Table 1.

Univariate and multivariate Cox regression models for patients with non–high-risk neuroblastoma based on EFS and OS considering clinical prognostic markers and the top five genomic classifiers

MarkerAvailable casesHR (95% CI)P
A: Multivariate Cox regression model for non–high-risk patients considering established prognostic markers and the top five SVM classifiers based on EFS
 SVM_th44n.s.
 SVM_th26n.s.
 SVM_th24n.s.
 SVM_th22n.s.
SVM_th103635.11 (3.04–8.59)<0.001
 Age (<18 vs. ≥18 mo)n.s.
 Stage0.001
  I (ref)122
  II or III1543.62 (1.75–7.50)
  IV341.34 (0.46–3.93)
  IVS533.87 (1.69–8.87)
 1p (no aberration vs. Imb/del)3630.4 (0.18–0.90)0.014
 Histology (F vs. UF)n.s.
B: Univariate Cox regression models for non–high-risk patients considering established prognostic markers and the top five SVM classifiers based on OS
 SVM_th4441318.16 (6.97–47.35)<0.001
 SVM_th2641319.2 (6.42–57.46)<0.001
 SVM_th2441318.03 (6.02–53.97)<0.001
 SVM_th2241316.81 (5.62–50.32)<0.001
 SVM_th1041329.24 (9.77–87.54)<0.001
 Age (≤18 vs. ≥18 mo)4138.55 (3.49–20.95)<0.001
 StageNot calculable
  I (ref)126
  II or III176
  IV49
  IVS62
 1p (no aberration vs. Imb/del)363n.s.
 Histology (F vs. UF)3066.31 (1.76–22.61)0.005
C: Multivariate Cox regression model for non–high-risk patients based on OS considering the clinical NB2004 risk estimation system and the SVM_th10 classifier
 NB2004 (LR vs. IR)382n.s.
 SVM_th1038225.54 (8.40–77.66)<0.001

NOTE: A, the Cox regression model based on EFS; B, the model based on OS (univariate models only). The lower number of cases in the model for EFS results from a reduction of those patients for whom all variables were available (n = 363). For each factor, the reference level to which the marker is compared is indicated first and underscored (e.g., “F vs. UF” for histology, “<18 vs. ≥18 mo” for age and “no aberration vs. imb/del of 1p”). Imb/del of 1p was defined according to the criteria of the European Neuroblastoma Quality Assessment Group (18). C, highlights the multivariate Cox regression model based on OS using the variables risk stratification according to the German neuroblastoma trial NB2004 (low risk vs. intermediate risk) and the potentially best-performing genomic predictor SVM_th10.