OBJECTIVES: To investigate association of host genomic variation and risk of infections during treatment for childhood acute lymphoblastic leukaemia (ALL).

METHODS: We explored association of 34,000 single-nucleotide polymorphisms (SNPs) related primarily to pharmacogenomics and immune function to risk of infections among 69 ALL patients on induction therapy.

RESULTS: Forty-eight (70%) patients experienced infectious events including 23 with positive blood cultures. Infectious events and positive blood cultures were associated significantly with 24 and 21 SNPs, respectively (P < 0.01). Classification and regression tree analysis demonstrated rs11033797 (OR51F1), rs2835265 (CBR1), rs28627172 (POLDIP3) and rs1129844 (CCL11) to be predictive of outcome. Among 61 patients for whom read-outs were available for all four SNPs, 40 of 41 patients with the worst SNP profile experienced at least one infectious event compared with five of the remaining 20 patients (Hazard ratio 9.0, 95% CI 3.4-23.5, which was unchanged after adjustments for neutrophil counts). Pathway analysis identified variations in 'G-protein-coupled receptor (GPCR) downstream signalling', 'Bile acid and bile salt metabolism' and 'Class I MHC-mediated antigen processing and presentation' to be highly predictive of infections.

CONCLUSIONS: Our data indicate that host genomic profiling may predict the risk of infections during induction therapy. This may facilitate development of individualised supportive care.