Integrated biomarker and treatment correlates of prognosis in infant non-WNT/non-SHH medulloblastoma: a multinational retrospective cohort study

Background: Infant medulloblastoma (age <3–5 years at diagnosis) is a major challenge in paediatric oncology. Clinical studies of molecularly defined non-WNT/non-SHH infant medulloblastoma (~60% of cases) have not been done, and this group is currently considered to have uniformly high risk. Understanding the potential for its biological subclassification and clinical stratification is an essential goal towards improved outcomes. This study therefore aimed to directly compare different therapeutic approaches in non-WNT/non-SHH infant medulloblastoma and assess the relationships between outcomes and clinicomolecular features. Methods: We assembled an international cohort of infants with non-WNT/non-SHH medulloblastoma identified from non-trial cohorts in nine countries. Inclusion criteria were age 0–5 years and a principal non-WNT/non-SHH medulloblastoma classification (group 3 [MB_Group3] or group 4 [MB_Group4]; confidence score >0·8) using DNA methylation array-based classification. Patient clinical and molecular data were collected from contributing institutions and centrally reviewed using standardised annotation protocols. A survival cohort was defined by the availability of complete progression-free and overall survival data, and grouped by whether principal upfront therapy included craniospinal irradiation or consisted of chemotherapy only. Chemotherapy was subclassified into high-dose, intraventricular methotrexate-based, and standard-dose regimens. We assessed the relationships between tumour molecular pathology, treatments received, and outcomes (progression-free and overall survival) using Kaplan–Meier plots, univariable log-rank tests, and Cox regression. Findings: The total collected cohort (n=375) comprised 262 males and 110 females (three patients had missing sex data), with a median age of 3·0 years (IQR 2·5–4·0). MB_Group3 tumours predominated (246 [66%] of 375 patients), among which molecular subgroups 4 (98 [40%]), 2 (75 [30%]), and 3 (46 [19%]) were most common. The remaining 129 (34%) of 375 patients had MB_Group4 tumours. 313 patients were included in the survival cohort. Upfront craniospinal irradiation was associated with significantly better 5-year progression-free survival (62% [95% CI 55–70]) than non-craniospinal irradiation approaches (including focal radiotherapy or chemotherapy-only strategies; 33% [23–44]; p<0·0001). When upfront chemotherapy-only approaches were used, high-dose chemotherapy produced better survival rates (5-year overall survival 60% [95% CI 42–87]) than standard-dose chemotherapy (27% [13–57]). Patients with MB_Group3 subgroup 4 represented a novel chemosensitive group, with 5-year progression-free survival of 64% (95% CI 44–95) when treated upfront with high-dose chemotherapy only (n=14). Patients with MB_Group3 subgroup 2 or 3 with MYC amplification (n=10) had 5-year progression-free and overall survival rates of 0% when treated with chemotherapy only. Patients with MB_Group3 subgroup 2 or 3 without MYC amplification who were treated upfront with chemotherapy only (n=10) had 5-year progression-free survival of 30% (12–77) and 5-year overall survival of 58% (34–100). Interpretation: Non-WNT/non-SHH infant medulloblastoma outcomes are associated with specific biomarkers and type of therapy received. This real-world experience identifies a favourable-risk group (MB_Group3 subgroup 4) with good prognosis and a very-high-risk group (MB_Group3 subgroup 2 or 3 with MYC amplification) with poor prognosis, providing a foundation for biomarker-driven MB_Group3 clinical trials. Funding: Cancer Research UK, Children with Cancer UK, Children's Cancer North, Star for Harris, JGW Patterson Foundation, Little Hero, and Blue Skye Thinking.