Is There Etiologic Heterogeneity between Subtypes of Childhood Acute Lymphoblastic Leukemia? A Review of Variation in Risk by Subtype

Abstract

Although substantial advances in the identification of cytogenomic subtypes of childhood acute lymphoblastic leukemia (ALL) have been made in recent decades, epidemiologic research characterizing the etiologic heterogeneity of ALL by subtype has not kept pace. The purpose of this review is to summarize the current literature concerning subtype-specific epidemiologic risk factor associations with ALL subtype defined by immunophenotype (e.g., B-cell vs. T-cell) and cytogenomics (including gross chromosomal events characterized by recurring numerical and structural abnormalities, along with cryptic balanced rearrangements, and focal gene deletions). In case-control analyses investigating nongenetic risk factors, home paint exposure is associated with hyperdiploid, MLL-rearranged, and ETV6-RUNX1 subtypes, yet there are few differences in risk factor associations between T- and B-ALL. Although the association between maternal smoking and ALL overall has been null, maternal smoking is associated with an increasing number of gene deletions among cases. GWAS-identified variants in ARID5B have been the most extensively studied and are strongly associated with hyperdiploid B-ALL. GATA3 single nucleotide variant rs3824662 shows a strong association with Ph-like ALL (OR = 3.14). However, there have been relatively few population-based studies of adequate sample size to uncover risk factors that may define etiologic heterogeneity between and within the currently defined cytogenomic ALL subtypes.

Figure 1

contains childhood acute lymphoblastic leukemia (ALL) incidence rates from the Surveillance, Epidemiology, and End Results program (SEER) 18 registries (2000–2014) for all ages (0–19 years) and stratified by age group (<1 year, 1–9 years, 10–19 years) (a). The cytogenomic subtypes of childhood B-ALL immunophenotype overall and by age at diagnosis categories (<1 year, 1–9 years, 10–19 years) are presented (b). The cytogenomic subtypes of childhood T-ALL immunophenotype overall are presented (c). Categories assumed to be mutually exclusive.

Figure 2

displays the odds ratios and 95% confidence intervals for associations by childhood acute lymphoblastic leukemia (ALL) immunophenotype and cytogenomic subtype for maternal and paternal exposures during the preconception period (a). Next, maternal and paternal exposures during pregnancy in association with childhood ALL by subtype are presented (b). Then, maternal, paternal, and birth characteristics of the child in association with childhood ALL by subtype are presented (c). Finally, childhood exposures in association with childhood ALL by subtype are displayed (d). Reference numbers are presented in brackets.

Figure 3

displays the allelic odds ratios and 95% confidence intervals for single nucleotide variants (SNVs) associated with childhood acute lymphoblastic leukemia (ALL) overall and by immunophenotype and cytogenomic subtype from discovery genome-wide association studies (GWAS) and replication analyses. Reference numbers are presented in brackets. Estimates reported for Caucasians only. Non-HD refers to non-hyperdiploid ALL.