In the past, it has been debated if peripheral blood (PB), rather than bone marrow (BM), could be used for MRD monitoring, regardless of the technique used (MFC or PCR). It is nowadays clear that the scenario is different between B-lineage and T-lineage ALL: in fact, in BCP-ALL, MRD levels tend to be 1–3 logs lower in PB than in BM (21, 22) and that bone marrow assessments cannot be replaced by blood analyses; at variance, in T-ALL, it has been shown—both in children and adults—that PB is a reliable source for MRD monitoring since there are no significant differences with BM and therefore they could be used as an alternative source. Nevertheless, also in T-lineage ALL MRD assessments are normally carried out on BM samples.

This approach takes advantage from the presence of proteic epitopes (antigens) in the nucleus, cytoplasm or surface of the cell, which are sequentially acquired during normal cell development. Antigens are differently expressed by B- and T-lymphoblasts, and their expression is assessed by quantification of the signal emitted by fluorochrome-conjugated-specific monoclonal antibodies (MoAb). The LAIP must be identified at diagnosis, before any therapy in each ALL case, by comparing the marker profile of leukemia cells to reference bone marrow samples, through various combinations of monoclonal antibodies against surface, cytoplasmic, or nuclear leukocyte antigens. A second approach is represented by the so-called “different from normal (DFN)” analysis, which defines leukemic blasts by recognizing immunophenotypic changes with respect to a normal counterpart population (either hematopoietic progenitors of similar lineage and maturational stage) thought the evaluation of antigenic patterns expression (23). This tool has the advantage of studying MRD without the need of a diagnostic immunophenotype, but it requires standardization and needs further implementation.

During the years, antibody-conjugated fluorochromes were developed to increase the number of “colors;” further advances in the field increased the possibility of studying a great number of functionally distinct lymphocyte populations in human PB (24–26). The introduction of violet lasers (405 nm) and of nanocrystals (called quantum dots) or organic polymers capable of conducting electrons (27) led to the current MFC capable of analyzing up to 18 colors in a single cell (28). At present, the most commonly used MFC panels comprise 6–8 MoAb combinations. The refinement of MFC has required a parallel advancement in hardware, software and reagents. Engineering advances in optics and signal processing (digital electronics) are areas of active development (24, 29). Different software packages are now available, including FACSDiva^TM [Becton Dickinson], Kaluza^TM [Beckman Coulter], FlowJo^TM [www.treestar.com], Cytobank^TM [www.cytobank.org], SPICETM [http://exon.niaid.nih.gov/spice/], SamSPECTRAL^TM [R-package], FLAMETM, SPADE^TM [www.cytospade.org], FlowType^TM [R-package], FlowCAPTM [flowcap.- flowsite.org], GemStone^TM [www.vsh.com], Infinicyte^TM [www.infinicyt.com].

Flow cytometry can be successfully applied to the majority of cases (>90%) and can reach a sensitivity of 10⁻³-10⁻⁴ (one leukemic cell out of 1000–10,000 normal cells) (Table 1 and Figure 1) (30). Flow cytometry analysis is quick, can release MRD evaluations in few hours, and is, therefore, also useful to assess the therapeutic response following the first 2 induction weeks (31). However, some limitations exist, such as the fact that the samples must be analyzed shortly after collection to avoid cell death, a problem when shipment is required for the centralized evaluation of MRD referral laboratories. Furthermore, (a) post-induction regeneration of normal lymphoid cells co-expressing some ALL-type antigens can lead to false positive results in B-ALL cases, (b) the bone marrow sample hypocellularity and, in some patients, phenotypic shift can induce erroneous or difficult interpretations (32, 33). The EuroFlow Consortium has optimized and standardized immunostaining protocols for the diagnosis, classification and prognostic subclassification of hematologic malignancies, as well as for the detection of MRD during the clinical follow-up; however, experienced operators are still needed to correctly evaluate MRD results (34, 35).

The advent of RQ-PCR (54) has represented a significant advancement with respect to conventional PCR based methods. However, the measurement of a dynamic process, such as the rate of target amplification, carries some intrinsic fluctuations that cannot be to fully eliminated: (a) non-specific amplification of spurious IG/TR rearrangements are hardly distinguishable from cases positive at a very low level (PNQ) in RQ-PCR, with an intrinsic risk of false positive/negative MRD detections; (b) the use of RQ-PCR can be limited by the lack of sufficient diagnostic material since the method is based on the comparison, for each experiment, with a standard curve based on neoplastic DNA collected at the onset of the disease, and this can limit the possibility of monitoring patients over time.

The novel next generation flow (NGF)-MRD approach takes advantage of innovative tools and procedures recently developed by the EuroFlow Consortium for sample preparation, antibody combinations (including choice of type of antibody and fluorochrome), and identification of B-cell precursor (BCP) pathway in the BM, which allows to define the degree of immunophenotypic deviation of BCP-ALL cells from normal BCP (also in regenerating BM). Also for T-ALL a comparable strategy is used to obtain reliable (evidence-based) antibody combinations, in order to discriminate from various types of normal T cells and other cells with cross-lineage marker expression (55, 56).

NGF-MRD is faster and reproducible, it has a greater applicability (>95%). Moreover, the costs of reagents and assays are estimated to be lower than those of NGS (57). However, it requires fresh material analyzed within 24 h after sampling. Finally, NGF-MRD strategies provide a full insight into the composition of normal cells and aberrant cells, and can help to better characterize ALL cell population changes such as treatment-induced immunophenotypic shifts (58, 59), heterogeneity in the blast cell population with a de-differentiation to immature stem like-cells and aberrations in other lineages.

A similar applicability associated with a significantly increased sensitivity for the novel EuroFlow-NGF approach vs. conventional flow-MRD has been described in multiple myeloma (59); NGF-MRD reaches a sensitivity close to 10⁻⁶, while conventional flow tools reach sensitivities in the range of 10⁻⁴-10⁻⁵ (Table 2 and Figure 3). The greater sensitivity of NGF-MRD is mostly due to the use of standardized approaches, including instrument setting, sample processing with bulk lysis procedure, immunostaining, data acquisition, and data analysis with standardized (even automated) gating strategies for definition of cell populations (59). However, the acquisition of a large number of cells is needed to reach the required sensitivity.

In the forthcoming decade, the new flow technologies will improve applicability and specificity of flow MRD measurements.

The digital PCR technology (ddPCR) (61) based on sample partitioning (mimicking limiting dilution) and Poisson statistics, has the potential to overcome the limitations of RQ-PCR. DdPCR (62, 63) is the third-generation implementation of conventional PCR that allows the quantitation of nucleic acid targets without the need of the calibration curves (64). As reported in several studies (65, 66), based on the dynamic nature of two methods, ddPCR appears more accurate than RQ-PCR since: (i) each sample is partitioned in droplets and each droplet is analyzed individually, (ii) small changes in fluorescence intensity are more readily detected, and (iii) the ratio between target DNA molecules to PCR reagents is substantially higher; this increases its amplification efficiency (67). Finally, the presence of inhibitors negatively affects the RQ-PCR efficiency but not that of ddPCR (68) (Table 2 and Figure 4).

Moreover, RQ-PCR, as an exponential process, is able to greatly amplify even small differences in reaction efficiency, leading to discrepancies in the final results with 68% confidence intervals (http://www.biorad.com/webroot/web/pdf/lsr/literature/Bulletin_6407.pdf).

ddPCR is an endpoint measurement with 95% confidence intervals, as reported by the manufacturer's application guide. The ddPCR technology has been applied to various fields of medical diagnostics, in particular in molecular oncology (62, 64, 65) and in prenatal diagnosis (69, 70). Several reports on the use of ddPCR in hematologic malignancies (71–74) are available in the literature. Three recent articles, including two from our Center, have compared ddPCR to RQ-PCR in adult patients affected by mature lymphoid malignancies and in Ph- ALL (75–77). These studies have established analytical parameters to investigate the applicability of ddPCR for MRD detection and concluded that ddPCR has a sensitivity, accuracy and reproducibility at least comparable to that of RQ-PCR. Regarding MRD evaluations, some discordances have been observed at very low disease levels; in this setting, ddPCR showed a good analytical performance to quantify those low positive samples defined as PNQ by RQ-PCR or to identify the false MRD+ cases. These results have been confirmed from our group in a wider comparative analysis including 175 patients with different lymphoid malignancies (78) (Figure 5). Recently, the clinical significance of ddPCR has been reported in a pediatric cohort of ALL patients (79). The authors showed that among “slow early responder” patients, most relapses occurred in cases with quantifiable ddPCR MRD at day +78, while patients with a negative or PNQ MRD by ddPCR at day +78 had a better outcome, emphasizing that high-risk treatment could be offered only to ddPCR quantifiable cases.

No established guidelines for ddPCR MRD analysis and interpretation have so far been defined. A major standardization effort is underway within the EuroMRD Consortium.

Several groups have shown the value of next-generation sequencing (NGS) technologies for MRD detection in precursor and mature B-cell tumors (80–82) NGS can be used to detect clone-specific IG/TR index sequences; clonal sequences detected at diagnosis can be re-detected and quantified in each follow-up sample. By using universal primers, this method allows to monitor all IG/TR gene rearrangements at the same time, providing a complete picture not only of the residual leukemia but also of the normal immune repertoire (82).

Sensitivity is a critical aspect in MRD detection. Methods allowing a sensitivity higher than 10⁻⁵ (routinely achieved by RQ-PCR) might be of interest to identify very low-level disease. Studies using the NGS platform in ALL and chronic lymphocytic leukemia have demonstrated that a sensitivity level of 10⁻⁶ (81, 83) is achievable when higher amounts of DNA are used (Table 2 and Figure 6). This is reflected in the possibility of detecting early clonal evolution, a relatively frequent occurrence in relapsed ALL (84).

Many authors have reported that NGS appears more specific than RQ-PCR in predicting relapse in ALL patients after induction (82) as well as after allogenic SCT (85). A comparative MRD analysis between RQ-PCR and NGS showed within the Berlin Frankfurt Münster (BFM) trials a change in the stratification risk, mainly due to different interpretations of the two techniques within low-positive samples (82). The NGS quantitative discrimination is always superimposable to the sensitivity, whilst the RQ-PCR quantitative range is usually inferior to the sensitivity threshold leading to cases which are defined as PNQ, as previously described (81). Another comparative analysis, performed at our Center, between RQ-PCR, ddPCR and NGS has documented a more precise prediction of relapse with the new methods compared to the standard technique and a change in the adult ALL risk stratification (86). However, NGS has a substantial intrinsic complexity and involves major costs. Furthermore, at the moment robust and broadly applicable NGS-based MRD standardized protocols are still not available in academic laboratories. The Euroclonality-NGS Consortium is working to standardize guidelines for analysis and data interpretation.

Through the integration of these technologic innovations, we are moving toward a system that can quickly integrate all the information necessary for a more precise evaluation of the response to treatment in ALL patients. ddPCR and NGS appear to be feasible and attractive alternative methods for MRD assessment that can help to classify more precisely cases that RQ-PCR is unable to detect or quantify.

The clinical impact of MRD is now widely accepted and is regarded today as the most important prognostic factor in the state-of-the-art management of ALL. MRD can provide different information, according to the timing in which it is performed (very early during treatment, after induction/consolidation, before and after SCT) and, more recently, it can be refined by the evaluation of additional genomic markers.

It is now widely acknowledged that MRD detection should be carried out with molecular methods, as it occurs in European countries; at variance, in the United States MFC, with few exceptions, is more commonly used, also for the lack of standardized guidelines for molecular analysis and data interpretation.

In the modern era, treatments include combination chemotherapy for the achievement of a complete hematologic and clinical remission (CR), followed by post-remission consolidation therapy with or without SCT, plus an effective central nervous system prophylaxis. Several pediatric and adult ALL study groups have established informative checkpoints in their respective treatment protocols. Among these, the initial MRD response to therapy is a relevant prognostic factor in both childhood and adult ALL (87, 88); indeed, MRD negativity at very early time points during induction therapy correlates with a particularly good outcome both in childhood and adult ALL, and is indicative of a high sensitivity to chemotherapy. In children, it has been shown that MRD negativity achieved as early as at day 15 correlates with an excellent outcome; in adults, the early evaluation of MRD is usually carried out at a later timing (i.e., end of induction, week 4) and it again correlates with better survival rates. Some studies have reported that patients with a very rapid tumor clearance after 2 weeks of therapy have a very good prognosis (31, 89). Along the same line, in a MFC-based trial, high levels of disease at day 14 of treatment (>30%) identified a small subgroup of patients with a particularly poor prognosis and a median event-free survival (EFS) and overall survival (OS) of only 9 and 21 months, respectively; however, MRD evaluation was not significant if evaluated at CR achievement (90). It must be again underlined that in European protocols the use of MFC for MRD evaluation is normally substituted by molecular techniques.

The other time points that hold prognostic significance are the end of induction/early consolidation. Indeed, the first pivotal study on molecular MRD analysis was carried out by the German Multicenter Study Group for Adult ALL (GMALL) on a large cohort of Ph- patients with standard-risk and high-risk features (n = 580 in CR) that showed that the molecular response to standard induction and consolidation treatment was the only significant prognostic factor for remission duration and survival in both risk groups (91). These data have been confirmed by others groups, regardless of the cut-off values, MRD technique, timing of MRD analysis and the target patient population chosen (92–94). The PETHEMA group evaluated the role of MRD (by flow-cytometry, cut-off: 5 × 10⁻⁴) in 326 adult high-risk Ph- ALL and confirmed that the only prognostic factor was represented by MRD persistence after induction and early consolidation (92). The GRAAL conducted an analysis on 955 patients to assess the role of SCT, taking into account also MRD after induction (week 6, cut-off: 10⁻³), and again showed that the role of MRD persistence is not abrogated by transplant procedures and that MRD-negative patients could be spared this approach (93). The Northern Italian Study Group (NILG) performed MRD evaluations starting from the end of induction (week 4) and then at weeks 10, 16, and 22 in an adult patient population to assess the feasibility and efficacy of liposome-encapsulated cytarabine for central nervous system prophylaxis along with the use of lineage-targeted systemic methotrexate blocks (2.5 and 5 g/m² in B- and T-lineage ALL, respectively) and other intensive pediatric-like elements (94). The early MRD response at weeks 4 (end of induction) and 10 had a profound prognostic effect. The relapse risk was very low (17% at 5 years) in the group of week 4 MRD responders and significantly lower (28%) than in non-responders (57%) when week 10 MRD results were examined. Similar results have been preliminary confirmed in the subsequent GIMEMA LAL 1913 (95).

Pivotal studies on the clinical role of MRD in Ph-ALL are summarized in Table 3.

The Ph chromosome, leading to the BCR/ABL1 rearrangement, defines the more frequent high risk ALL subset in adults. It is found in 25–30% of patients and its incidence increases with age progression (101). Ph+ ALL was previously regarded as the subgroup with the worse outcome: the introduction of TK inhibitors (TKI) such as imatinib, dasatinib, and ponatinib (102–106), has led to the achievement of a CR in virtually all patients, to improve disease-free survival (DFS) and OS, and to increase the percentage of patients who can undergo a SCT. A common therapeutic approach for adult Ph+ ALL patients is based on the use of a TKI, with or without systemic chemotherapy for CR induction, followed by consolidation and when possible SCT. As in Ph- ALL, MRD has an important role in the management of this disease. The GIMEMA has clearly shown that the degree of MRD reduction correlates with improved DFS, regardless of the inhibitor used (107). Lee and colleagues showed that also the timing of MRD clearance is important for patients' stratification: in fact, patients who display a very early clearance have a significantly better outcome (108).

Pivotal studies on the clinical role of MRD in Ph+ ALL are summarized in Table 4.

In Ph+ ALL, MRD can also drive therapeutic decisions: indeed, a persistent MRD positivity and/or its reappearance can underlie the presence of resistant mutations, particularly the T315I for which alternative approaches, including novel TKIs (namely ponatinib) and/or therapies based on the combination of TKI with immunotherapeutic strategies, particularly blinatumomab have been evaluated (116).

An open issue in this setting is represented by the cases who are persistently MRD negative; in fact, while in the past SCT was considered mandatory in all Ph+ ALL patients, if clinically fit, it is currently debated if these patients might be spared this procedure, in line with the clinical policy routinely applied to Ph- ALL (102–107, 113, 115).

Finally, there is growing interest in evaluating MRD using more than one marker: in pediatric Ph+ ALL, Hovorkova and colleagues (53) showed that roughly 20% of children have significantly higher BCR/ABL1 levels (assessed by evaluating both genomic DNA and RNA fusion levels) than IG/TR or IKZF1 deletions, indicating that the BCR/ABL1 signal arises from other hemopoietic cells. Along the same line, Cazzaniga et al. (117) showed that MRD positivity after induction and consolidation (evaluated by IG/TR) is strongly associated with relapse. A formal comparison between IG/TR and BCR/ABL1 evaluation showed comparable results in terms of relapse risk: however, concordance between the two techniques was only 69% and, again, BCR/ABL1 levels were significantly higher than IG/TR. In the adult setting, Clappier et al. (118) evaluated IG/TR and BCR/ABL1 levels (using genomic DNA and RNA) and again proved that there is an overall poor concordance between genomic DNA BCR/ABL1 and IG/TR results (correlation coefficient = 0.51), while a good concordance is detected between genomic and RNA BCR/ABL1 levels (correlation coefficient = 0.8). Furthermore, the authors showed that there are two subsets of patients: the first group with concordant MRD results and the second displaying discordant results among IG/TR and BCR/ABL1, with BCR/ABL1 levels being higher. Interestingly, discordant cases harbor more often the p210 protein isoform, and have less frequently IKZF1 deletions, again suggesting that the BCR/ABL1 signal derives from other cells rather than lymphoblasts and that these cases might resemble a “CML-like” subset. At present, however, it is not defined which marker is most suitable for therapeutic decisions.

As mentioned, MRD after induction/early consolidation is the most important decision-making parameter for on allogeneic transplant, a procedure still aggravated by transplant-related mortality and toxicity, observed in about 20% of patients (119, 120). In addition, several studies have analyzed the prognostic impact on outcome of a MRD (+) status at the time of SCT (121, 122) and others have shown the prognostic relevance of pre-transplant MRD assessment in adults (98, 101, 123). In particular, Bassan et al. (98) showed that patients with MRD levels ≥10⁻³ at week 16 and/or week 22 (i.e., after consolidation) had a worse post-transplant outcome with a 6 years relapse incidence of 64% compared to 23% in patients with MRD levels <10⁻³. A recent metanalysis on 21 published reports, including over 2,000 patients, confirms that a pre-transplant positive MRD is a significant negative predictor of relapse-free survival (RFS), event-free survival (EFS), and OS; as expected, a positive MRD prior to transplant was not associated with a higher rate of non-relapse mortality (124). Taken together, these results show that MRD evaluation before transplant is extremely useful for treatment intensification, since we can offer the opportunity to adequately use immunotherapeutic compounds (e.g., blinatumomab, inotuzumab, and in the future possibly CAR-T cells) aimed at obtaining a MRD negative status.

With regard to the post-transplant setting, MRD monitoring has been much less frequently used after SCT because donor chimerism monitoring provides an alternative for early relapse detection; nevertheless, it has been shown that a IG/TR-based MRD assessment allowed an earlier and more specific detection of an impending relapse compared to chimerism analysis, showing that MRD positivity was an independent significant predictor of risk of relapse (125). Another MFC-based trial showed that patients with evidence of MRD after SCT had significantly worse outcomes compared to patients without evidence of MRD (126). In the pediatric context, pre-SCT MRD resulted the only independent prognostic factor in a multivariate analysis and after-SCT MRD is considered a reliable marker for early detection of impending relapses (127, 128).

At relapse, molecular evaluation of IG/TR markers previously used for MRD monitoring might be useful to confirm the persistence of the same clone. However, it has some potential pitfalls, mostly represented by the phenomenon of clonal evolution. The analysis of all molecular markers at first diagnosis and at the time of relapse may reveal a different origin of the predominant clone (84, 129). Indeed, the loss of a molecular marker at relapse, or the expansion of a clonal marker expressed at the subclonal level at diagnosis, is a relatively frequent event: therefore, MRD assessment should be carried out with several targets (84). Finally, it must be reminded the clonal evolution can vary according to the site of relapse (medullary or extramedullary).

The extensive characterization of the genetic bases of ALL is leading to an attempt of combining MRD with other markers: a pivotal study was carried out by the GRAAL group (130) which evaluated 423 young adults with Ph- ALL in first remission (both B- and T-lineage ALL), demonstrating that a higher risk of relapse was associated with MRD persistence and can be further refined by the presence of IKZF1 deletions in B-lineage ALL, and by the absence of NOTCH1/FBXW7 mutation, and/or by the presence of N/K-RAS mutation and/or PTEN gene alteration in T-cell ALL.

Likewise, in childhood ALL, the generation of a risk score based on the combination of MRD at day 33, the presence of IKZF1 intragenic deletion and P2RY8-CRLF2, which proved more discriminative of outcome that MRD evaluation alone, has been recently reported (131).

Novel therapies, such as monoclonal antibodies, bispecific T-cell engagers, or chimeric antigen receptor T cells (CART), are an exciting advancement in the immunotherapeutic treatment of relapse/refractory BCP-ALL. These new therapeutic approaches make MRD an almost perfect therapeutic target, considering that MRD+ patients harbor significantly less leukemic cells and therefore a more manageable clinical profile than cases in hematologic relapse.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.