First Relapse, Second Chance: Determinants of Survival After First Relapse in Pediatric Acute Lymphoblastic Leukemia

Improvements in survival among children and adolescents and young adults (AYA) with acute lymphoblastic leukemia (ALL) are attributable to increased understanding of leukemia biology and well-defined prognostic factors that dictate risk-adapted upfront therapy, including age, presenting white blood cell count, central nervous system (CNS) involvement, cytomolecular subtype, and end-of-induction (EOI) measurable residual disease (MRD).¹  Historically, children and AYAs with ALL who experience relapse remain a challenge to treat.^2,3  And at relapse diagnosis, prognostic factors associated with overall survival (OS) are not as well defined. A prior Children’s Oncology Group (COG) study detailing outcomes for relapsed ALL patients enrolled in clinical trials between 1988 and 2002 reported a 20.5% relapse rate and a five-year OS of 36% for the entire cohort (37.5% for B-cell ALL [B-ALL]),⁴  consistent with outcomes in other international groups.^5,6  Together, these studies have identified a few general prognostic factors in the relapse setting, including immunophenotype (B vs. T), time to relapse, site of relapse, and MRD after re-induction; however, evaluation of additional clinical, cytogenetic, and biological risk factors have been limited due to small numbers and inadequate data.

Susan R. Rheingold, MD, and colleagues recently analyzed the largest dataset to date of newly diagnosed pediatric and AYA patients (n=16,115) with B-ALL, T-cell ALL (T-ALL), and infant B-ALL (diagnosed <1 year of age) enrolled in 12 frontline COG clinical trials between 1996 and 2014, with the goal of describing clinical features at first relapse and identifying prognostic indicators for survival after relapse. In this updated cohort, overall relapse rates were lower at 12.7%. Among 13,771 B-ALL patients, 1,715 (12.5%) relapsed, a significantly higher percentage of whom were classified as National Cancer Institute (NCI) high risk (19%, 754/3,974) as compared to NCI standard risk (9.8%, 961/9,797; p>0.0001). Among 2,109 T-ALL patients, 227 (11.2%) relapsed, while 111 of 325 infants with B-ALL (34.2%) relapsed, including 90 of whom had KMT2A-rearranged (KMT2A-r) ALL.

Relapse risk was significantly higher for males, infants, and patients aged 10 years or older at initial diagnosis, with further increased risk in those aged 16 or older. Leukemia phenotype had different distribution of relapse sites, as well as time to relapse. Half of B-ALL relapses occurred late (defined as >36 months from diagnosis; median time to relapse of 34.6 months), with 72.5% involving the bone marrow and 32.9% involving CNS. This differed from T-ALL patients, among whom 64.8% experienced early relapse (defined as <18 months from diagnosis) and 82% relapsed in less than 36 months from diagnosis (median time to relapse of 13.8 months), with 56.5% involving bone marrow and 47.1% involving CNS. T-ALL patients had higher rates of isolated CNS (iCNS) relapse (31.4% vs. 21.7% in B-ALL patients), with more than 80% of these occurring less than 18 months from diagnosis. Notably, 67.9% of these patients had CNS-3 status at original diagnosis and received cranial radiation as part of their frontline regimen. Among the infant B-ALL cohort, median time to relapse was 13.9 months, with 77.8% involving bone marrow and 23.1% involving CNS.

For all relapsed patients, the five-year OS from relapse diagnosis was 48.9%±1.2% (52.5%±1.3% for B-ALL and 35.5%±3.3% for T-ALL; Figure). Unfortunately, this study demonstrated no improvement in survival for infants with B-ALL compared to the prior cohort, with OS at 21.5%±3.9%, an even more dismal OS of 9.0±3.2% for those with early relapse, and no survivors among infants who relapsed prior to 1 year of age (Figure).

This study identified additional risk factors for B-ALL patients that independently predicted survival outcomes. Although EOI MRD greater than or equal to 0.01% during frontline therapy has previously been associated with increased relapse risk, this study demonstrated that on multivariable analysis, EOI MRD of greater than or equal to 0.1% was associated with shorter time to relapse and inferior five-year OS after relapse of 37.7±2.6%, compared to 58.2±1.6% in those who had MRD of less than 0.1% (p<0.005; Table). Cytogenetic subtype in B-ALL patients was also found to be an important prognostic factor for survival that was closely related to time to relapse, supporting leukemia biology as a driver of clinical behavior. Post-relapse OS for B-ALL was significantly better for those with favorable cytogenetics, including a five-year OS of 74.4±3.1% for ETV6-RUNX1-positive ALL and 70.2±3.6% for patients with simultaneous trisomies of chromosomes 4 and 10 (median times to relapse of 43 months for both subtypes), compared to those with unfavorable cytogenetics, including a five-year OS of 14.2±6.1% for those with hypodiploid ALL and 31.9±7.7% for those with KMT2A-r ALL (median times to relapse of 12.5 and 18 months, respectively; Figure, Table).