Skip to main content
. 2014 May 4;128(1):137–149. doi: 10.1007/s00401-014-1276-0

Table 4.

Multivariable Cox regression model for event-free survival including molecular subgrouping, age at diagnosis, M stage, residual disease, histopathological subtype and MYC status

Variable Available cases HR 95 % CI p*
Age at diagnosis N/S***
 <4 vs. >4 33 vs. 133
M_Stage 0.045
 M1–M4 vs. M0 66 vs. 100 2.064 0.998–4.269
Residual tumor N/S***
 >1.5 cm2 vs. <1.5 cm2 21 vs. 145
WHO classification N/S***
 Desmoplastic/nodular vs. classic 33 vs. 119
 MBEN vs. classic 5 vs. 119
 Anaplastic vs. classic 8 vs. 119
 Large cell vs. classic 1 vs. 119
MYC_status N/S***
 Amplified vs. balanced 7 vs. 159
450k subgrouping 0.032
 Group_3 vs. Group_4 45 vs. 70 2.141 1.042–4.400
 SHH vs. Group_4 38 vs. 70 1.329 0.497–3.556
 SHH vs. Group_3 38 vs. 45 0.621 0.234–1.644
 WNT vs. Group_4 13 vs. 70 NE

Estimated hazard ratio (HR) with 95 % confidence interval (CI) and p value of the likelihood ratio test for omnibus test

NE not estimable (because there are no events in this group)

* p value of the likelihood ratio test for omnibus test. For pairwise comparisons, confidence intervals instead of p values are given (p value of Wald test ≤0.05 if and only if confidence interval does not contain 1)

*** N/S not selected in the final multivariable model (inclusion: p value Score test ≤0.05, exclusion: p value likelihood ratio test >0.1)