Incorporation of nonchemotherapeutic agents in pediatric acute lymphoblastic leukemia

With current risk-stratified regimens, >80% of pediatric patients with acute lymphoblastic leukemia (ALL) are expected to be long-term, event-free survivors and ∼90% are ultimately cured of their disease.^1,2  However, the cytotoxic chemotherapy regimens that have led to these high cure rates are associated with many acute and long-term complications, especially for those with high-risk disease. In addition, for patients who experience relapse, salvage therapy remains toxic and suboptimally effective.^3,4  Thus, despite the relatively favorable overall prognosis for pediatric patients with ALL, there is a compelling need to develop more effective and less toxic therapies.

Over the last several years, comprehensive genomic analyses of childhood ALL have identified lesions that are potentially amenable to targeted treatment approaches, some of which are being tested in subsets of high-risk de novo patients. In addition, promising activity has been observed with several immunotherapeutic approaches, including monoclonal antibodies, immunotoxins, bispecific T-cell–engaging (BiTE) antibodies, and chimeric antigen receptor (CAR) T cells (Table 1).

Philadelphia chromosome–positive (Ph+) ALL, characterized by the presence of the BCR-ABL1 fusion (a constitutively activated tyrosine kinase), occurs in ∼3% to 5% of childhood ALL.⁵  Historically, patients with Ph+ ALL had a poor outcome when treated with standard chemotherapy regimens and thus were allocated to more intensified treatments, including hematopoietic stem cell transplantation (HSCT) in first complete remission (CR). Even with intensified therapy, outcomes for Ph+ ALL were inferior to other pediatric ALL subtypes. In a retrospective analysis of 610 pediatric patients with Ph+ ALL treated between 1995 and 2005 by 10 study groups, the 7-year event-free survival (EFS) was 32% and overall survival was 45%.⁶ 

The development of tyrosine kinase inhibitors (TKIs) targeting the BCR-ABL1 fusion, such as imatinib and dasatinib, has transformed the treatment of pediatric Ph+ ALL. Several clinical trials have been conducted to test the addition of TKIs to intensive chemotherapy regimens (in many cases, with the omission of allogeneic HSCT).^7-9  The success of these trials provides a paradigm for adding targeted therapies to cytotoxic chemotherapy backbones in order to improve cure rates for high-risk patient subsets.

In a trial conducted between 2002 and 2006, the Children’s Oncology Group (COG) evaluated the use of imatinib in combination with intensive multiagent chemotherapy in children and adolescents with Ph+ ALL, with some patients proceeding to HSCT depending on the availability of an HLA-matched related donor.⁹  With a median follow-up of 5.2 years, the disease-free survival for those treated with continuous dosing of imatinib but without HSCT was 70% ± 12% (n = 28), which was not significantly different (P = .60) from that of imatinib-treated patients who underwent HSCT from a related donor (65% ± 11%; n = 21) or unrelated donor (59% ± 15%; n = 13).⁹ 

The successor COG Ph+ ALL study AALL0622, which enrolled patients between 2008 and 2012, evaluated the feasibility of administering dasatinib on a similar chemotherapy backbone. Dasatinib was started on day 15 of induction (rather than after completion of induction as in the previous trial with imatinib), which resulted in an improved CR rate and a higher proportion of patients with low levels of end-induction minimal residual disease (MRD),¹⁰  thus indicating that earlier introduction of TKIs (midinduction) was both feasible and favorably impacted early response. The combination of dasatinib with chemotherapy was well tolerated, without excessive toxicity noted. The overall 3-year EFS was 79% ± 6% (n = 58 evaluable patients),¹¹  similar to the 3-year EFS observed in the previous trial with imatinib.

The European Intergroup Study of Postinduction Treatment of Philadelphia Chromosome–Positive ALL (EsPhALL), which enrolled 178 pediatric patients with Ph+ ALL between 2004 and 2009, tested the safety and efficacy of postinduction imatinib (administered in a discontinuous schedule) utilizing a different cytotoxic chemotherapy regimen than the contemporaneous COG Ph+ ALL protocols. Patients in the EsPhALL trial were risk-stratified based on absolute blast count after a 7-day prednisone prephase and marrow status after induction chemotherapy; all “poor risk” patients received imatinib and “good risk” patients were randomly assigned to receive imatinib or not. For good-risk patients, analyses by treatment received showed that 4-year disease-free survival was 75% for those who received imatinib and 56% for those who did not (P = .06).⁷ 

Results from these trials support the feasibility and efficacy of incorporating TKIs into multiagent therapy for pediatric Ph+ ALL and show that this combination may reduce the need for HSCT in first CR. However, many issues remain unsettled. For instance, the cytotoxic backbones of both the COG and EsPhALL regimens are intensive, leading to a high frequency of acute and potentially long-term treatment complications. In addition, the biologic mechanisms leading to relapse (the primary cause of treatment failure) remain unclear. In studies of adults with Ph+ ALL treated with imatinib or dasatinib, BCR-ABL1 kinase domain mutations conferring resistance to TKI have been observed in the majority of patients at the time of relapse.^12-14  There are limited data regarding the frequency of these mutations in pediatric Ph+ ALL; however, in one study of 9 imatinib-treated pediatric patients with available samples at the time of relapse, only 1 patient had evidence of a BCR-ABL1 kinase domain mutation associated with TKI resistance,¹⁵  suggesting that the frequency of these resistant clones may be lower than in adult Ph+ ALL. Thus, further studies are necessary to optimize the therapy of pediatric patients with Ph+ ALL, including investigations of different TKIs and chemotherapy backbones (to both improve cure rates and reduce treatment-related toxicities), refinement of prognostic factors to identify high-risk patients who may still benefit from HSCT and/or other novel therapies, the role of postchemotherapy and/or post-HSCT TKI, and the mechanisms underlying treatment resistance, including more extensive characterization of the frequency of kinase domain mutations at diagnosis and relapse. In addition, the potential long-term sequelae of prolonged TKI therapy in children, including abnormalities of bone metabolism and growth impairment,¹⁶  remain to be elucidated.

Recent studies have identified a subset of patients with B-ALL characterized by a gene expression profile similar to Ph+ ALL, but without the BCR-ABL1 fusion. This subgroup, termed BCR-ABL1-like (or Ph-like) ALL, occurs in up to ∼15% of pediatric patients with B-ALL and has been shown to be an independent predictor of adverse outcomes.^17,18 BCR-ABL1-like ALL is more common in older children and adolescents,¹⁷  which may in part explain the increased relapse risk that has historically been observed in this age group compared with younger patients.¹⁹ 

Detailed genomic analyses have identified kinase-activating alterations in >90% of patients with BCR-ABL1-like ALL. These alterations appear to impact a limited number of signaling pathways, notably the ABL-class and JAK-STAT pathways, suggesting the potential for targeted interventions.¹⁸  One subset, harboring ABL-class fusions involving ABL1, ABL2, CSF1R, or PDGFRB, was shown to be sensitive in vitro to dasatinib.^18,20  Other subsets have been identified with in vitro sensitivity to the JAK1/JAK2 inhibitor ruxolitinib, including BCR-ABL1 cases with rearrangements of JAK2, EPOR, and/or CRLF2 (with or without concomitant JAK1 or JAK2 mutations).^18,20,21 

Given the success of adding TKIs to Ph+ ALL, clinical trials are now underway to assess whether adding either dasatinib or ruxolitinib to cytotoxic chemotherapy backbones in first CR will reduce the relapse risk of pediatric patients with BCR-ABL1-like ALL who harbor genetic alterations that are potentially targetable by one of these agents. As other potentially targetable fusions and mutations are discovered in this subset of patients, it is possible that additional therapies could be investigated in a similar manner (eg, adding crizotinib to chemotherapy in patients with ETV6-NTRK3 fusion)¹⁸ ; however, the rarity of many of these aberrations will likely require international collaborative trials for definitive testing.

Several immunotherapeutic approaches are under active investigation in childhood ALL. Although they vary in their mechanisms of action, all of these approaches target cell surface antigens expressed on B lymphoblasts, including CD19 and CD22 (nearly universally expressed in B-ALL) and CD20 (less frequently expressed). Immunotherapeutic strategies that have been tested in B-ALL can be subdivided into the following categories: (1) naked (or unconjugated) monoclonal antibodies, such as rituximab and epratuzumab, typically combined with cytotoxic chemotherapy backbones; (2) immunotoxins (toxin-conjugated antibodies), such as inotuzomab ozogamicin; (3) bispecific antibodies, such as blinatumomab, which bind to both B lymphoblasts and T cells, promoting autologous T-cell killing of leukemia cells; and (4) CAR T cells. Clinical trials of these immune-based therapies have yielded promising results in patients with relapsed or refractory B-ALL, raising the possibility that these treatments could have a role in upfront therapy in newly diagnosed patients to prevent relapse and reduce exposure to cytotoxic chemotherapy. Although immune-based therapies hold great promise for patients with B-ALL, their success is dependent on the near uniform expression of the target surface antigen; leukemias with more variable expression of these antigens (either inherent or downregulated in response to the targeted treatments) may evade these therapies. In addition, currently available immune-based therapies do not target antigens typically expressed in T-ALL; thus, these patients (∼10% to 15% of pediatric ALL cases) are not yet candidates for these treatments.

Rituximab is a chimeric monoclonal antibody that binds to CD20, inducing cell killing by antibody-dependent cytotoxicity, complement-dependent cytotoxicity, and apoptosis. The role of rituximab in improving outcomes for both adult and pediatric patients with B-cell non-Hodgkin lymphomas is well established,^22,23  and its use in those diseases is now considered standard of care. Given these compelling findings, as well its favorable toxicity profile when combined with cytotoxic chemotherapy, rituximab has also been tested in CD20-positive B-ALL, primarily in adults.²⁴ 

The randomized multicenter Group for Research on Adult Acute Lymphoblastic Leukemia 2005 (GRAALL-2005) trial enrolled 209 adult patients (aged 18 to 59 years) with newly diagnosed CD20-positive, Ph-negative B-ALL.²⁵  Results from the GRAALL-2005 study demonstrated that the addition of rituximab to chemotherapy was associated with superior EFS (2-year EFS of 65% vs 52%; hazard ratio, 0.66; P = .04) without an increase in the frequency of serious adverse events.²⁵  Interestingly, fewer patients in the rituximab group experienced an allergic reaction to l-asparaginase. Thus, in addition to a direct antileukemic effect, rituximab may reduce relapse risk indirectly by inhibiting production of antiasparaginase antibodies, leading to increased exposure to l-asparaginase.

Only 30% to 50% of cases of B-ALL are CD20 positive, which may limit rituximab’s role in treating the disease. However, Dworzak et al²⁶  analyzed changes in CD20 expression during the first month of treatment in >200 pediatric patients with B-ALL. They found that CD20-positivity significantly increased from 45% at diagnosis to 75% at day 8 of induction (as ascertained by peripheral blood flow cytometry) and 81% by the end of induction (in patients with detectable marrow MRD at that time point). These findings were supported by in vitro experiments demonstrating that prednisolone induced CD20 upregulation in lymphoblasts. The favorable results of the GRAALL-2005 study, along with data suggesting that CD20 expression may increase during induction treatment, provide a rationale for testing rituximab in all B-ALL cases after initiation of therapy, regardless of initial CD20 expression. The St. Jude Total XVI trial, which opened in 2017, is randomly assigning all patients with newly diagnosed B-ALL to receive 1 dose of rituximab or not during the first month of treatment to determine rituximab’s impact on end-induction MRD levels, incidence of asparaginase allergy, and EFS.

CD22 is nearly universally expressed in pediatric B-ALL and thus is a more attractive target for immunotherapeutic approaches than CD20. Epratuzumab is a humanized anti-CD22 monoclonal antibody that is internalized after binding to cell surface CD22, modulating B-cell activation and signaling.²⁷  In a phase 1/2 trial conducted by COG in patients with B-ALL with first marrow relapse, epratuzumab (either 4 or 8 doses) was given in conjunction with a standard reinduction regimen.²⁸  The rate of the second CR was similar for patients receiving 4 or 8 doses of epratuzumab and was unchanged from that observed in historic controls receiving the same reinduction regimen without epratuzumab. Rates of MRD negativity (defined as MRD <0.01%) were not significantly different compared with historic controls.

The ongoing International Study for Treatment of Standard Risk Childhood Relapsed ALL 2010 is further investigating whether the addition of epratuzumab can improve outcomes in pediatric patients with B-ALL in first relapse. In this trial, children and adolescents with isolated extramedullary relapses or late marrow-involved relapses are randomly assigned to receive epratuzumab or not during postinduction consolidation therapy.

Inotuzumab ozogamicin (InO) is a humanized anti-CD22 monoclonal antibody conjugated to calicheamicin, the same agent used in the CD33-targeted immunotoxin gemtuzumab ozogamicin, which has been shown to be active in AML.²⁹  After it binds to CD22, InO is rapidly internalized and calicheamicin is released, inducing DNA damage and subsequent apoptosis. In early-phase trials in adults with relapsed or refractory B-ALL, InO was initially administered as a single dose every 21 days, with a complete marrow response rate (with or without peripheral blood recovery) of 57%.³⁰  A subsequent trial used a weekly dosing schedule in an attempt to reduce toxicity, with a similar overall response rate (66%).³¹ 

In a phase 3 trial in adults with relapsed or refractory B-ALL (A Study of Inotuzumab Ozogamicin Versus Investigator's Choice of Chemotherapy in Patients With Relapsed or Refractory Acute Lymphoblastic Leukemia [INO-VATE ALL]), 218 patients were randomly assigned to receive either InO or 1 of 3 standard chemotherapy regimens (fludarabine/cytarabine with granulocyte colony-stimulating factor; cytarabine plus mitoxantrone; or high-dose cytarabine).³²  The patients assigned to InO had superior outcomes, with significantly higher rates of CR (81% vs 30%; P < .001) and progression-free survival (hazard ratio, 0.45; P < .001). Among the patients who achieved CR (with or without complete hematologic recovery), the proportion with low MRD was significantly higher in the InO group (78% vs 28%; P < .001).

Hepatic veno-occlusive disease/sinusoidal obstructive syndrome (VOD/SOS) has been of concern with InO. In the INO-VATE ALL trial, VOD/SOS occurred in 15 of 139 patients (11%) in the InO arm compared with 1 of 120 patients (0.8%) in the control arm.³²  The frequency of VOD/SOS was highest in those who underwent HSCT either before or after receiving InO.

InO has not been extensively studied in pediatric patients with B-ALL.³³  A COG-led phase 2 study in patients with relapsed or refractory pediatric B-ALL opened to accrual in 2017.

Blinatumomab is a BiTE antibody targeting CD19, a surface antigen that is nearly always expressed on B lymphoblasts. BiTE antibodies are genetically engineered to combine 2 variable regions with different specificities; in the case of blinatumomab, one region binds CD3 (T cells) and the other binds CD19 (B lymphoblasts).³⁴  Treatment with blinatumomab results in T cells being brought into close proximity to B lymphoblasts, initiating a cytotoxic response to the leukemic cell that does not require T-cell specificity to the tumor and bypasses resistance mechanisms often employed by B lymphoblasts to avoid immune surveillance, such as downregulation of HLA antigens.

Blinatumomab has been shown to be active as a single agent in both adult and pediatric patients with relapsed or refractory B-ALL. In a multicenter phase 2 study of blinatumomab, 81 of 189 adult patients (43%) who received blinatumomab (by continuous infusion over 4 weeks, every 6 weeks) achieved marrow CR (with or without full recovery of peripheral blood counts) after 2 cycles.³⁵  In a phase 3 trial comparing blinatumomab to standard chemotherapy (similar in design to the INO-VATE ALL trial described above) in adults with relapsed, refractory B-ALL, blinatumomab was associated with superior marrow CR rates (44% vs 25%; P < .001) and overall survival.³⁶  In a pediatric phase 1/2 trial, 27 of 70 patients (39%) treated at the recommended phase 2 dose achieved CR with single-agent blinatumomab; 52% of those achieving CR were MRD negative.³⁷ 

Administration of blinatumomab, especially in patients with high disease burden at the time of treatment, has been associated with cytokine release syndrome (CRS), which is characterized by fever and flu-like symptoms and, in severe cases, hypotension, capillary leak, and respiratory distress.^35-37  CRS is thought to be mediated by elevated levels of cytokines associated with T-cell activation, including interleukin (IL)-6, IL-10, and interferon-γ³⁸ ; thus, the presence of some degree of CRS is indicative of a successful drug effect. In the pediatric phase 1 trial, 3 of 4 dose-limiting toxicities during the first cycle were attributable to severe CRS, including 1 fatal episode. Subsequently, a stepwise increase in the blinatumomab dosage, with a lower dose administered for the first 7 days of the 4-week continuous infusion, was evaluated and found to be more tolerable.³⁷  Eight of 70 patients (11%) treated in this manner experienced CRS, but only 4 (6%) experienced grade ≥3 CRS. Treatment with tocilizumab, an antibody that blocks IL-6 from binding to its receptor, has been shown to effectively treat severe CRS.³⁸ 

In addition to being evaluated in trials in patients with relapsed or refractory disease with extensive marrow involvement at the time of treatment, blinatumomab has also been tested in high-risk patients with B-ALL with a lower disease burden. In one of the first published trials of blinatumomab, 16 of 20 adult patients (80%) in morphologic CR but with MRD positivity (defined as ≥10-4) achieved MRD negativity after 1 cycle of treatment; 8 of these 16 patients went on to receive an allogeneic HSCT, without any subsequent relapses observed.³⁹  Of note, blinatumomab-related CRS and neurotoxicity were less frequent and severe than when administered to patients with higher disease burden. Thus, blinatumomab could potentially be integrated into multiagent regimens during postinduction treatment phases in high-risk patients, particularly those with higher MRD levels. Blinatumomab is currently being investigated in this way in a randomized phase 3 trial conducted by COG for pediatric patients with B-ALL in first relapse.

Infusion of genetically engineered autologous CAR T cells is an emerging and very promising immunotherapeutic approach in B-ALL. CAR T-cell therapy involves collecting autologous T cells from patients (via leukapheresis), genetically engineering the collected cells ex vivo so that they express CARs specific for the target leukemia-associated antigen, and then reinfusing these cells into the patient, typically after a chemotherapy preparative regimen. Initial trials in B-ALL have focused on testing CAR T cells that have been engineered to couple an anti-CD19 domain to intracellular T-cell signaling domains, thus redirecting cytotoxic T cells to C19-expressing cells.^40,41 

In 2014, investigators from the Children’s Hospital of Philadelphia and the Hospital of the University of Pennsylvania published the results of a phase 1/2 trial of CD19-directed CAR T-cell therapy in 30 children and adults (25 of whom were aged ≤22 years) with multiply relapsed or refractory CD19-positive ALL.⁴¹  Patients were infused with autologous T cells transduced with a CD19-directed CAR lentiviral vector incorporating the 4-1BB costimulatory domain (CTL019). CR was obtained in 90% of patients, including 15 of 18 (83%) who had previously received allogeneic HSCT. The 6-month EFS rate was 67%, with most patients showing persistence of CAR T cells and B-cell aplasia through 6 months. All 30 patients experienced some degree of CRS; 8 patients (27%) had severe symptomatology (need for vasopressors and/or respiratory support), which was effectively treated with tocilizumab. In a follow-up report including additional patients (a total of 53 patients aged 4 to 24 years) and longer follow-up, the CR rate was 94% (with 90% of patients having MRD <0.01%, as assessed by flow cytometry) and 12-month EFS was 45%.⁴² 

Other CAR T-cell products, utilizing different vectors and/or costimulatory molecules, have also been evaluated in pediatric patients. The National Cancer Institute (NCI) Pediatric Oncology Branch tested a CD19-targeted CAR T-cell product containing a CD28 costimulatory domain that was prepared using a retroviral vector, with a resultant CR rate of 70%.⁴³  Favorable CR rates in patients with relapsed or refractory pediatric B-ALL have also been observed utilizing alternative CD19-directed CAR T-cell products in early-phase trials conducted at Memorial Sloan Kettering Cancer Center and Seattle Children’s Hospital/Fred Hutchinson Cancer Research Center.^44,45  The NCI is also studying a CD22-targeted CAR T cell.⁴⁶  Although encouraging CR rates have been reported in nearly all of these trials, the duration of CAR T-cell persistence and the frequency and timing of severe side effects have varied, possibly owing to differences in the CAR T-cell product itself (viral vector, costimulatory molecule) as well as the preparative regimen administered before infusion.

CRS is frequently observed after CAR T-cell infusions and can be severe (grade ≥3) in up to 20% to 40% of cases.^41,45,47  Both corticosteroids and tocilizumab have been used to manage CRS; tocilizumab has become the preferred initial approach because of concern that corticosteroids may more significantly interfere with the antitumor effect of the CAR T cells. Severe neurotoxicity, including aphasia, altered mental status, seizures, and cerebral edema, has been reported in adult and pediatric CAR T-cell trials.^45,47  In the majority of patients, even the most severe neurotoxicity appears to be reversible; however, fatal cases have been reported.

In addition to these side effects, potential barriers to widespread application of CAR T cells include the logistical complexity in collecting autologous T cells from patients, successfully manufacturing the genetically modified product, and the waiting time between collection and infusion (typically several weeks). In a multicenter trial of CTL019 conducted at 9 US sites, 29 of 35 enrolled participants (83%) ultimately received CAR T cells; of the 6 patients who were enrolled but did not receive CAR T cells, 2 did not receive the treatment as a result of manufacturing failures and 4 died before the infusion could be performed.⁴⁷  The target cell dose was met in 73% of CAR T-cell products.

The extremely promising early outcome results of CAR T-cell trials highlight the potential of this treatment approach to transform therapy for patients with relapsed and high-risk de novo B-ALL. However, many questions remain unanswered, including the duration of response (and whether this treatment can be considered definitive therapy, replacing prolonged chemotherapy and/or HSCT), mechanisms of relapse, and long-term sequelae. Further clinical investigation is needed in order to determine the optimal integration of CAR T cells into the care of pediatric patients with ALL beyond the multiply relapsed or refractory setting.

Over the last decade, a number of nonchemotherapeutic therapies have emerged with the potential to improve outcomes for children and adolescents with ALL. Recent trials of TKIs in Ph+ ALL demonstrate how targeted therapies can be successfully incorporated into multiagent chemotherapy regimens; ongoing studies will clarify whether these same agents improve outcomes for the BCR-ABL1-like subset. Immunotherapeutic approaches hold great promise in B-ALL; further studies are needed to determine how best to integrate them into current treatment regimens, prevent the development of resistance, address logistical challenges in their administration, more effectively treat and prevent significant toxicities, and expand their applicability to patients with T-ALL.

Lewis B. Silverman, Dana-Farber Cancer Institute/Boston Children’s Hospital, 450 Brookline Ave, Boston, MA 02215; e-mail: lewis_silverman@dfci.harvard.edu.