BACKGROUND:MYCN amplification (MYCN-A) is an established adverse prognostic factor in neuroblastoma. The extent to which the prognostic impact of MYCN-A depends on other factors has not been fully characterized. PATIENTS AND METHODS:Using the International Neuroblastoma Risk Group database, we constructed Cox models of overall survival (OS) to obtain hazard ratios of the effect of MYCN-A within subgroups defined by other prognostic factors. Cox models assessed the degree to which the prognostic impact of MYCN-A was modulated by each other covariate. We used absolute hazard ratio (HR) differences to construct classification trees to identify subgroups with greatest differential prognostic effect of MYCN-A. RESULTS:In a cohort of 6223 patients with known MYCN status, the OS hazard ratio associated with MYCN-A was 6.3 (95% confidence interval 5.7-7.0, P < .001). Age at diagnosis conferred the largest HR absolute difference for MYCN-A between subgroups (HR absolute difference 16.6; HRs for MYCN-A of 19.6 for <18 months, 3.0 for ≥18 months). MYCN-A remained significantly prognostic of OS after controlling for other factors, abrogating their prognostic strength. Patients whose outcome was most impacted by MYCN status were those who were <18 months, had high mitosis karrhyohexis index (MKI) and low ferritin. CONCLUSION:The prognostic strength of MYCN-A varies depending on which patient subgroup defined by other neuroblastoma risk factors is examined, with greatest strength in patients with otherwise favorable features. MYCN-A has little effect within some subgroups, aiding clinical decision-making if MYCN status cannot be assessed. Subgroups where MYCN-A has large effect may be prioritized for agents targeting Myc family proteins.