however, with varied odds ratio (OR) in terms of different ethni-
cities, ages, and ALL subtypes. For instance:1.ARID5BandPIP4K2ASNPs exhibit consistent significance in
all ethnicities, while CDKN2A SNPs are only significant in
Caucasians, due to the ethnicity-specific allele frequency.- Although significant in all ethnicities forARID5BSNPs, both
odds ratios and risk allele frequencies of rs10821936 for ALL
susceptibility increase in the order of Africans, Caucasians, and
Hispanics [59, 95], which is in parallel with ALL incidence
among ethnic/racial groups (incidence rate: Africans<Cauca-
sians<Hispanics) [96, 97]. - Age at diagnosis was also considered to be another impact factor,
and higher RAF of topARID5Bsignal has been observed in
younger patients in an ethnicity- and subtype-independent man-
ner [59, 98], which is opposite to SNP inGATA3[90, 93, 99]. - Risk alleles ofARID5BSNPs are enriched in hyperdiploidy
[95, 100], whileGATA3SNPs are enriched in BCR-ABL1-like
subtype [90].
However, the SNPs described above are all located in intronic
or intergenic regions, most of which are considered to be tags for
causal variants through lineage disequilibrium, calling for large
efforts on post-GWAS [101, 102]. In addition, exome array was
also used to identify the ALL-related SNPs that can alter protein
coding, only one common missense variant atCDKN2Awas iden-
tified as the top signal in independent studies [92, 103], and
patients with heterozygous genotype of rs3731249 tend to loss
the expression of wild-type allele in their leukemia cells, either
through loss of heterozygosity or posttranscriptional inactivation
[92, 104]. Although some rare variants can be detected in the gene-
based burden test, no causal variants from exome array can explain
the association of the common SNPs inARID5B, etc. New theory
indicates that most of the causal variants for GWASs tend to induce
the alteration of gene expression through interrupting the tran-
scription factors binding on promoters or enhancers. For ALL, two
potential causal variants have been revealed forARID5B and
CDKN2B, respectively [105, 106].
Array-based GWAS can also be used for identifying the genetic
basis of drug response, especially for adverse drug reactions. As
described above, variant allele of SNPs inTPMTwas found to
induce loss of enzyme activity, thus resulting in high toxic metabo-
lite content in plasma. However, frequencies of these SNPs are
higher in Caucasians and blacks than East Asians, who paradoxically
have a higher rate of mercaptopurine-induced myelosuppression
[107]. With GWAS studies for mercaptopurine dosage adjustment,
two SNPs reached genome-wide significance with one indicating
TPMT*3C and a novel missense SNP inNUDT15[108]. With398 Heng Xu and Yang Shu