Background

Glucocorticoids are an essential component of current treatment regimens for childhood acute lymphoblastic leukemia (ALL), and are a component of treatment for many autoimmune and inflammatory conditions. A common therapy-related and dose-limiting toxicity in ALL is glucocorticoid-induced osteonecrosis (ON). The mechanism and underlying genetic basis of ON risk are likely multifactorial.

Methods

We performed the largest genome-wide association study (GWAS) of ON to date in 2285 children with ALL who were treated on the Children's Oncology Group (COG) AALL0232 frontline protocol for high risk B-lineage ALL. We also studied two validation cohorts: 361 children with ALL treated on the frontline St. Jude Total XV protocol for ALL, and 309 children and adults treated with corticosteroids for oncologic and non-oncologic indications in the Vanderbilt University Medical Center BioVU DNA repository. Single nucleotide polymorphism (SNP) genotyping was performed using Illumina and Affymetrix Human Gene Chip Arrays. Additional SNPs were imputed using 1000 Genomes as the reference genome with MaCH-Admix and IMPUTE2. Testing for association was performed by Cox proportional hazard and logistic regression, with meta-analysis across cohorts using Stouffer's Z-score method. Gene based and pathway analyses were performed using SNP-set Kernel Association Test (SKAT) and Ingenuity Pathway Analysis (IPA), respectively.

Results

In the discovery GWAS for AALL0232, adjusting for gender, age, ancestry and treatment, the SNP with the strongest association was at 9q31.1 (rs28584318, odds ratio [OR] = 2.00, P = 4.86 x 10-7), near the GRIN3A locus (glutamate [NMDA] receptor subunit 3A). The same SNP was validated in both the St. Jude and Vanderbilt cohorts after adjusting for covariates (OR = 1.81 and 2.26, P = 0.0734 and 0.0073 respectively). Overall, patients with the variant genotype at rs28584318 in 9q31.1 had a higher incidence of ON compared with those patients who did not: AALL0232, St. Jude Total XV and Vanderbilt BioVU (15.8% vs 9.7%, 21.8% vs. 20.4%, 35.6% vs 24.0%). In the meta-analysis with both discovery and validation cohorts combined, the SNP at 9q31.1 (P = 5.34 x 10-8), and an additional variant in GRIK1 (glutamate receptor, ionotropic, kainate, rs2154490 P = 7.24 x 10-7) were most associated with risk of ON. Pathway analysis of the genes implicated in the discovery and validation cohorts also identified the glutamate pathway as the top overlapping canonical pathway present in all three cohorts (COG AALL0232 P = 0.045; St. Jude Total XV P = 0.011; Vanderbilt P = 0.037).

Conclusions

These findings demonstrate an association between loci in the glutamate pathway and glucocorticoid-induced ON. The glutamate pathway could be linked to ON via several mechanisms, including that glutamate receptors are involved in the mechanical signaling of bone and regulation of osteoblasts and osteoclastogenesis, and are affected by glucocorticoids. That this pathway was also implicated in a non-ALL cohort suggests glutamate pathway variation may modulate risk for ON across a wide array of patients treated with glucocorticoids.

Disclosures

Hunger:Sigma Tau Pharmaceuticals: Honoraria; Jazz Pharmaceuticals: Honoraria. Evans:St. Jude: In accordance with institutional policy (St. Jude), I and/or my spouse have in the past received a portion of the income St. Jude receives from licensing patent rights related to TPMT polymorphisms as clinical diagnostics. Patents & Royalties.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution