Children with acute lymphoblastic leukemia (ALL) are usually treated according to risk groups defined by both clinical and laboratory features. The intensity of treatment required for favorable outcome varies substantially among subsets of children with ALL. Risk-based treatment assignment is utilized in children with ALL so that patients with favorable clinical and biological features who are likely to have a very good outcome with modest therapy can be spared more intensive and toxic treatment, while a more aggressive, and potentially more toxic, therapeutic approach can be provided for patients who have a lower probability of long-term survival.[1,2,3] Certain ALL study groups use a more or less intensive induction regimen based on a subset of pretreatment factors, while other groups give a similar induction regimen to all patients. All groups modify the intensity of postinduction therapy based on a variety of prognostic factors.
Risk-based treatment assignment requires the availability of prognostic factors that reliably predict outcome. For children with ALL, a number of clinical and laboratory features have demonstrated prognostic value, some of which are described below. The factors described are grouped into the following categories:
As in any discussion of prognostic factors, the relative order of significance and the interrelationship of the variables are often treatment dependent and require multivariate analysis to determine which factors operate independently as prognostic variables.[5,6] Because prognostic factors are treatment dependent, improvements in therapy may diminish or abrogate the significance of any of these presumed prognostic factors.
A subset of the prognostic and clinical factors discussed below is used for the initial stratification of children with ALL for treatment assignment. At the end of this section are brief descriptions of the prognostic groupings currently applied in ongoing clinical trials in the United States.
- Age at diagnosis
Age at diagnosis has strong prognostic significance, reflecting the different underlying biology of ALL in different age groups. Young children (aged 1-9 years) have a better disease-free survival (DFS) than older children, adolescents, or infants.[1,7,8] The improved prognosis in younger children is at least partly explained by the more frequent occurrence of favorable cytogenetic features in the leukemic blasts including hyperdiploidy with 51 or more chromosomes and/or favorable chromosome trisomies, or the ETV6-RUNX1 (t[12;21], also known as the TEL-AML1 translocation).[7,9] The outcome for adolescents has improved significantly over time.[10,11,12] Multiple retrospective studies have suggested that adolescents aged 16 to 21 years have a better outcome when treated on pediatric versus adult protocols.[13,14,15] (For more information about adolescents with ALL, see the Postinduction Treatment for Childhood Acute Lymphoblastic Leukemia Subgroups section of this summary.)
Infants with ALL have a particularly high risk of treatment failure. Treatment failure is most common in infants younger than 6 months and in those with extremely high presenting leukocyte counts and/or a poor response to a prednisone prophase.[16,17,18,19] Infants with ALL can be divided into two subgroups on the basis of the presence or absence of translocations that involve the MLL gene located at chromosome 11q23.[18,19,20] Approximately 80% of infants with ALL have an MLL gene rearrangement.[18,20,21] The rate of MLL gene translocations is extremely high in infants younger than 6 months; from 6 months to 1 year the incidence of MLL translocations decreases but remains higher than that observed in older children. Infants with leukemia and MLL translocations have very high white blood cell (WBC) counts, increased incidence of central nervous system (CNS) involvement, and a poor outcome.[18,19] Blasts from infants with MLL translocations are typically CD10 negative and express high levels of FLT3.[18,19,20,22] Conversely, infants whose leukemic cells show a germline MLL gene configuration frequently present with CD10-positive precursor-B immunophenotype. These infants have a significantly better outcome than infants with ALL characterized by MLL translocations.[18,19,20] Infants diagnosed within the first month of life have higher WBC counts, higher incidence of MLL translocations, significantly higher relapse rate, and poorer overall survival compared with infants older than 1 month at diagnosis.
- WBC count at diagnosis
Patients with B-precursor ALL and high WBC counts at diagnosis have an increased risk of treatment failure compared with patients with low initial WBC counts. A WBC count of 50,000/�L is generally used as an operational cut point between better and poorer prognosis, although the relationship between WBC count and prognosis is a continuous rather than a step function. There are conflicting data regarding the prognostic significance of presenting leukocyte counts in T-cell ALL.[6,24,25,26,27,28,29]
- CNS involvement at diagnosis
The presence or absence of CNS leukemia at diagnosis has prognostic significance. Patients who have a nontraumatic diagnostic lumbar puncture may be placed into one of three categories according to the number of WBC/�L and the presence or absence of blasts on cytospin as follows:
- CNS1: Cerebrospinal fluid (CSF) that is cytospin negative for blasts regardless of WBC count.
- CNS2: CSF with fewer than five WBC/�L and cytospin positive for blasts.
- CNS3 (CNS disease): CSF with five or more WBC/�L and cytospin positive for blasts.
Compared with patients classified as CNS1 or CNS2, children with ALL who present with CNS disease (i.e., CNS3) at diagnosis are at a higher risk of treatment failure (both within the CNS and systemically). The adverse prognostic significance associated with CNS2 status, if any, may be overcome by the application of more intensive intrathecal therapy, especially during the induction phase.[30,31] A traumatic lumbar puncture (?10 erythrocytes/�L) that includes blasts at diagnosis appears to be associated with increased risk of CNS relapse and indicates an overall poorer outcome.[30,32] To determine whether a patient with a traumatic lumbar puncture (with blasts) should be treated as CNS3, the Children's Oncology Group (COG) uses an algorithm relating the WBC and red blood cell counts in the spinal fluid and the peripheral blood.
- Testicular involvement at diagnosis
Overt testicular involvement at the time of diagnosis occurs in approximately 2% of males, most commonly in T-cell ALL. In early ALL trials, testicular involvement at diagnosis was an adverse prognostic factor. With more aggressive initial therapy, however, it does not appear that testicular involvement at diagnosis has prognostic significance.[34,35] For example, the European Organization for Research and Treatment of Cancer (EORTC, [EORTC-58881]) reported no adverse prognostic significance for overt testicular involvement at diagnosis. The role of radiation therapy for testicular involvement is unclear. A study from St. Jude Children's Research Hospital (SJCRH) suggests that a good outcome can be achieved with aggressive conventional chemotherapy without radiation. The COG has also adopted this strategy for boys with testicular leukemia that resolves completely by the end of induction therapy. COG considers patients with testicular involvement to be high risk regardless of other presenting features, but most other large clinical trial groups in the United States and Europe do not consider testicular disease to be a high-risk feature.
- Down syndrome (trisomy 21)
Outcome in Down syndrome children with ALL has generally been reported as somewhat inferior to outcomes observed in non-Down syndrome children.[36,37,38,39] The lower event-free survival (EFS) and overall survival (OS) of children with Down syndrome appear to be related to higher rates of treatment-related mortality and the absence of favorable biological features.[36,37,38,39,40] Patients with Down syndrome and ALL have a significantly lower incidence of favorable cytogenetic abnormalities such as ETV6-RUNX1 or trisomies of chromosomes 4 and 10. In a report from the COG, among B-precursor ALL patients who lacked MLL translocations, BCR-ABL1, ETV6-RUNX1, or trisomies of chromosomes 4 and 10, the EFS and OS were similar in children with and without Down syndrome.
In some studies, the prognosis for girls with ALL is slightly better than it is for boys with ALL.[41,42,43] One reason for the better prognosis for girls is the occurrence of testicular relapses among boys, but boys also appear to be at increased risk of bone marrow and CNS relapse for reasons that are not well understood.[41,42,43] However, in clinical trials with high 5-year EFS rates (>80%), outcomes for boys are closely approaching those of girls.[31,44]
Survival rates in black and Hispanic children with ALL have been somewhat lower than the rates in white children with ALL.[45,46] This difference may be therapy-dependent; a report from SJCRH found no difference in outcome by racial groups. Asian children with ALL fare slightly better than white children. The reason for better outcome in white and Asian children compared with black and Hispanic children is at least partially explained by the different spectrum of ALL subtypes. For example, blacks have a higher incidence of T-cell ALL and lower rates of favorable genetic subtypes of ALL. However, these differences do not completely explain the observed racial differences in outcome.