Introduction

The prognosis of children and adolescents with acute lymphoblastic leukemia (ALL) has dramatically improved in the last half century, with current 5-year event-free survival (EFS) rates exceeding 80%, according to the most recently reported data [1]. This remarkable success has been driven both by the use of complex multiagent chemotherapy regimens and by recognition of clinical, biological, and treatment response characteristics that identify children at higher probability of treatment failure, allowing the administration of risk-adapted therapy. However, there are still subsets of children with ALL in whom the probability of EFS with chemotherapy alone remains unsatisfactory, because of the high chance of disease recurrence. In these very-high-risk patients, as well as in many relapsed patients, allogeneic hematopoietic stem cell transplantation (HSCT) continues to play a major curative role through the combination of intense preparative radiotherapy/chemotherapy and graft-versus-leukemia (GvL) effect.

Defining the appropriate use of HSCT in pediatric ALL is a dynamic process, requiring constant and careful assessment of the likelihood of cure with chemotherapy to identify the subset of children for whom transplant offers a better treatment option. Indeed, advances in chemotherapy schemes continue to occur in parallel with important changes in transplantation techniques, donor availability, and supportive care measures. In addition, the exponential growth of genomic medicine and immunotherapy approaches in the last decade is dramatically changing the treatment perspective of ALL. On the one hand, genome sequencing studies have led to better molecular characterization of pediatric leukemias, this translating into improved risk stratification and identification of targetable genetic lesions [2, 3]. On the other one, impressive clinical results obtained with immunotherapeutic approaches, such as bispecific T cell engagers (BiTEs), targeted immunotoxins, and chimeric antigen receptor (CAR) T cells, have substantially revolutionized the treatment of high-risk B cell precursor (BCP) ALL [4,5,6]. In light of these considerations, the role of HSCT in childhood ALL is likely to continue to change as continuous advances are made in biology, chemotherapy, immunotherapy, and transplantation in future years.

In this review, we will define the current best and promising future practices of HSCT for pediatric ALL by discussing (1) current indications for HSCT, (2) type of donor and conditioning regimen to be used in this setting, and (3) novel immunotherapeutic approaches that may be employed before or after HSCT to improve leukemia-free survival (LFS).

To Whom and When Should Allogeneic HSCT Be Offered in Pediatric ALL?

Typically, only a minority of children with ALL in first complete remission (CR1) are considered for HSCT. Most study groups define these patients as having an estimated EFS probability lower than 50%, according to specific biologic features and response to induction treatment. Primary induction failure (PIF) is rarely seen in pediatric ALL (< 2% of patients), but continues to represent a strong indication for HSCT in CR1, being one of the most unfavorable prognostic factors. A large retrospective study conducted by Schrappe et al. reported outcomes for pediatric patients with ALL experiencing PIF [7], the 10-year survival rate being only 32%. Patients older than 10 years, those with T cell phenotype, or those with an M3 marrow (i.e., > 25% blasts) at the end of induction were considered at particular risk. A significant survival advantage was observed in children older than 6 years receiving HSCT from an HLA-identical sibling compared with chemotherapy (59% compared to 39% of those not transplanted) and in T-ALL after any HSCT (5-year EFS of 40–45% versus 26% for chemotherapy) [7].

Minimal residual disease (MRD) measured at selected time points is currently the single most powerful prognostic factor in childhood ALL [8•]. It may be measured through flow cytometry or through detection of specific rearrangements of immune receptor genes IgH/TCR (heavy immunoglobulin/T cell receptor) [9]. The Italian Association of Pediatric Hematology-Oncology (AIEOP)-Berlin-Frankfurt-Muenster (BFM) 2000 study used prospective molecular (i.e., PCR-measured) MRD evaluation at day 33 (end of induction (EOI), timepoint 1 (TP1)) and day 78 (end of consolidation (EOC), timepoint 2 (TP2)) for treatment stratification in 3184 BCP-ALL patients. In this study, MRD predicted outcome more precisely than either genetic factors, white blood cell (WBC) count or prednisone response. BCP-ALL patients with MRD < 0.01 % at both TPs (standard risk (SR) patients) had a 5-year EFS of 92%, compared with 50% if MRD was > 0.1% at TP2 (high risk (HR)) on day 78 and 78% in intermediate-risk (IR) patients [10•]. In the same study, TP2-MRD was found to be the best predictor of outcome also in T-ALL. Indeed, HR patients had a much worse outcome, with a 7-year EFS of 50%, regardless of other factors, as compared with other risk groups (91% for SR, 81% for IR; P < 0.001) [11]. On the basis of such data, the successor AIEOP-BFM ALL 2009 study used a combination of PCR-MRD response at days 33 and 78, genetic features, and prednisone response to allocate BCP-ALL patients to different risk groups and formulate indications for performing HSCT (Table 1).

Table 1 Indications to HSCT in CR1 in the BFM-AIEOP 2009 protocol
Full size table

Children with severe hypodiploidy (< 44 chromosomes) have a particularly poor outcome progressively worsening with the decrease of chromosome number. A retrospective analysis on 139 hypodiploid ALL patients with < 45 chromosomes, treated by ten different national ALL study groups between 1986 and 1996, reported an 8-year EFS of 38.5% and overall survival (OS) of 49.8%. Patients with fewer than 44 chromosomes fared significantly worse than those with 44 chromosomes (EFS, 30% versus 52%; P = 0.01; OS, 37% versus 69%; P = 0.017) [12]. However, controversial results have been recently reported for this subset, making indications for HSCT in CR1 less clear. Indeed, data from a small series of patients treated at single institution suggest that children with hypodiploid ALL and negative MRD at the end of induction may be cured through risk-directed chemotherapy alone [13]. Another report by the Children’s Oncology Group (COG) observed that, while chemotherapy alone results were still suboptimal in hypodiploid ALL, HSCT in CR1 did not confer a survival benefit, thus indicating the need for alternative therapeutic strategies in this subset of patients [14].

In addition to aneuploidy, a number of chromosomal translocations or structural aberrations have been demonstrated to have a prognostic impact in childhood BCP-ALL. Rearrangements involving the MLL gene on chromosome 11q23 are observed in up to 80% of infants ALL, while they are more rarely detected in children older than 1 year [15]. The majority of infants with MLL-rearranged ALL are characterized by a high white blood cell count, together with a pro-B cell phenotype of leukemic cells, which lack CD10 expression. The prognosis of these patients is relatively poor with only half of cases being cured with conventional chemotherapy [16, 17]. In an attempt to improve the outcome of infant ALL, a number of collaborative groups have investigated the use of HSCT in CR1. In a COG study, EFS for HSCT and chemotherapy alone were identical (49%; P = 0.6) [18], and the authors concluded that routine HSCT in CR1 may not be required for infants. However, that analysis included both lower risk and very-high-risk infants (younger children with high WBC), this mixture potentially diluting the impact of HSCT in the highest risk subgroup. Indeed, the Interfant group found that HSCT offered an advantage for a subgroup of infants with MLL-rearranged ALL with two additional unfavorable prognostic features: age less than 6 months and either poor response to steroids or leukocytes ≥ 300 × 106/μL at diagnosis [19•] Many cooperative groups consider patients older than 1 year with MLL-AF4 BCP-ALL candidates to receive an allograft in CR1.

Intrachromosomal amplification of chromosome 21 is a recurrent lesion found in up to 3% cases of pediatric ALL and has been associated with increased risk of relapse when treated with standard therapy [20]. Although intensive therapy including transplantation has led to improved survival in studies from the UK, whether HSCT has to be offered to patients who become MRD negative early in treatment still remains unclear [21].

BCP-ALL harboring the translocation t(17;19), which results in the TCF3-HLF fusion gene, represents a rare subset (< 1% of all BCP-ALL cases) characterized by low sensitivity to conventional chemotherapy and a dismal prognosis. Pre-clinical data suggests that these patients may be particularly sensitive to treatment with BCL-2 inhibition [22]. Moreover, recent observations documenting high expression of CD19 on leukemia cell surface suggest that immunotherapy with blinatumomab is effective in lowering the leukemia burden in this high-risk subset of ALL [23]. Once achieved CR, early consideration of HSCT for these patients is warranted.

Patients with T-ALL and poor MRD clearance at TP2 are considered candidate to receive an allograft in CR1. Indeed, in contrast with BCP-ALL, leukemia T cell blasts so far cannot benefit from approaches of immunotherapy, this contributing to the lower chance of rescue when a relapse occurs. This observation corroborates the choice of offering an allograft in CR1 to those patients that, mainly because of poor MRD clearance, are predicted to have a chance of EFS in the order of 50% or less.

Allogeneic HSCT is used much more commonly after ALL relapse (i.e., in ≥ 50% of patients) than during in CR1 [24•]. Protocols for treatment of relapsed ALL are based on patient stratification according to the time of relapse, the site of relapse, and the immunological lineage of ALL. Isolated BM relapse carries the worst prognosis, isolated central nervous system (CNS), and testicular or other extramedullary sites of relapse having a better prognosis, with combined BM and extramedullary relapse being associated with an intermediate prognosis [25,26,27]. Moreover, children with T cell ALL BM relapses have a much worse prognosis than BCP-ALL, irrespective of the time elapsing between diagnosis and recurrence [28].

More recently, persistence of MRD after induction/consolidation therapy (i.e., after 5 and 12–13 weeks from the beginning of treatment for relapse) has been shown to influence prognosis in children with relapsed ALL [29,30,31,32].

Based on the above observations, most recent relapse protocols have integrated MRD evaluation in the decision process for treatment stratification and indication to HSCT [29, 32, 33]. Notably, none of the classifications in use considers other risk factors, such as genetic abnormalities, this representing a relevant limitation. Indeed, it has been shown that, in intermediate-risk relapsed patients with low MRD (currently considered to have a good prognosis), deletion of IKZF1 and alteration of TP53 identify patients with significantly worse outcome [34]. Likewise, patients with relapsed BCP-ALL and either t(1;19) or hypodiploidy or MLL rearrangement should all be offered HSCT, since their probability of definitive cure without transplantation is low.

According to recommendations of the worldwide largest International Study for Children and Adolescents with Relapsed ALL (IntReALL), HSCT should be offered to any child with either HR features (very early isolated extramedullary relapse of BCP-ALL or T-ALL, early isolated or any very early bone marrow relapse of BCP-ALL, any bone marrow relapse of T-ALL) or poor clearance of blast after induction therapy for reaching a CR2 (i.e., high levels of MRD) (Table 2). Indeed, these patients (accounting for approximately two thirds of the overall population of children who relapse) have a probability of survival of 30% without HSCT [24•]. All patients who have experienced two or more relapses are candidates to an allograft, irrespectively of the type of donor available.

Table 2 Indications to HSCT for relapsed ALL in the IntReALL 2010 protocol
Full size table

Which Is the Best Donor to Be Used in Pediatric ALL?

Historically, an HLA-identical related donor (or matched family donor (MFD)) was considered the best donor for patients in need of an allograft. However, given the theoretical 25% probability of finding such a donor [35], use of alternative donors (e.g., unrelated donors (UD)) has been investigated. Locatelli et al. demonstrated in 2002 that after the introduction of high-resolution molecular typing for both class I and II HLA loci, EFS of UD-HSCT in children with ALL in CR2 improved, mainly due to a decreased transplant-related mortality (TRM) [36]. Registry data from the Center for International Blood and Marrow Transplant Research (CIBMTR) showed that, although TRM was higher after HSCT from a mismatched unrelated donor (MMUD) or cord blood (UCB) donor, there were no differences in LFS between MFD and MUD, MMUD, or UCB [37]. These data were further confirmed by the ALL-SCT-BFM-2003 Trial [38•], comparing allogeneic HSCT from MFD or UD in 411 high-risk ALL patients. Indeed, this was the first prospective study to compare HSCT from different donor types in well-defined cohorts of children with high-risk ALL by using a standardized transplantation protocol, as well as homogeneous supportive care. Although TRM was higher in the UD group, LFS was superimposable to that of patients transplanted from a MFD. Since the probability of finding a suitable/compatible MUD within the international volunteer registries is conditioned by the patient’s ethnic background, with a markedly lower probability for patients belonging to ethnic minorities [39], in the last decade, the use of haploidentical donors increased significantly. Indeed, this kind of donor is characterized by several desirable features, including availability for virtually all patients, choice of the best donor from a panel of candidates, immediate accessibility to the allograft, no delay in obtaining the graft, and easy access to donors’ adoptive cell therapies in case they are required after transplantation. Recently, the Rome pediatric group demonstrated in a prospective monocentric study [40•] and in a large multicenter retrospective study [41] that TCRαβ-depleted haploidentical HSCT (αβhaplo-HSCT) is a suitable therapeutic option for children with ALL in need of transplantation. Indeed, in the first study, LFS was superimposable in the three groups analyzed, namely MFD, MUD, and αβhaplo-HSCT [40•]. In the second study, LFS was comparable between MUD and αβhaplo-HSCT, while it was lower when HSCT was performed using a MMUD [41]. Notably, the risk of both acute and chronic GvHD after αβhaplo-HSCT was significantly lower than after MFD/UD HSCT, this potentially representing a further advantage in terms of patients’ acute morbidity and long-term quality of life.

Post-transplantation, early administration of two doses of cyclophosphamide is an option largely used to prevent acute GvHD in unmanipulated haploidentical HSCT, thanks to the ability of the drug to kill preferentially cycling, donor-derived alloreactive T cells, while sparing hematopoietic stem/progenitor cells [42]. Many studies have shown the efficacy of unmanipulated haplo-HSCT in adults with ALL [43, 44]; however, data on pediatric patients are limited to few reports enrolling small numbers of patients [45, 46], this precluding the possibility of drawing firm conclusions on what can be offered by this transplant option in children with ALL.

UCB transplantation (UCBT) has been largely employed in the past for treating patients with either relapsed or high-risk CR1 ALL children [47, 48]. Although registry data showed that unrelated UCBT is able to offer a long-term outcome similar to that observed using an unrelated adult volunteer as donor, the last few years have witnessed a progressive decline in the number of UCBT performed for children with hematological malignancies, including ALL [49].

What Kind of Conditioning Regimen and GvHD Prophylaxis Should Be Used?

Conditioning regimens are given before the infusion of the graft with two goals: further reduction of leukemia burden (ideally disease eradication) and immune suppression to overcome host rejection of the graft. The choice of the optimal conditioning regimen for any given patient is based on disease-related factors, such as diagnosis and disease phase, as well as patient-related factors (e.g., age, type of donor, source of stem cells, and presence of comorbidity conditions) [50].

Total body irradiation (TBI)-based regimens are considered the standard conditioning for pediatric patients affected by ALL [51, 52]. High-dose cyclophosphamide (Cy) (120 mg/kg) has been associated since the first studies to TBI [53]. A large retrospective study comparing TBI + Cy and busulfan (Bu) + Cy showed improved OS and LFS when the former conditioning was used. Since relapse is the main cause of treatment failure, intensification of this type of conditioning with a number of cytotoxic agents has been investigated [54]. In particular, the use of thiotepa (TT) (10 mg/kg), a drug that is characterized by cell cycle-independent effect, good CNS diffusion, and limited extramedullary toxicity, has been shown to be safe and effective [55]. Another drug frequently associated with TBI and TT is the purine analogue fludarabine (Flu) [40•]. With the limitation related to the nature of the study, a retrospective registry analysis from the Japan Society for Hematopoietic Cell Transplantation identified melphalan (LPAM) as the best single agent to be coupled with TBI, since this conditioning resulted into the lowest relapse incidence as compared with other combinations of radiotherapy and cytotoxic agents [54]. The same study showed improved outcome, especially when an alternative donor is used, with the addition of etoposide (VP16) to TBI and Cy [54]. The combination of TBI and high-dose VP16 (60 mg/kg) [56], although characterized by considerable mucosal and dermatologic toxicity [57], compared favorably with TBI + Cy for children in CR2 in a large CIBMTR registry study [58]. This kind of regimen has been used in the ALL-SCT-BFM-2003 trial [38•], confirming the good toxicity profile, and is currently investigated in the multicenter, randomized controlled trial (RCT) “FORUM” (NCT01949129). Given the well-known long-term toxicity of TBI in children [59, 60], this study, conducted by the international BFM consortium, is aimed at comparing non-TBI-based conditioning (Bu-based or treosulfan-based) with TBI + VP16. So far, the only RCT comparing Bu with TBI in children with ALL suggested comparable rates of relapse in the two arms, although transplant-related mortality was higher in the Bu group [61]. Thus, the results of the FORUM trial are urgently awaited. Although reduced intensity conditioning (RIC) regimens have been developed in the last years to reduce treatment-related toxicities, myeloablative conditioning (MAC) still remains the gold standard for pediatric patients undergoing HSCT for malignant disorders. Indeed, for children not eligible to MAC, RIC based on Bu and Flu has been studied, resulting in low TRM, but burdened by a high incidence of relapse [62] (see Table 3 for further details).

Table 3 Outcomes of different conditioning regimen
Full size table

Anti-human T lymphocyte globulin (ATLG) is used in patients undergoing HSCT to tune bi-directional alloreactivity, with the aim of reducing the incidence and severity of the two main immune-mediated complications of the procedure, namely graft rejection and acute/chronic graft-versus-host disease (GvHD). Four RCTs comparing the use of pre-transplant ATLG administration to regimen without serotherapy have been conducted in adult patients, showing a significant reduction in the incidence of chronic GvHD when ATLG is added to the standard prophylaxis [63]. A recent trail RCT conducted in the USA showed an increased risk of relapse in adult patients allocated to receive ATLG [64]. The only RCT in children addressing the dosage of ATLG in HSCT from UD has been conducted by Locatelli and colleagues, who compared two different doses of rabbit anti-human T cell line (Jurkat) globulin (Grafalon®, Neovii Biotech), namely 15 mg/kg versus 30 mg/kg over 3 days [65••]. The authors showed that low-dose ATLG can reduce the incidence of life-threatening infections, without significantly affecting the incidence of acute and chronic GvHD, as well as that of recurrence of the original disease, thus resulting into an improved OS and EFS. They concluded that ATLG at 15 mg/kg should, thus, be regarded as the standard serotherapy regimen for UD allogeneic HSCT in pediatric patients affected by malignant disorders.

A correct tuning of GvHD prophylaxis remains a cornerstone of the transplant strategy for children with ALL. Indeed, several authors have demonstrated that the occurrence of GvHD can reduce the incidence of relapse [66, 67]; however, high-grade GvHD is associated with a lower OS due to an increased risk of TRM [67]. The COG has deeply investigated the role of GvHD, together with that of pre- and post-HSCT MRD detection, in the ASCT0431 protocol. In this study, it was clearly shown that that pre-HSCT MRD < 0.1% and acute GvHD occurring before day + 55 were independently associated with decreased relapse and improved EFS [66, 67], thus also identifying an optimal time window to initiate interventions to prevent relapse occurrence (e.g., rapid tapering of immunosuppression, donor lymphocyte infusions, or other cellular therapies [68]). The occurrence of chronic GvHD of limited severity has been shown to prevent the risk of disease recurrence, this resulting into an improved probability of OS in children with hematological malignancies, including those with ALL [69].

Children with ALL experiencing leukemia recurrence have a dismal outcome. A minority of them can be rescued by a second allograft [70, 71], whose efficacy is better if the time elapsing between the first transplant and relapse exceeds 6–12 months. To be effective, a second allograft can be performed only if a new state of morphological CR is achieved. Donor lymphocyte infusion (DLI) has a limited value in patients with overt relapse, while it can be more effective in patients with persistence/re-positivization of MRD after HSCT [72]. Either aggressive or low-dose chemotherapy can be used before DLI to reduce the leukemia burden and promote donor T cell expansion.

HSCT in Unique Pediatric Populations

Two categories of pediatric patients with ALL deserve special consideration in terms of indication to HSCT, considering its less-clear role in these subgroups of children: patients affected by Down syndrome (DS) and patients with Philadelphia chromosome-positive (Ph+) ALL.

Children with DS are at 10 to 30% greater risk for developing acute leukemia than children with a normal constitutional karyotype; relapse and survival rates in case of ALL are worse than those of children without DS [73]. DS patients when transplanted require special consideration, because of possible pre-existing comorbidities and the potentially increased transplantation-related toxicity, resulting into increased TRM [74,75,76]. The main toxicity reported in studies of DS children undergoing HSCT involves respiratory complications, including infections, pulmonary hemorrhages, and upper airway obstruction following severe mucositis [75, 76]. Pulmonary vascular abnormalities and obstructive sleep apnea leading to pulmonary hypertension are common in children with DS and can predispose these patients to the development of the aforementioned complications. However, more recent reports suggest relapse, rather than TRM, as the most frequent cause of treatment failure and mortality in this population of patients, with a higher incidence of post-transplant relapse for children with DS (54%) than for non-DS patients [77, 78]. All these reports, however, are limited by the small sample size and the consequent impossibility of performing a subgroup analysis to define the role of conditioning regimens and GvHD prophylaxis on relapse and TRM incidence. It is important to note, however, that the risk of chemotherapy-induced toxicities accounts for a reduction of disease control before HSCT in this population of patients, with a higher proportion of children with DS receiving the transplantation in a non-remission status that might affect the final patients’ outcome. Concerning the risk of conditioning chemotherapy- or radiotherapy-associated toxicities, contrasting results have been published; indeed, three retrospective series showed that myeloablative doses of either TBI, Cy, or Bu did not impact the 100-day mortality [75, 77, 78]. Data on the risk of acute GvHD are also unclear, incidence ranging between 25% and 100% across different studies, whereas rates of chronic GvHD do not differ significantly from non-DS children [74, 77, 78]. Large controversies have been raised about the use of methotrexate as GvHD prophylaxis, given the well-known poor tolerance of these patients to the drug, although the reduced doses used in this setting as compared to those used in leukemia chemotherapy protocols might likely induce a lower toxicity [79]. So far, the limited data available do not allow drawing firm conclusions on the risk/benefit ratio associated with HSCT in patients with DS. Larger, prospective studies are necessary to define clear indications to HSCT and to identify the optimal conditioning regimens and GvHD prophylaxis approaches for this peculiar, fragile population.

Children and adolescents with Ph + ALL represent the second population of ALL patients deserving special consideration. Considered for a long time one of the poorest-prognosis subgroup of ALL, their prognostic perspective has been completely revolutionized by the introduction of imatinib and other tyrosine kinase inhibitors (TKIs) in front-line therapy. Through the blockade of the fusion protein associated with the characteristic t(9;22), TKIs significantly improved outcomes in pediatric Ph + ALL patients, with or without HSCT [80,81,82]. In 2014, COG reported outcomes of 91 children treated with the combination of imatinib and intensive chemotherapy, showing a 5-year DFS of 70% in the chemotherapy + imatinib group, 65% for patients receiving a MFD HSCT, and 59% for children transplanted from an UD, these results suggesting that many patients could be maintained on TKI treatment only, avoiding allo-HSCT [83]. A more recent study based on the use of the second-generation TKI dasatinib combined with intensive chemotherapy, reserving HSCT to high-risk patients with a MFD, showed comparable 5-year EFS between the two groups (61% versus 67%) [84]. Taken together, these data suggest the possibility of avoiding HSCT in a subgroup of pediatric patients with Ph + ALL in CR1, namely low-risk patients (i.e., those with low PCR-MRD at the end of induction). Identification of predictive factors influencing the outcome can further refine the selection of patients that can skip HSCT without impairing long-term survival. Whereas the role of MRD remains clear in predicting relapse also for these patients, the role of additional cytogenetic mutations needs to be further elucidated [85,86,87]. Lastly, the efficacy of the different TKIs available could also influence the outcome, as shown by studies comparing the last generation ponatinib with dasatinib and showing the superiority of the first one [88, 89].

What Is the Role of HSCT in the Immunotherapy Era?

In the last 15 years, novel treatment approaches aimed at exploiting the ability of the immune system to recognize and eliminate leukemia blasts have been developed and proved to be extremely effective in patients affected by refractory/relapsed (r/r) BCP-ALL [24•]. Different strategies have been developed, including BiTEs, CAR-expressing T cells, and the so-called antibody drug conjugates (ADCs).

Blinatumomab, the CD19-directed BiTE, was shown to be extremely effective in children with r/r CD19-positive BCP-ALL in an international, multicenter phase I–II study, showing 39% of CR within the first cycle and a median LFS of 4.4 months (increased to 7.5 months in the MRD-negative subpopulation) [90••]. Earlier use of blinatumomab is currently under investigation in the IntReALL consortium, as consolidation therapy before HSCT in children with high-risk first relapse, in order to reduce the pre-transplantation MRD levels. Promising results have been shown also by the ADC inotuzumab, a CD22-specific mAb conjugated with the cytotoxic agent calicheamicin, in a cohort of 51 children with r/r ALL treated on a compassionate use basis; 67% of patients achieved CR, this observation providing a strong basis for implementing a prospective, multicenter, phase I/II study [91]. Lastly, CD19-CAR T cells are currently the most advanced T cell therapy tested in clinical trials on patients with BCP-ALL. Initial clinical reports and clinical trials have demonstrated striking responses, with durable remissions obtained in patients with r/r ALL who were considered incurable [92••, 93]. These results led to the recent approval first by the Food and Drug Administration and subsequently by the European Medicines Agency of two CD19-directed CAR T cell therapies (Kymriah™ and Yescarta™), for treatment of children and young adults with r/r BCP-ALL and of adult patients with r/r large B cell lymphoma, respectively [94•].

In this scenario, the role of HSCT will likely have to be redefined. It must be noted, however, that, although effective, with durable responses in a subgroup of patients, CAR T therapy and other targeted therapies do not prevent the phenomenon of antigen escape, which renders leukemic blasts resistant to these approaches [95]. Well-designed RCTs will have to compare directly the long-term leukemia control offered by CAR T cell approaches versus HSCT, showing whether the former can substitute allogeneic HSCT. Interestingly, in the Memorial Sloan Kettering Cancer Center experience with CD19-CAR T cells, amongst the 32 patients obtaining MRD-negative CR, no significant difference in OS between patients who either did or did not subsequently undergo HSCT was found (P = 0.89) [93]. In the ELIANA trial, substantially identical OS was reported for patients who did not receive HSCT as compared to those who underwent HSCT, the persistence of tisagenlecleucel in peripheral blood being the main factor associated with durable response [92••]. The limited and non-homogeneous data reported to date do not allow drawing firm conclusions, and the question remains to be properly addressed in future well-designed trials. Moreover, head-to-head comparisons of different immunotherapy strategies are lacking, precluding any definitive conclusion on the relative roles and merits of each approach. Future clinical trials and research developments will ultimately determine the best strategy to treat children with high-risk or r/r BCP-ALL.

Conclusions

For many years, allogeneic HSCT from an HLA-identical sibling has represented the treatment of choice for many children with high-risk features or r/r ALL. In the last two decades, results of transplantation from an UD have progressively improved over time, becoming similar to those obtained using a MFD when a highly compatible donor can be identified. More selective strategies of T cell depletion, such as those based on the physical removal of αβ T cells from the graft, have rendered HSCT from an HLA-haploidentical donor a suitable option to virtually transplant any child in need of an allograft. Several important questions are under investigation, including that of the advantage/limitations offered by a TBI-containing conditioning regimen in comparison to those based on the combination of cytotoxic agents. Through the conduction of well-designed trials based on the intention-to-treat approach analysis, the next few years will witness if allogeneic HSCT can be replaced by CAR T cells in BCP-ALL. Till these studies will be finalized, HSCT remains the standard-of-care treatment for every child with CR1 ALL carrying high-risk features predicting leukemia recurrence and for those experiencing high-risk first relapse or multiple recurrences.