Polymorphism in the PAI-1 (SERPINE1) gene and the risk of osteonecrosis in children with acute lymphoblastic leukemia

To the editor:

The success of modern therapy for childhood acute lymphoblastic leukemia (ALL) has led to an increased focus on avoidance of treatment-related morbidity in long-term survivors. It is clear that the prognosis for older patients in particular has been improved by protocols based on effective asparagine depletion and intensive exposure to corticosteroids, particularly dexamethasone.^1,2  This improvement has come at a cost of increased rates of osteonecrosis (ON), which affects up to 15% of older children and young adults treated in recent trials.^3,4  Some collaborative groups are sufficiently concerned about bone toxicity that they are opting to use alternative corticosteroid regimens in selected patients, at the potential expense of antileukemia efficacy.⁵ 

A more appealing approach might be to highlight patients at genetic risk of development of ON. Polymorphism in the plasminogen activator inhibitor-1 (PAI-1) gene (rs6092; AA/GA genotype) was reported by the Children's Oncology Group (COG) to be associated with an approximately 2.8-fold increased risk of ON in patients treated as part of the Children's Cancer Group CCG1882 study.⁶  This effect was independent of traditional risk factors for ON development. It was suggested that increased PAI-1 levels associated with this polymorphism may lead to reduced fibrinolysis, leading to increased intraosseous venous pressure, compromise of blood flow, and bone death.

We assessed the prevalence of the rs6092 polymorphism in patients with and without ON treated as part of the Medical Research Council UKALL 2003 trial. Treatment regimen for National Cancer Institute high-risk cases in ALL 2003 is identical to CCG 1882, with the exception that dexamethasone (6 mg/m²) replaces prednisolone, and all patients receive PEG-asparaginase. We sought to develop a simple assay to identify patients at increased risk of ON, who may benefit from avoidance of intensive dexamethasone treatment. We identified 43 cases of symptomatic radiologically confirmed ON and 67 age-, sex-, and treatment-matched controls (Table 1) and tested constitutional DNA using a qPCR-based assay for allelic discrimination. Results of gene sequencing of a representative 7 samples were concordant with the results of the assay, confirming its reliability. The combined incidence of AA and GA genotype was similar in ON cases (20.9%) and controls (23.9%) (Table 1). There was no significant association of the polymorphism with ON with an odds ratio of 0.81 (95% CI = 0.33-2.00) for ON development.

It is possible that other genetic modifiers in our cohort could obscure any putative effect of the PAI-1 polymorphism, although it is notable that a recently reported genomewide association study of ON-related single nucleotide polymorphisms also failed to show an association between rs6092 and ON.⁷  Monogenic prediction of disease is not straightforward, particularly in the context of a disorder such as ON, which has a multifactorial pathogenesis that remains poorly understood. Further work is necessary to characterize the genetic basis of therapy-related morbidity in ALL. Better tools of prediction would allow prospective identification of patients in whom timely treatment modification might reduce the risk of serious pathology.