Introduction

The treatment of childhood acute lymphoblastic leukemia (ALL) has steadily improved over the past several decades. In children aged 2–10 years, 5-year survivals are now in the 90% range [1–3]. In contrast, outcomes in adolescents and young adults with ALL have been less successful. However, recent advances in therapy appear to have significantly improved results in these patients. Nevertheless, a number of treatment-related controversies and challenges remain.

The standard treatment of adolescents and young adults with ALL usually consists of four phases. The first, induction therapy, produces rapid cytoreduction and attainment of complete remission (CR), and is generally performed on an inpatient basis. In some protocols, this is now preceded by a steroid pre-phase. Induction therapy generally uses combinations of vincristine, corticosteroids, an anthracycline, and of additional agents, most commonly asparaginase, cyclophosphamide, or methotrexate.

Post-remission therapy includes three phases—central nervous system (CNS) prophylaxis (which is sometimes at least partly incorporated into other phases), systemic intensification (or consolidation), and maintenance therapy. CNS prophylaxis always includes repeated doses of intrathecal chemotherapy, and may also include certain high-dose systemic chemotherapy agents with high CNS penetration (methotrexate, cytarabine), and/or cranial radiation. Intensification, either inpatient or outpatient-based, generally continues for 3–8 months depending on the regimen. Maintenance therapy usually consists of daily 6-mercaptopurine and weekly methotrexate, sometimes combined with corticosteroids and vincristine. This phase usually continues for at least 2 years from diagnosis. Patients deemed to be at very high risk of relapse may be referred during treatment for allogeneic hematopoietic stem cell transplantation (HSCT) in first CR.

The current review will highlight the most recent advances and state-of-the-art approach to the management of ALL in adolescents and children. The major therapeutics challenges will be outlined, and some promising new approaches will be discussed.

ALL in Adolescents

Historically, children ≥10 years of age have been reported as having an inferior prognosis as compared to younger children, with overall survival (OS) and event-free survival (EFS) in the 45–70% range [4–6]. This is at least partly related to a higher incidence of adverse prognostic factors, including high WBC, t(4;11), and BCR-ABL abnormalities, and a lower frequency of favorable factors such as hyperdiploidy and TEL-AML [5]. A number of studies in the past decade (summarized in Table 1) have shown superior results using pediatric-based regimens, which utilize higher cumulative doses of asparaginase, vincristine, and corticosteroids such as dexamethasone, compared to adult-based regimens. Boissel et al. [7] retrospectively compared patients age 15–20 years treated with a pediatric-based regimen (FRALLE 93) to the adult-based LALA 94 protocol. The 5-year EFS was substantially higher with the FRALLE regimen (67% vs 41%, P < 0.001). A similar analysis by MRC showed an advantage in 5-year EFS using a pediatric versus adult-based regimen (67% vs 49%, P = 0.01) [8]. The Dutch group also reported superior results using a pediatric as compared to an adult-based regimen [9].

Table 1 Summary of recent studies using pediatric-based protocols in adolescents and young adults with acute lymphoblastic leukemia
Full size table

Subsequently, The Dana Farber Consortium reported a 5-year EFS of 78% for adolescents treated with their pediatric-based DFCI regimen, with no difference in outcomes for patients aged 10–15 versus 15–18 [10]. Stock et al. [11] compared patients aged 16–20 treated with pediatric-based CCG protocols over a 6-year period to those treated using the adult-based CALGB protocol. The CCG regimens utilize comparatively much higher cumulative doses of non-myelosuppressive drugs, including dexamethasone, vincristine, and asparaginase. The 5-year EFS and OS with the CCG protocol were 63% and 67%, respectively, as compared with 34% and 46% with the CALGB protocol (P < 0.001 for each). More recently, the Children’s Oncology Group reported 5-year EFS and OS of 72.5% and 80.3%, respectively, for standard-risk patients with their CCG1961 protocol, which uses early post-induction intensification with asparaginase [12•]. Finally, St. Jude’s Research Hospital recently reported substantial improvements in outcome with adolescents, compared to those receiving prior regimens, using a protocol which intensifies the doses of dexamethasone, asparaginase, and vincristine, with 5-year EFS and OS of 87% and 93%, respectively [13•]; outcomes in adolescents were comparable to those of younger children.

Taken together, these studies indicate that adolescents with ALL can achieve outcomes approaching those of younger children, with 5-year survivals in excess of 70%, using intensified pediatric-based regimens, without HSCT.

ALL in Young Adults

Traditionally, ALL patients over age 18 have been treated with protocols specifically designed for adults, owing to a widespread assumption that most adults are not capable of tolerating intensive pediatric-based regimens. Using such adult-based regimens, 5-year survivals in the 35–55% range have been reported [14, 15]. For example, the MRC-ECOG study reported 5-year OS of 52% in standard-risk young adults (defined as age 18–35, Philadelphia chromosome [Ph] negative with low white blood count presentation) [15].

The encouraging results in adolescents have led to a number of attempts to extend pediatric-based protocols into the young adult population, variably defined as ranging from age 18–20 up to 30–35 years. The PETHEMA group reported on 46 patients age 19–30 with low-risk ALL treated with a pediatric-inspired regimen, achieving 6-year OS and EFS which were not significantly different from those obtained in their adolescent group receiving the same regimen (Table 1) [16]. Using the pediatric-based GRAALL-2003 protocol, a 42-month EFS of 63% was obtained in Ph-negative patients aged ≤45 years, as compared to 35% for a similar aged group, with similar baseline characteristics, who had previously been treated with the adult-based LALA-94 regimen (P < 0.001) [17].

Our group at Princess Margaret Hospital (PMH) in Toronto recently reported results using a pediatric regimen based on the DFCI protocol [18•]. In 42 Ph-negative patients age 18–35, the 5-year relapse-free survival (RFS) and OS were 77% and 83%, respectively. A subsequent retrospective comparison in the subset of patients with T-ALL indicated a substantial improvement in survival compared to a historical control group receiving a variety of adult-based protocols [19]. These results are similar to those reported by the Dana Farber group [20]. A subgroup analysis in the Princess Margaret study found that patients who were not able to receive at least 80% of the anticipated dose of asparaginase during intensification had a significantly higher cumulative relapse risk and lower OS; this is similar to findings by the Dana Farber group in the pediatric population [21], and emphasizes the importance of asparaginase intensification in achieving optimal results.

Taken together, these studies indicate that pediatric-based protocols are producing superior outcomes in younger adults with Ph-negative ALL than previously utilized adult regimens, without HSCT; results are now approaching those in the adolescent population.

Prognostic Factors

Despite encouraging results with recent studies in adolescents and young adults, certain subgroups are still experiencing higher failure rates with treatment. One of the challenges is identifying such patients early, in order to intervene with other treatment strategies such as HSCT.

Patients with high initial white blood counts (WBC), most often defined as >30 × 109/L for B-ALL and >100 × 109/L for T-ALL, are well-recognized to be at higher risk of relapse. Although most early studies identifying this factor used adult-based protocols, subgroup analyses in larger studies show that these continue to be important prognostic factors with pediatric-inspired regimens. For example, Boissel et al. [7] found increasing WBC to be significantly associated with inferior EFS on multivariate analysis (P < 0.001) in their adolescent group treated with the FRALLE 93 regimen. The Children’s Oncology Group reported high WBC (defined as >50 × 109/L) to be an adverse prognostic factor in adolescents [12•]. The Princess Margaret series also showed significantly inferior 5-year RFS and OS in adult patients with high WBC presentations [18•].

The influence of immunophenotype on prognosis is more controversial. Most studies have found that the prognosis for T-ALL does not differ significantly from that of the more common B subtype. However, results from the adult-based UKALL/EGOG protocol indicated that T-ALL was associated with the superior 5-year survival (48% vs 41%). Using the DFCI protocol, there was a trend toward higher 5-year EFS in adolescents with T-ALL [11]; a similar trend was found in our adult series, although this did not reach statistical significance [18•].

CD20 expression has been associated with inferior survival, using the adult-based hyper-CVAD protocol [22]. However, this may be a regimen-dependent effect; a retrospective analysis of Princess Margaret adult patients, most of whom had received a pediatric-based regimen, did not find any association between CD20 expression and outcome [23]. The absence of CD10 expression has also been associated with adverse outcome, primarily due to its close association with the presence of MLL rearrangements.

Cytogenetic and molecular abnormalities are also associated with prognosis in Ph-negative ALL. Although hyperdiploidy has been associated with more a favorable outcome in younger children, it appears to be more of a standard risk factor in adolescents and adults [24]. Conversely, a number of adverse risk factors have been identified in adults with B-ALL, including t (4;11), t(1;19), complex karyotype, and near triploidy [24–26]. However, discordant results have been found between different studies. Most analyses in adult ALL have been found using adult-based protocols, and more studies are needed to evaluate their potential prognostic with pediatric-based regimens. Despite this, it does appear that MLL rearrangements, particularly those associated with t (4;11), remain a significant adverse prognostic factor even with these regimens [18•]. Most studies in T-ALL are limited by small numbers; however, recent data suggest that complex karyotypes may be associated with inferior survival, while NOTCH mutations, which occur in about 60% of cases, may have a more favorable outcome [27].

HSCT in Ph-Negative ALL

The role of allogeneic HSCT in young patients with Ph- ALL is controversial. An older retrospective comparison in patients age 15–45 from the International Bone Marrow Transplant Registry did not show any difference in leukemia-free survival between chemotherapy alone versus matched sibling HSCT in CR-1 [28]; the lower relapse rate in transplanted patients was offset by a higher treatment-related morality. The LALA-94 study also did not find any difference in survival between standard-risk patients assigned to HSCT compared to chemotherapy alone [29]. In contrast, more recently the MRC/ECOG study, using a similar design but larger numbers, found a 63% 5-year OS with HSCT versus 52% with chemotherapy (P = 0.02), in standard-risk adults up to age 35 [16]. The 10-year cumulative relapse rate was 24% in transplanted patients versus 49% in the chemotherapy-treated group.

One major limitation of the MRC/ECOG study was the use of an adult-based chemotherapy regimen. While there have been no prospective studies to date comparing HSCT with a pediatric-based regimen in adolescents or younger adults, the 5-year OS obtained with these protocols, as outlined previously, have been in ranges comparable, or superior, to the transplanted arm in the MRC/ECOG study. Because of this, many centers which have adopted these pediatric-based protocols, including our own, are not transplanting standard-risk adolescents or young adults in CR-1, and thereby avoiding the substantial morbidities and potential mortality associated with HSCT.

Nevertheless, certain high-risk groups may benefit from HSCT in CR-1. The French group reported, as part of the LALA-94 study, that patients with t(4;11) and t(1;19) had a significantly superior survival with HSCT, as compared to chemotherapy alone [30]. Given that patients with MLL rearrangements remain a high-risk group with pediatric-based regimens [18•], this is a reasonable recommendation for younger ALL patients with these abnormalities. For those considered at high risk, unrelated HSCT is another option, and recent data suggest the OS with closely matched unrelated donors is comparable to that of matched sibling transplants [31].

For adult patients who relapse, the prognosis is dismal, with failure rates approaching 100% using conventional therapy. HSCT is the only approach to date which has been capable of salvaging such patients. However, studies have shown that salvage rates are low; MRC/ECOG data showed a 5-year OS of 23% in patients undergoing matched sibling HSCT following relapse, and only 16% with unrelated HSCT [32]. Therefore, the identification of patients at higher risk of relapse in first CR is of major importance.

Ph-Positive ALL

The Philadelphia chromosome, resulting in BCR-ABL transcripts, is present in approximately 8–10% of adolescents and 15–20% of young adults with ALL. It has been recognized for many years as a poor prognostic feature; despite CR rates exceeding 80%, most patients relapse, with median DFS of about 10 months and 5-year survival below 20%, using chemotherapy alone [33, 34]. HSCT has been widely used for young patients in CR-1, and most studies demonstrate a survival advantage compared to chemotherapy alone [33–35].

Most protocols for young patients now incorporate a BCR-ABL–targeted tyrosine kinase inhibitor (TKI), such as imatinib or dasatinib, into combination chemotherapy protocols. A number of studies using these combinations have been published. The M.D. Anderson group, combining imatinib with hyper-CVAD, reported CR rates of 93% [36]; the 3-year OS was 66% for patients undergoing SCT in CR versus 49% for non-transplanted patients. These results were substantially superior to their retrospective results with chemotherapy alone. The French GRAAPH-2003 study obtained similar results combining imatinib with induction and post-remission therapy [37]; results were also superior to historical results with LALA-94 in the pre-imatinib era. An Italian study reported 5-year OS of 38% and DFS of 39% with imatinib combined with chemotherapy; these results were also significantly superior to a historical cohort receiving chemotherapy alone [38]. Finally, the UKALL/ECOG2993 trial recently reported an improved 3-year OS of 42% after the addition of imatinib to treatment versus 25% in the pre-imatinib era [39].

Since a number of relapses were associated with the emergence of BCR-ABL mutations potentially sensitive to newer TKIs, dasatinib has been combined with chemotherapy in younger patients [40]. Although similarly high early CR rates have been reported, relapses have occurred, in many cases associated with the emergence of T315I mutations, and it is unclear whether results are superior to those with imatinib plus chemotherapy.

A number of other questions remain, particularly the role of allogeneic HSCT in the era of TKIs. It appears that the use of TKIs, by increasing CR rates and duration, permits a higher proportion of patients to proceed to HSCT [37]. However, HSCT is still hampered by transplant-related mortality, in the range of 20–30% [38]. Nevertheless, in a retrospective comparison, the UKALL/ECOG study found a 60% 3-year OS in patients undergoing HSCT, as compared to 28% in those treated with chemotherapy + imatinib without HSCT [39]. Most experts still recommend HSCT for younger Ph + ALL patients in CR-1 who have a suitable related or unrelated donor.

Another unanswered question is whether combining TKIs with a pediatric-based chemotherapy regimen would result in improved outcomes. Our center has been combining imatinib with our PMH pediatric-based regimen. Although CR rates and OS appear at least comparable to those reported with other regimens, we have noted unusually high rates of neurotoxicity and hepatotoxicity, suggesting possible interactions with vincristine and asparaginase, respectively (manuscript in preparation). Based on our findings we have deleted asparaginase from our Ph + ALL protocol and are substituting vinblastine for vincristine in the intensification phase.

Management of Treatment-Related Complications

The improved treatment outcomes with pediatric-based regimens in adolescents and young adults have come with a cost, namely treatment-related toxicities, some of which are substantial. Some of these, such as infections, are inherent to all intensive multi-agent chemotherapy regimens, while others are related to the specific agents which have been intensified in the pediatric regimens.

Venous thromboembolism (VTE) is an increasingly common complication, likely related to asparaginase-induced depletion of antithrombin III. The Dana Farber group recently reported an incidence of 19% in patients age 10–20, and 25% in those age 20–30 [41]. The most common sites, in order of frequency, are central venous catheters/upper extremities, lower extremities, and sinus vein [41]. The Princess Margaret group found an overall VTE incidence of 23% during the weekly asparaginase intensification phase [18•]. Because of this, we have now instituted routine prophylaxis with low-molecular weight heparin (LMWH) during this phase in all patients. There are retrospective data in children with ALL which indicate that LMWH may reduce the frequency of asparaginase-related VTE in high-risk patients [42], but data are lacking in adults. Asparaginase can be safely restarted once therapeutic anticoagulation is instituted and the clot has stabilized [18•, 41].

Osteonecrosis (ON) has emerged as a common complication, likely related to the high doses of corticosteroids. The UKALL XII study found a 10-year incidence of 29% in patients age 15–20, as compared to only 8% in adults > age 20 [43]. However, using the dexamethasone-intensive pediatric-based regimen, we found a 32% incidence of MRI-documented ON in adults [18•]. The most frequent site is the femoral head, but ON involving the femoral shaft, knee, shoulder, ankle, and elbow has been seen [18•, 43]. These have resulted in major morbidity, particularly when involving more than 30% of the femoral head; such patients have an 80% risk of subsequent collapse, resulting in debilitating pain and frequently requiring arthroplasty [44].

The pathogenesis of corticosteroid-induced ON is unclear. One study in childhood ALL found a correlation between ON and certain plasminogen activator-1 (PAI-1) polymorphisms [45]. Other studies indicate that corticosteroids induce endothelial cell activation, leukocyte adhesion, and expression of genes expressing procoagulant molecules [46, 47], which may promote vascular occlusion leading to ON. A small retrospective study of ON in childhood ALL suggested that bisphosphonates may reduce symptoms and improve function but do not influence radiologic progression [48]. Potential strategies to prevent or reduce the severity of ON will depend on further studies to better delineate the underlying mechanisms, as well as evaluating potential biomarkers for early detection.

Vincristine-related peripheral neuropathy is also common [18•]. This may necessitate reduction or omission of the vincristine, although we have found that substitution with vinblastine can prevent further progression of the neuropathy and does not adversely impact prognosis [18•]. Asparaginase-related acute pancreatitis is seen in up to 5% of cases [11, 18•]. Corticosteroid-induced hyperglycemia and myopathy, and liver enzyme elevations secondary to 6-mercaptopurine, methotrexate, and asparaginase, are also common. Hypersensitivity reactions to E. coli asparaginase can be dealt with by substituting Erwinia asparaginase; according to one recent report, this substitution does not appear to adversely affect prognosis, at least in children [49]. Pegylated asparaginase is now being increasingly used in pediatric-based protocols, permitting a reduced frequency of administration; results to date suggest at least equivalent efficacy and toxicity compared to the E. coli formulation [50].

The diagnosis of ALL, and superimposed treatment-related toxicities, have a profound psychological impact on adolescents and young adults. Being faced with a life-threatening illness, at a time when they are just beginning to plan and develop their future lives, can lead to severe anxiety and depression. At a time when body self-image is of great importance, alopecia and corticosteroid-related changes in body habitus can affect their confidence and sense of well-being. Other toxicities already mentioned, such as ON and neuropathy, as well as prolonged fatigue and nausea, can have major impacts on quality of life. Fertility issues also come into play, particularly with alkylator-containing regimens. These impacts have not been well-studied to date but, as cure rates improve, these issues are assuming much greater importance and require further study. A current US cooperative group phase 2 study in adolescents and young adults (C-10403) is evaluating psychosocial issues, in addition to clinical outcomes, in these patients.

New Approaches

MRD Detection

The difficulty in salvaging relapsed patients has led to attempts to identify patients in CR-1 with otherwise standard-risk factors who are at higher risk of relapse. The most promising approach is minimal residual disease (MRD) detection. There are two major techniques to evaluate MRD, molecular and immunophenotypic. Molecular monitoring may involve the detection of specific fusion genes by real-time quantitative polymerase chain reaction (RT-PCR); this technique is well established for BCR-ABL and MLL rearrangements. For the majority of ALL cases where these are not present, MRD monitoring involves identification of the junctional regions of the rearranged immunoglobulin or T-cell receptor gene for each patient, followed by sequencing to create allele-specific oligonucleotides [51]. Although this technique is highly sensitive, in the range of 0.01–0.001%, it is also laborious and expensive.

The second technique involves the identification of leukemia-associated immunophenotypes (LAP) using multigated-flow cytometry. Studies in children have shown that LAP can be identified in at least 90% of cases, with a sensitivity of up to 0.01%. One limitation of this technique is that phenotypic switches occurring post-treatment may escape detection.

MRD monitoring has been extensively studied recently in children and adolescents with ALL, and has been demonstrated to be of prognostic value [52•, 53]. Bassan et al. [54], using molecular techniques in adult ALL, found a highly significant correlation between MRD and 5-year DFS and OS. The Polish leukemia group, using immunophenotyping, reported that an MRD level ≥0.1% post-induction was an independent predictor of relapse in adults [55]. Although these studies strongly suggest that MRD detection is of prognostic value, it has not yet been demonstrated that early intervention in these high-risk patients with treatment intensification or HSCT can improve the outcome in adults. One small study found that, although clearance of MRD can occur post-HSCT, the relapse rate in MRD + patients is higher as compared to MRD- patients [56].

New Therapies

Several new agents have recently been reported as showing promising results. Approximately 40% of B-ALL cases are CD20-positive. The German GMALL Group recently reported results from a phase 2 study combining rituximab with chemotherapy in adults with CD20+ ALL. Compared to historical controls receiving chemotherapy alone, the group receiving rituximab achieved a higher rate of molecular CR at 16 weeks, as well as a superior 5-year continuous CR rate and OS [57]. However, this has not yet been studied in a randomized prospective manner. The M.D. Anderson group has also reported superior results in CD20+ ALL with the addition of rituximab to their hyper-CVAD protocol [58].

Blinatumomab is a bispecific tumour-engaging (BiTE) antibody, with anti-CD19 and CD3 components. Topp et al. [59] treated 21 adult ALL patients with MRD + post-chemotherapy; 16 became MRD-, and 78% remained MRD- at 1-year follow-up. Further studies are in progress with this promising agent.

Conclusions

Despite the many challenges, there appear to have been significant recent advances in the therapy of adolescents and young adults with ALL. The most notable is the more widespread adoption of pediatric-inspired protocols in Ph-negative patients. Despite the absence of randomized controlled trials, the weight of evidence suggests that results are superior as compared to adult-based protocols. Standard-risk patients treated with these protocols probably do not require HSCT in CR-1.

Certain high-risk patients do require additional treatment. For Ph + patients, BCR-ABL dependent TKIs such as imatinib or dasatinib should be added to chemotherapy from the time of induction, and continued through post-remission phases. We still recommend HSCT in CR-1 for these patients. Other high-risk patients, such as those with MLL rearrangements, especially t(4;11), and possibly those with t(1;19), should also be referred for HSCT in CR-1. Emerging evidence indicates that MRD + patients post-induction are also high risk and should be considered for additional post-remission strategies in CR-1. However, each center and group needs to determine the appropriate threshold and timing for MRD detection using their particular regimen and monitoring technique.

Future approaches should involve attempts to reduce short and long-term toxicities associated with treatment using pediatric-based regimens as well as HSCT-associated morbidity and mortality. More work is also needed to determine the optimal ways of eliminating MRD using HSCT and non-transplant approaches.