Acute leukemia is the most common malignancy diagnosed in children. An estimated 3,000 cases of leukemia in children younger than 15 years occur in the United States each year. Based on mortality statistics, the overall incidence is estimated between 35 and 49 per million children younger than 15 years. Approximately three-fourths of the cases are acute lymphoblastic leukemia (ALL), and most of the remaining cases of leukemia are acute nonlymphocytic leukemia. In the United States, childhood ALL has a peak incidence approaching 80 per million among white children between 2 and 5 years of age, with a less prominent peak approaching 35 per million in black children. The highest rates have been reported among Hispanics, Filipinos, and Chinese, and the lowest rates among African-Americans. The reason for this difference is unexplained. Childhood ALL occurs more frequently in boys than in girls, and the difference increases with age. Geographic variation in incidence, rates, and subtype of leukemia (i.e., leukemic clusters) has been reported in the United States and worldwide.

The molecular basis for leukemic transformation in humans is unknown. In normal bone marrow, undifferentiated pluripotent progenitor cells with a capacity for self-renewal give rise to committed progenitor cells. These are morphologically recognizable cells that give rise to the erythroid, myeloid, megakaryocytic, eosinophilic, and monocytic-macrophage series. In the clonal expansion theory, leukemia arises from a damaged progenitor cell that has the propensity for unlimited self-renewal or has lost the ability to differentiate along the lines of normal committed progenitor cells. Proposed molecular mechanisms for leukemic induction include activation of a protooncogene or the creation of a fusion gene with oncogenic properties, or the loss or inactivation of genes whose proteins suppress leukemia.

A variety of environmental, genetic, viral, and immunologic factors may contribute to the development of disease (Table 300.1). Of the possible environmental factors, ionizing radiation has been the most extensively studied. The increased incidence of leukemia in survivors within 1,000 meters of the atomic bomb explosions during World War II has
been well documented. ALL developed most frequently in those younger than 15 years at the time of exposure. Exposure to radiation in utero, usually from diagnostic examinations, has been associated with a small but statistically significant risk for development of childhood leukemia. Radiation dose and inherited susceptibility play contributing roles in the development of radiation-associated malignancies. The relative risk of contracting leukemia or other forms of cancer in persons exposed to electromagnetic fields has been investigated in several studies, but no definitive cause and effect association has been established.

Several chemical agents are known to induce or promote leukemia in animals. Except for benzene exposure, however, little is known about the importance of such agents in humans. With the exception of second malignancies in children previously treated with chemotherapy, usually including alkylating agents or epipodophyllotoxins, with or without radiation therapy, no clear association for chemical carcinogenesis has been established in children.

Based on studies in monozygotic twins, a prenatal origin of childhood leukemia has been proposed. Using neonatal blood spots from Guthrie cards, investigators were able to demonstrate the presence of t(12;21)(p13;q22) TEL/AML1 fusion sequences from six of nine infants and a pair of twins who subsequently developed lymphoblastic leukemia. This important leukemic-associated clonal mutation has now been demonstrated to occur prenatally; however, postnatal events are required to promote clonal expansion to the clinical disease state.

Several hypotheses have been proposed to suggest a role for infection, particularly in utero or during early infancy as a leukemogenic risk factor. No single infectious agent emerges as a clear risk factor. Socioeconomic status, geographic isolation with sudden shifts in population mix or density, and other community characteristics have been suggested as cofactors contributing to abnormal patterns of infection that lead to an increase risk for leukemia.

An unusual susceptibility to leukemia has been associated with certain heritable diseases, chromosomal disorders, and constitutional syndromes. Children with trisomy 21 (i.e., Down syndrome) have at least a 10- to 15-fold increased risk for developing leukemia compared with normal children. The greatest risk period for leukemia in children with Down syndrome is before the age of 5 years. Before 3 years of age, acute nonlymphoblastic leukemia predominates. Increased chromosomal fragility may predispose these patients to leukemic transformation. Cases of childhood leukemia have been associated with several heritable syndromes, including Klinefelter syndrome, Rubinstein-Taybi syndrome, Poland syndrome, Shwachman syndrome, neurofibromatosis, and Kostmann congenital agranulocytosis. The relation between these syndromes and leukemia and other childhood cancers requires further definition.

Several immunodeficiency states have an associated increased risk for lymphoma and leukemia. These conditions include the syndromes of Wiskott-Aldrich, X-linked agammaglobulinemia, severe combined immune deficiency, and ataxia-telangiectasia. The loss of cellular immune surveillance capability for tumor antigens and the inability to self-regulate lymphoproliferative processes may contribute to malignant transformation in these patients. When a child with an identical twin develops leukemia, the risk for leukemia in the other twin is approximately 20%, but the risk diminishes with age. Fraternal twins and siblings of children with leukemia have an estimated fourfold greater risk for leukemia than children in the general population. However, an annual risk that increases from 4 per 100,000 to 16 per 100,000 is not of major clinical significance and should not be a cause for alarm for parents. Epidemiologic studies do not indicate an increased frequency of leukemia in children of leukemic parents, in children breast-fed by mothers who subsequently develop leukemia, in recipients of blood products from donors who develop leukemia, or in households with pets with leukemia.

Childhood leukemia is a heterogenous disease. Morphologic, immunologic, biochemical, and cytogenetic features are used to characterize the disease, estimate prognosis, and develop successful therapeutic strategies. Under normal conditions, less than 5% of the nucleated marrow is composed of blast forms. Blasts are primitive, undifferentiated-appearing precursor cells not normally seen in the peripheral circulation, except in unusual circumstances. With the Wright-Giemsa stain, blasts can be recognized by their large size and high nuclear-to-cytoplasmic ratio. The nuclear membrane is round or clefted, and the nuclear chromatin appears fine and homogeneous with an occasional small nucleolus. The leukemic lymphoblast frequently reacts with the periodic acid-Schiff stain but not with myeloperoxidase and Sudan black.

Using a morphologic classification system developed by a French-American-British (FAB) collaboration, approximately 85% of the children with ALL have lymphoblasts of L1 morphology. Fewer than 15% of the patients have lymphoblasts of L2 morphology. Lymphoblasts with L3 morphology are identical to Burkitt lymphoma cells, ordinarily possess surface immunoglobulin, and are associated with a distinct karyotypic abnormality. Specific immunologic phenotypes have not otherwise been associated with L1 or L2 morphology. The FAB classification may have some prognostic value. Childhood ALL with L1 morphology has a high remission induction rate and more prolonged survival, whereas patients with L3 disease have a worse prognosis.

Immunologic marker analysis has allowed lineage assignment and maturational staging of the lymphoid leukemias and has offered some insight into the pathology of these diseases (Table 300.2). Approximately 65% of the children with ALL
have early pre–B lymphoblasts. Using monoclonal antibodies directed at specific antigen sites defined as clusters of differentiation (CD), these lymphoblasts were found to express CD10, CD19, CD20, CD22, and the HLA-DR antigen (Ia). These lymphoblasts lack surface (sIg) and cytoplasmic (cIg) immunoglobulins, and do not react with monoclonal antibodies directed at T-cell antigens.

Pre–B-cell ALL represents approximately 18% to 20% of the new cases of ALL. Morphologic and immunologic features are similar to the early pre–B-cell ALL, except for the presence of heavy-chain (cIg) immunoglobulin within the cytoplasm.

B-cell ALL (B-ALL) is rare in children, representing 1% of all cases. The lymphoblasts are characterized by their Burkitt-like appearance and express sIg.

T-cell phenotypes represent 13% to 15% of childhood ALL. Monoclonal antibodies corresponding to different stages of intrathymic differentiation are used to identify these patients, with one-third to one-half of T-ALL cases reacting with antigens of the early thymocyte state (i.e., CD2, CD5, CD7).

Approximately 1% to 3% of patients fail to react with any antigen test system and are classified as undifferentiated, null, or stem cell leukemias.

The immunologic subtypes may be important for predicting response to conventional therapy. Patients with early pre–B-ALL experience an increased remission induction rate and prolonged remission and survival. Patients with pre–B-ALL and the t(1;19)(q23;p13) translocation may not enjoy the same degree of long-term disease control as patients with early pre–B-ALL. Patients with T-cell disease are frequently older (average age, 8 to 12 years) and are boys (male-to-female ratio, 4:1). They frequently present with a leukocyte count of more than 100,000/μL, a mediastinal mass, normal hemoglobin concentration, hepatosplenomegaly, and adenopathy. This disease is more difficult to treat and cure than other forms. Children with B-ALL have an aggressive leukemia (cell doubling time, 24 hours) and require very intensive therapies to achieve a cure. Infants (less than 1 year of age) frequently fail to express reactivity to any lymphoid antigens, are CD10 negative, and have a poor prognosis.

Approximately two-thirds of the patients with ALL have karyotypic abnormalities involving the leukemic cell. These alterations are broadly defined as changes in chromosome number (i.e., ploidy) or chromosome structure (i.e., translocations, deletions, inversions). Ploidy can be determined by classic enumeration of chromosome number from metaphase preparations or by analysis of DNA content by flow cytometry. Prognostic significance has been suggested for certain cytogenetic subgroups. Patients with hyperdiploidy (more than 53 chromosomes per cell) without structural abnormalities and patients with trisomy of chromosomes 4, 10, and/or 17 have a more favorable prognosis with conventional therapy than other groups. Most newly diagnosed ALL patients have a diploid or near diploid chromosome complement. Patients with pseudodiploid or hypodiploid chromosome numbers have a poor prognosis. Multiple leukemic stem lines occur in approximately 9% to 15% of patients. The significance of these complex chromosome combinations is not understood.

Several specific chromosome translocations have been recognized in childhood ALL and have significance for disease ontogeny and clinical outcome (Table 300.3). The genetic basis for ALL is presented in Box 300.1.