Approximately 30% of medulloblastoma patients present with metastatic disease [1] and have a poorer prognosis. There is a clear worse outcome for image-defined intracranial disease dissemination (Chang stage M2) or spread to the spine (Chang Stage M3), but microscopic spread to the cerebrospinal fluid (CSF) (Chang stage M1) is independent of the presence of macroscopic metastasis [16,17]. Chang stages M1-M4 are thus considered high-risk [18]. Multicentre trials showed a significant rate of false staging of patients; described as not having metastasis on local imaging reports but revised on central review; this is reflected in a lower-than-expected survival in patients not undergoing central review [19,20]. Quality assurance thus mandates central review for all patients in this trial. M4 disease is exceptionally rare and the approach to its management must be individualised.

The recognised histological variants of medulloblastoma in the WHO classification of CNS tumours (2016 and 2021) are as described above: classic medulloblastoma, desmoplastic/nodular medulloblastoma, medulloblastoma with extensive nodularity and large cell/anaplastic medulloblastoma [2]. Current treatment strategies use histology as a tool for risk-stratification. LCA medulloblastoma, although briefly classified separately as large-cell and anaplastic medulloblastoma, have now been re-grouped as one entity in the WHO 2016 classification due to the difficulty in differentiating these rare variants, which often show mixed phenotypes [1]. Large-cell medulloblastoma is characterised by predominant monomorphic cells with large, round vesicular nuclei, single prominent nucleoli and variable amounts of eosinophilic cytoplasm [21]. Highly aggressive behaviour has been described in several reports [22,23]. Severe cytological anaplasia is also recognised to be a negative prognostic factor [24].

The extent of resection is still currently considered as a prognostic variable in medulloblastoma when overt metastatic disease is excluded by initial staging; however, its influence on PFS and OS is not clear. Apart from the CCG-921 trial, undertaken in the pre-magnetic resonance imaging (MRI) era, there are roughly an equal number of studies that identify, or fail to identify, an association between the increased extent of resection and OS. In the biggest randomised trial so far reported for non-metastatic medulloblastoma patients by Packer et al. in 2006, the 15 patients with post-operative residual disease did not have a significantly worse prognosis than the others [20]. In the St Jude medulloblastoma-96 trial the “high” risk group represented by those six patients with only residual disease (non-metastatic) reported having 100% EFS/OS [7]. The presence of residual post-operative disease was prognostic in the SIOP PNET 4 trial [25], but more recent prognostic analyses of 184 medulloblastoma cases treated with HIT (German-speaking countries cooperative group) protocols did not reveal a role of residual disease in a multivariate evaluation [26]. In an analysis of 125 consecutive patients in a single Italian institution, the eight children with only residual disease did not have a statistically different EFS and OS from the patients without residual disease [27]. A recent report from the Paediatric Oncology Group (POG) 9631 protocol, exploring the role of concomitant oral etoposide during craniospinal irradiation, once again did not find residual disease as prognostic factor [28]. Furthermore, it is probable that the prognostic benefit of a total resection is attenuated after accounting for a molecular subgroup affiliation [4].

Considering all of these data, there is a paucity of supportive evidence that intensifying therapy to the craniospinal axis improves local control in the setting of subtotal resection. Presently, the SIOP-E group recommends that a residual tumour without any other high-risk factors should be treated similarly to standard-risk disease.

The discovery of molecular disease subgroups represents the most fundamental recent advance in our biological understanding of medulloblastoma. The current international consensus recognises four subgroups—WNT, SHH, Group 3 and Group 4 [29] and further subtypes within these subgroups were recently described [3,5,30,31]. Each subgroup is defined empirically by genome-wide transcriptomic and DNA methylation patterns [5,32] and characterised by distinct clinico-pathological and molecular features. WNT and SHH are synonymous with WNT (wnt/wingless pathway) and SHH (sonic hedgehog pathway)-activating mutations, respectively [33,34]. Childhood WNT patients (<16 years at diagnosis) consistently show a favourable prognosis (>90% survival) [7,35,36,37]. In addition, significant biological heterogeneity is evident within each non-WNT subgroup, for instance, TP53 mutations associate with a poor outcome in SHH [5,38]. The loss of p53 function is thought to confer resistance to chemotherapy [39,40], and effective anti-tumoural treatments have yet to be established for this group, which represents approximately 10 patients in Europe per year. In contrast, Group 3 and Group 4 harbour few mutations but multiple DNA copy number alterations [34]. Importantly, subgrouping and TP53 status are now integral to the World Health Organization (WHO) MB classification and are considered the ‘standard-of-care’ [2]. In addition to WNT- and SHH/TP53-mutated tumours, the presence of MYC or MYCN amplification were consistently identified as independent prognostic factors in trials-based studies [14,35,40]. MYC/MYCN amplification is also significantly associated with metastasis and LCA histology [41]. Schema that incorporate these combined factors significantly outperform risk-stratification using clinical factors alone [4,35]. The prognostic significance of MYC/MYCN amplification and histology is likely to be relevant only in the context of molecular subgrouping (e.g., MYC amplification in Group 3 tumours; MYCN amplification in SHH but not Group 4 tumours); therefore, clearer risk groups may become apparent as these refined prognostic associations are validated [5,42].

MYCN amplification was considered a high-risk factor in the original SIOP-PNET5-MB protocol, based on its association with a poor prognosis in studies undertaken across the disease prior to the identification of the four consensus molecular subgroups [35,41]. Two large retrospective studies have since been undertaken which assessed the prognostic impact of MYCN amplification with reference to these subgroups [5,42]. In both studies, MYCN amplification was associated with the SHH and Group 4 subgroups and displayed different clinical outcomes in each. In SHH, MYCN amplification was associated with a poor prognosis and commonly co-occurred with other high-risk factors (LCA pathology, TP53 mutation, M+ disease). In contrast, MYCN amplification in Group 4 was not associated with a worse prognosis. These associations have since been validated in investigations of two groups of standard-risk patients (i.e., M− and R− with classic or desmoplastic pathology and no evidence of MYC amplification) from the HIT-SIOP-PNET4 clinical trial cohort and a UK research cohort [4,43].

Finally, emerging biological risk factors have the clear potential to further understand disease heterogeneity and improve the stratification of risk in medulloblastoma (e.g., novel molecular subgroups and/or whole-chromosome aberration patterns within Group 3/4 tumours [3,5,30,31,32] M+ in Group 4 tumours [4]). These require urgent evaluation and/or validation in the clinical trials setting, alongside biomarker discovery studies that focus on understanding heterogeneity within the high-risk medulloblastoma clinical group.

Familial disease/germline mutations describe a notable proportion of medulloblastomas (5–10%); predominantly Gorlin (PTCH1/SUFU mutation in SHH patients), Turcot (Adenomatous-polyposis-coli (APC) in WNT patients), Li-Fraumeni (TP53 in SHH patients) and Fanconi’s Anaemia (BRCA2/PALB2, subgroup unknown); they are associated with systemic radio- and chemosensitivity and must also be considered in therapy selection [30].

Although SHH subgroup patients with somatic TP53 mutations are treated on high-risk protocols, chemotherapy-related toxicity and secondary malignancies are of great concern in patients with germline TP53 mutations [44]. Alkylating drugs especially seem to exert a high geno-toxic stress in TP53-deficient backgrounds [45]. In a historic cohort of n = 37 patients with SHH-activated, germline TP53-mutated medulloblastoma treated with surgery, chemotherapy and radiotherapy, 3- and 5-year EFS were 20% and 16%, respectively, and no long-term survivors were detected (Milde, personal communication). No difference in OS and PFS was detected when patients were treated with chemotherapy before RT as compared to RT immediately after surgery, suggesting that chemotherapy before radiotherapy (i.e., a delay of radiotherapy) does not significantly influence the outcome. Thus, there is currently no consensus on the treatment of SHH-activated, germline TP53-mutated medulloblastoma patients, and specific clinical studies are required for this patient group.

Children and young people currently eligible for trials of high-risk medulloblastoma are summarised in Table 1.