Background: Genome-wide association studies (GWAS) have identified validated neuroblastoma susceptibility loci associated with different tumor subtypes. Approximately 20% of neuroblastomas are MYCN-amplified, but it is unknown whether patients with MYCN-amplified tumors have different genetic predispositions than other patients. To investigate this, we performed a multi-ethnic GWAS on 763 MYCN-amplified cases and 3440 MYCN-nonamplified cases as controls.
Methods: Neuroblastoma patients with clinical and genomic information were obtained from the Children’s Oncology Group. Genotypes from three cohorts typed on different Illumina platforms were subjected to quality control and imputation, and analyzed separately using SNPTEST. Meta-analysis was performed using a fixed-effects model in METAL.
Results: By meta-analysis, we identified two loci significantly associated with MYCN-amplification. The first was at the previously identified high-risk BARD1 locus indexed by rs2070096 (Pmeta=3.51×10-13). The second was a novel locus at 3p21 indexed by rs112645395 (Pmeta=8.83×10-13). In GTEx, three variants in tight linkage (r2>0.75) with rs112645395 are cis-eQTLs in tibial nerve for KIF15 (P≤2.6×10-5), a critical protein in mitotic pole separation, with the risk allele associated with decreased gene expression. We then used a subset of 1662 patients with recorded risk status to test if previously identified high-risk susceptibility loci differed by MYCN status. Variants in BARD1 and CASC15-S were enriched in MYCN-amplified neuroblastoma but not MYCN-nonamplified high-risk neuroblastoma (BARD1: Prs2070096/MYCN-amp=1.54×10-5 vs Prs2070096/MYCN-nonamp=0.34, CASC15-S: Prs1543310/MYCN-amp=7.90×10-5 vs Prs1543310/MYCN-nonamp=0.10). Conversely, LMO1 was significant only in patients with MYCN-nonamplified high-risk tumors (Prs110419/MYCN-amp=0.38 vs Prs110419/MYCN-nonamp=3.44×10-5).
Conclusions: We demonstrate MYCN-amplified and MYCN-nonamplified high-risk neuroblastoma arise under the influence of distinct germline susceptibility alleles. These results also suggest the etiology of MYCN-amplified neuroblastoma may result from attenuated DNA repair and/or impaired mitotic chromosome segregation. Further functional analysis of BARD1, CASC15-S, and KIF15 may provide insight into the pathogenesis of MYCN-amplification and identify potential therapeutic targets.