Interaction between birth characteristics and variants in the childhood lymphoblastic leukemia risk.

Background: The incidence rate of childhood acute lymphoblastic leukemia (ALL) differs worldwide, and the interplay between hemostasis actors and the maladaptive responses to environmental exposures has been explored. It has been proposed that endogenous cortisol, induced by different triggers, would eliminate pre-leukemic clones originated . Herein, we tested if the interaction between , and (players in glucocorticoid secretion) and birth characteristics would be associated with ALL risk.

Methods: Children aged <10 years were enrolled within the EMiLI project (period: 2012 to 2020). The study had three steps: (1) observational analysis of birth characteristics ( = 533 cases and 1,603 controls); (2) genotyping to identify single-nucleotide variants ( = 756 cases and 431 controls); and (3) case-only to test gene-environment interactions ( = 402 cases). Genetic syndromes were exclusion criteria. The controls were healthy children. The distribution of the variables was assessed through Pearson's chi-square test. Logistic regression (LR) tests were run fitted and adjusted for selected covariate models to estimate the association risk. Formal interaction analysis was also performed. Genotyping was tested by qPCR with TaqMan probes () or by high-resolution melting ( and ). Hardy-Weinberg equilibrium (HWE) was accessed by the chi-square test. The genotype-risk association was tested in co-dominant, dominant, and recessive models. The gene-environment interaction odds ratio (iOR) was assessed in case-only.

Results: Low birthweight, C-section, and low maternal schooling were associated with increased risk for ALL, adjOR 2.11, 95% CI, 1.02-4.33; adjOR 1.59, 95% CI, 1.16-2.17; and adjOR 3.78, 95% CI, 2.47-5.83, respectively, in a multiple logistic regression model. rs1893219 A>G was negatively associated with ALL (AG: OR = 0.68; 95% CI = 0.50-0.94 and GG: OR = 0.60; 95% CI = 0.42-0.85), while for rs37972 C>T, TT was positively associated with ALL (OR = 1.91; 95% CI = 1.21-3.00). The combination of genotypes for (AA) and (TT) increased ALL risk (OR = 2.61; 95% CI = 1.16-5.87). In a multiplicative interaction, rs1893219 A>G was associated with children whose mothers had less than 9 years of schooling (iOR = 1.99; 95% CI = 1.11-1.55).

Conclusion: Our study has demonstrated a significant association between rs1893219 A>G (reduced risk) and rs37972 C>T variants (increased risk) and childhood ALL susceptibility. Based on this evidence, genes controlling the HPA axis activity may play a role in leukemogenesis, and further investigation is needed to substantiate our findings.