amgen, Ascentage, Astex, Bristol Meyers Squibb, Genentech, Hikma, Jazz, Johnson and Johnson, Novartis, Kite, Pfizer, Takeda, and Wugen. The remaining authors declare no competing financial interests.

Correspondence: Nicholas J. Short, Department of Leukemia, The University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, Unit 428, Houston, TX 77030; email: nshort@mdanderson.org.