Osteonecrosis in childhood ALL

Kawedia et al report a 71% incidence of osteonecrosis (ON) in a cohort of children and young adult acute lymphoblastic leukemia (ALL) patients who underwent routine magnetic resonance imaging (MRI) screening at various time points on study.¹  A majority of cases were asymptomatic. Older age, increased treatment intensity, low serum albumen, elevated lipid levels, and decreased dexamethasone clearance were risk factors associated with an increased incidence of osteonecrosis.

Kawedia et al present the first large series of pediatric patients (N = 364) with ALL who were screened via MRI for ON at specified time points in the chemotherapy treatment program. The authors identified older age and increased intensity of treatment as important prognostic factors for ON. Both of these risk factors for ON have been demonstrated to be significant in other trials. In addition, in this study, lower serum albumen levels, elevated serum cholesterol, and decreased clearance of dexamethasone were prognostic factors for symptomatic ON. However, the study raises many more questions than it provides answers.

A key question is whether or not routine screening of ALL patients for ON should be performed. In this study, the overall incidence of ON was 71.8% while the incidence of symptomatic ON was 17.6%. One hundred forty-one patients were discovered to have asymptomatic ON on the initial screen (38.7%). One hundred thirty of these patients had a second MRI performed. Thirty-four of these patients (26%) eventually developed symptomatic ON compared with 27 of 194 patients whose initial-screening MRI did not show ON (P = .008). An additional 62 patients who had no ON on initial screening subsequently developed asymptomatic ON as their maximal grade of ON. Thus, the predictive value of an initial finding of asymptomatic ON for the development of symptomatic ON is relatively low.

The incidence of symptomatic ON in patients less than and greater than 10 years of age in this study was 28/272 (10%) and 41/92 (44.6%), respectively. These incidences of symptomatic ON are much higher than the rates of symptomatic ON observed in other studies in which screening was not performed. In the Children's Oncology Group (COG) high-risk ALL study 1961, the incidence of symptomatic ON in patients < 10 years, 10-15 years, and 16+ years was 0.96%, 9.8% and 19.9%, respectively.²  In this study, using discontinuous dexamethasone during delayed intensification phases reduced the incidence of ON by approximately 50%.³  In the current COG 0232 study, the rate of AVN was 15.2% for patients > 10 years of age at 2 years. The incidence of symptomatic ON in COG ALL 0232 patients < 10 years of age was 2.6%.⁴ 

There is a problem comparing the incidence of ON in this study to that seen in other studies where screening was not performed. In this study by Kawedia and colleagues, 42 of the 69 patients with “symptomatic” ON had a maximum grade 2 ON, which is defined as pain without interference with function. Only 27 cases were identified in which the ON led to reduced function. It is unclear whether grade 2 ON would have been identified if not for screening. The Kawedia et al trial uses high cumulative doses of dexamethasone and vincristine. In the presence of asymptomatic ON, pain that might otherwise have been attributable to vincristine neuropathy or steroid myopathy might well have been ascribed to ON. In addition, parents and physicians may have been sensitized by a diagnosis of asymptomatic ON to note any pain symptoms in the affected region. The incidence of grade 3 and 4 ON seems comparable with that reported for other studies.

In cases where asymptomatic ON is diagnosed by routine MRI, there is no evidence extant to suggest that any intervention such as reducing further steroid exposure or starting therapy such as a statin might decrease the incidence of progression to symptomatic ON. In the study by Kawedia et al, management of patients with symptomatic ON was not standardized, but left to the discretion of the treating physician. One could conceive of a possible trial design in which patients with high serum lipids and low serum albumen and/or decreased dexamethasone clearance might be eligible for an intervention such as statin therapy in an effort to reduce the risk of grade 3 and 4 ON.

In my opinion, this study does not provide evidence that screening asymptomatic patients for ON by MRI provides clinical benefit, and screening should not be performed outside the context of a well-designed clinical trial. If screening is performed, it will be critical to control how therapy is changed if asymptomatic ON is discovered, because this study suggests that a large percentage of patients will have such findings. Screening should not be performed on a routine basis in patients < 10 years of age. For older patients, one could consider screening only those patients at increased risk for developing symptomatic ON but, however, we lack clearly defined risk factors for symptomatic ON and we have no evidence that any intervention would be effective in patients identified by screening.