Developing an advanced risk stratification model for pediatric intracranial ependymoma based on the prospective trial E-HIT2000 and subsequent registries

Abstract

BackgroundCurrent treatment strategies for pediatric intracranial ependymoma do not consider molecular heterogeneity. Here, we evaluated molecular group-specific determinants of outcome and developed an improved risk stratification model. Methods Patients aged 0-21 years with localized intracranial ependymoma were enrolled into the prospective clinical trial E-HIT2000. Treatment included maximum safe surgery, local radiotherapy, and chemotherapy, stratified according to age, histology and, following a major amendment, residual tumor. Clinical data were analyzed in a pooled molecularly annotated cohort with data from patients treated analogously within subsequent registries. Results For 291 trial patients, the 5-year progression-free survival (PFS) and overall survival (OS) were 62 ± 3% and 81 ± 2%, respectively. For the molecularly annotated pooled cohort (n = 228), 5-year PFS/OS were: posterior-fossa group A ependymoma (EPN-PFA) (n = 146): 45 ± 4%/77 ± 4%; posterior-fossa group B ependymoma (EPN-PFB) (n = 19): 90 ± 7%/100%; supratentorial ependymoma, ZFTA fusion-positive (EPN-ZFTA) (n = 59): 64 ± 7%/86 ± 5%; supratentorial ependymoma, YAP1 fusion-positive (EPN-YAP1) (n = 4): 50 ± 25%/100%. Patients with EPN-PFA without molecular risk factors (1q gain, and/or subtype EPN-PFA1c/d/e, 2a), with complete resection, and postoperative radiotherapy showed favorable outcomes (5-year PFS/OS 75 ± 10%/92 ± 7%). For patients with EPN-PFA with molecular risk factors, prognosis was poor irrespective of residual tumor status (5-year PFS/OS: 33 ± 6%/64 ± 6%). Among EPN-ZFTA, 11/59 tumors were classified as EPN-ZFTA with alternative fusions, associated with inferior PFS (5-year PFS/OS: 36 ± 15%/91 ± 9%). For EPN-ZFTA-RELA, homozygous deletions of CDKN2A were associated with unfavorable outcomes (4-year PFS/OS: 19 ± 16%/57 ± 18% vs. 79 ± 7%/97 ± 3%, P =. 0001). Finally, we developed a novel stratification model that discriminates standard and intermediate risk patients from those at high risk (P