ALL and osteonecrosis

In this issue of Blood, Karol et al establish a link between glutamate receptor polymorphisms and osteonecrosis, using a large discovery cohort and 2 validation cohorts. One validation cohort included patients under treatment of acute lymphoblastic leukemia (ALL) and the second is patients receiving glucocorticoid treatment of various other indications.¹  This battered old clinician does not pretend to follow all the mathematics.

Osteonecrosis is the bane of contemporary therapy for ALL in adolescents and young adults, leading to lifelong pain and disability for some.²  Patients may undergo surgery, such as hip replacement, and obtain symptomatic relief, but artificial hips wear out and must be replaced every 10 to 20 years. These patients may be cured from their leukemia, but we have exchanged one disease for another.

Osteonecrosis appears more common adolescent and young adults, more common in young women than young men, and less common in African Americans. The incidence of symptomatic osteonecrosis varies from 10% to 30% and depends somewhat on the thoroughness of the treating physicians.¹  Pain and loss of range of motion appear prior to any x-ray changes and diagnosis is best made with magnetic resonance imaging (MRI). The course is quite variable, complicating assessment of candidate interventions.

The Children’s Oncology Group (COG) decreased dexamethasone in the Delayed Intensification phase from 21 to 14 days (7 days on, 7 days off, and 7 days on), with no erosion of outcome and almost halved the incidence of osteonecrosis.³  The UKALL 2011 study currently tests a similar intermittent dexamethasone schedule in induction. Results are eagerly awaited.

Hopes that MRI screening would lead to early detection and useful intervention have been confounded by a high incidence of transient marrow edema. Enthusiasm for statins as a preventive agent has gone nowhere. Early surgical intervention remains controversial. Prostaglandins or bisphosphonates to slow progression and ameliorate pain remain anecdotes. Only the seriousness and the urgency of the problem stand unchallenged.⁴ 

Karol et al find that a polymorphism at single nucleotide polymorphism rs 10989692 near the glutamate receptor GRIN3A locus was associated with a doubling of the incidence of osteonecrosis.¹  The allelic frequency is 0.106, so ∼20% of patients have ≥1 affected allele. Using the discovery cohort of COG white patients, back of the envelope calculations indicate an osteonecrosis incidence of 21.5% for affected patients with G/A or A/A and 10.7% for G/G patients (see Table 1 in Karol et al). The polymorphism accounts for ∼17% of cases of osteonecrosis, ∼1 case in 6. White patients with the G/G genotype have 109 cases in place of the expected 132 cases, a reduction in incidence from 12.9% to 10.7%, and remain at substantial risk for symptomatic osteonecrosis.

Karol et al are impressive for their sample size and scientific rigor. They present compelling arguments for biological plausibility. However, no prior genome-wide association study investigation has linked glutamate receptor genetic variations and osteonecrosis. A variety of other plausible polymorphisms have been inconsistently implicated with similar hazard ratios, involving PAI-1, glucocorticoid metabolism, antifolate pharmacodynamics, fibrinolysis, and lipid and albumin homeostasis.^5-8  Few candidate polymorphisms appear in >1 report. As with all retrospective studies, prospective confirmation is needed.⁹