Abstract

Approximately 40% of patients treated with chemotherapy for pediatric AML will relapse after initial treatment and 8–10% will die of toxic side effects of chemotherapy. This highly variable response to standardized treatment protocols is due in part to genetic variation among patients in the metabolism of chemotherapeutic agents and in the response to the DNA damage they induce. We have analyzed polymorphisms in the coding region of the DNA alkylation repair gene O6-Methylguanine-DNA Methyltransferase (MGMT) in 356 patients with de novo pediatric AML to determine whether naturally occurring MGMT variant alleles correlate with leukemia susceptibility, treatment efficacy, and the incidence of adverse effects of chemotherapy. The polymorphism rs2308321 (Ile143Val) has been shown to modify biochemical properties of the MGMT protein, with altered substrate specificity of the rare valine allele (G genotype) compared with the isoleucine allele (A genotype) (

Margison et al,
Carcinogenesis
,
26
;
1473
,
2005
). All patients were treated on Children’s Cancer Group protocols 2941 and 2961. Outcomes were examined stratified by genotype. All patients received intensively timed induction therapy with IDA-DCTER (idarubicin, dexamethasone, Ara-C, thioguanine, etoposide and daunomycin) given on days 0 to 3 followed by DCTER given on days 10 to 13. On recovery of ANC and platelet counts, patients were randomized to consolidation therapy with a further course of IDA-DCTER/DCTER or IDA-FLAG (idarubicin, fludarabine, Ara-C and G-CSF). Patients received intensification with a single course of high dose Ara-C and L-asparaginase unless a family donor was available when allogeneic transplant was performed after busulfan and cyclophosphamide. Black children were significantly more likely to have a homozygous wild-type AA genotype than white children (p=0.002). When compared to a healthy control population, there was no correlation between MGMT polymorphisms and the risk of AML overall. However, in a case-case comparison of genetic sub-types of AML, children with a variant allele (AG or G/G genotype) were significantly more likely to have monosomy 7 than AA cases (9% vs 1%; p=0.01); in contrast, children with trisomy 8 were significantly more likely to have an AA genotype (8% vs 0%; p=0.028). The presence of at least one G allele was associated with a significant reduction in the incidence of relapse. There was no effect on treatment-related mortality (TRM). In a multivariate Cox model adjusted for race, the increased risk of relapse (RR) associated with the variant G allele remained significant (hazard ratio 1.72 95% CI 1.03–2.86; p=0.037). These results suggest that genotyping at the rs2308321 polymorphism in the MGMT gene could be useful in the risk stratification of patients with AML prior to therapy.

A/AA/G+G/G
N% ± 2 SEN% ± 2 SEp-value
5yr OS from end of course 1 280 54 ± 6 76 71 ± 11 0.031 
5yr RR from end of course 1 280 43 ± 6 76 27 ± 11 0.014 
5yr TRM from end of course 1 280 14 ± 4 76 14 ± 8 0.11 
A/AA/G+G/G
N% ± 2 SEN% ± 2 SEp-value
5yr OS from end of course 1 280 54 ± 6 76 71 ± 11 0.031 
5yr RR from end of course 1 280 43 ± 6 76 27 ± 11 0.014 
5yr TRM from end of course 1 280 14 ± 4 76 14 ± 8 0.11 

Author notes

Disclosure: No relevant conflicts of interest to declare.