A SERPINE1 (PAI-1) Single Nucleotide Polymorphism Predicts Osteonecrosis in Children Treated for Acute Lymphoblastic Leukemia: Results of the Children’s Oncology Group (COG) Study AALL03B2.

As intensity of glucocorticoid use has increased, osteonecrosis has become an increasingly frequent complication of modern intensive chemotherapy for pediatric patients with acute lymphoblastic leukemia (ALL). Although patient age > 10 years is a well-known risk factor, other host risk factors that may predispose to this complication are not well described. We tested whether 12 polymorphisms in 11 candidate genes were associated with osteonecrosis among up to 344 patients ≥ 10 years old treated on the Children’s Cancer Group protocol CCG 1882 for newly diagnosed high-risk ALL, adjusting for gender, age, and treatment arm (augmented versus standard treatment). Genes tested (TYMS, ER, MTHFR, ABCB1, BGLAP (osteocalcin), ACP5, LRP5, ESR, SERPINE1, VDR, PTH, and PTHR) were chosen because they are candidates based on the putative mechanisms underlying osteonecrosis risk, including involvement in antifolate pharmacodynamics, drug pharmacokinetics, thrombosis risk, and possible hormonal influence on glucocorticoid effects. All children received dexamethasone during the delayed intensification phase of therapy, with doses varying by treatment arm (one dexamethasone pulse in the standard chemotherapy arm vs. two dexamethasone pulses in the augmented chemotherapy arm). Of the 11 candidate polymorphisms, a single nucleotide polymorphism (dbSNP rs6092) in the SERPINE1 gene (plasminogen activator inhibitor type 1 or PAI-1) was the only polymorphism associated with an increased risk of osteonecrosis in both univariate (p=0.0039) and multivariate (p=0.0019) analyses (adjusting for gender, age, and treatment arm), with an odds ratio of 2.51 (95% CI 1.32 to 4.74). Overall, 23.9% of the 88 children heterozygous or homozygous for the SERPINE1 T allele, versus 11.1% of the 234 children homozygous for the C allele, developed osteonecrosis. The combined CT or TT genotypes were present in a greater percentage (27%) of whites than in blacks (12%, p = 0.016), consistent with a higher risk (p = 0.003) of osteonecrosis in whites than blacks that was originally reported for CCG-1882 (J Clin Oncol18:3262–72, 2000). In a multivariate analysis, only the SERPINE1 polymorphism and sex predicted AVN, with the incidence of osteonecrosis lower in boys (odds ratio 0.49, 95% confidence interval 0.25 to 0.96, p = 0.034). No statistically significant gene-gene interactions between SERPINE1 and any other tested genotype were noted. SERPINE1 polymorphisms and PAI-1 serum levels have previously been associated with a clinical pro-thrombotic state. We conclude that polymorphisms in SERPINE1 may contribute to the risk of osteonecrosis in patients treated with glucocorticoids.