Yet another susceptibility variant for ALL: what’s next?

In this issue of Blood, Qian and colleagues¹  add ERG variants to a growing list of common host DNA polymorphisms that have been associated with an increased risk of childhood acute lymphoblastic leukemia (ALL). It has long been known that Hispanics more frequently develop childhood ALL. In addition to GATA3, PIP4K2A, and ARID5B, this increased risk is now also accounted for by ERG variants with the frequency of risk variants being positively related to the proportion of Native American Ancestry.

Susceptibility for ALL covers a wide continuum that ranges from rare, but highly penetrant, cancer prone syndromes with a 10- to 100-fold (or even more) increased risk of ALL to common germline DNA variants that mediate only a modest (1.2- to 3.0-fold) increased odds ratio.²  The former group includes both syndromes dominated by their nonmalignant phenotype, such as Down syndrome and ataxia telangiectasia, as well as pure cancer syndromes, such as Li-Fraumeni syndrome. At the other end of the spectrum, the risk genes include ARID5B, PIP4K2A, IKZF1, CDKN2A, CEBPE, GATA3, and now also ERG. Many of the risk genes associated with ALL are transcription factors involved in hematopoietic development. Several of these are frequently affected by somatic mutations in ALL, such as IKZF1, CDKN2A, and ERG, but the common variants that associate with ALL risk in most cases reside in noncoding regions.

Adding to this, the susceptibility variants and cancer-prone syndromes are often strongly associated with certain subsets of ALL, such as ARID5B variants being associated with high-hyperdiploid ALL, GATA3 variants with Philadelphia-like ALL, and Li-Fraumeni syndrome with hypodiploid ALL. Of interest, the gap between the common susceptibility variants and the rare cancer-prone syndromes is slowly closing with the demonstration that 1% to 3% of ALL patients harbor deleterious coding variants in genes previously only linked to rare cancer-prone syndromes, such as ETV6³  and TP53.⁴ 

Whereas some of the common variants, like those residing in ARID5B, have been associated with an increased risk of ALL across multiple ethnicities, others like ERG seem to be more race restricted. However, it remains uncertain whether this reflects the broader, yet undefined, ethnicity-dependent genomic context within which they mediate their biological effect, or whether it reflects interactions with environmental risk factors for ALL that are influenced by certain behavioral profiles.⁵ 

The mapping of the natural history of ALL has mostly used the ETV6/RUNX1 and high-hyperdiploid ALL subsets as prototypes as they are the most common ALL subsets and furthermore frequently initiated prenatally. Thus, clone-specific markers representing preleukemic cells can be detected in neonatal blood spot samples or Guthrie cards.⁶  The preleukemic cell burden at birth can then taper off, or the preleukemic cells can persist and acquire the additional somatic mutations necessary for development of overt ALL, and likely mediated by activation-induced cytidine deaminase and recombination-activating genes.⁷  Models of these dynamics have been based on epidemiological and animal research that links ALL risk with reduced microbial exposure in early life.⁸  Thus, one of the most consistent epidemiological findings is the association of daycare center attendance with a 20% to 25% reduced risk of childhood ALL.⁹  Importantly, immune system maturation in early postnatal life may not just reflect infectious exposures because abnormal profiles of inflammatory markers can already be detected in neonatal blood spot samples.¹⁰ 

Although traditional genome-wide association study analyses have granted us some insight into the biological pathways involved in leukemogenesis, a deeper understanding of childhood leukemia development will require integration of large-scale screening of cord blood samples,^6,10  mapping the infectious burden in early life through population-based registers that provide data on known risk factors for ALL, such as birth weight, birth order, sibship size, and daycare center attendance, as well as hospital admissions and antibiotic use, and linkage of these data with genomic profiling of both host and tumor DNA.