1. Introduction

25 years after Eibl had detected the first TP53 mutations in tumor probes from medulloblastomas ^[1], in 2016 the World Health Organization (WHO) introduced a revolutionary paradigm shift in the classification of these and other brain tumors. Molecular profiling of the transcriptome as a new diagnostic system superseded histology (Table 1) ^[2][3]. A recent update in 2021 also included further analysis of the methylome for epigenetic markers. Over a century after the neurosurgeon Cushing had pioneered modern brain tumor surgery ^[4] and also developed a classification of brain tumors (Table 1) ^[5], the histologic diagnosis was based on hematoxylin-eosin (HE) stains from formalin-fixed, paraffin embedded (FFPE) tissue sections and microscopic evaluation by a pathologist, typically detecting a “blue cell” tumor reflecting the high nuclear ratio to cytoplasm, as well as neuroblastic Homer-Wright rosettes. Immunohistochemistry further improved the diagnostic spectrum by using both monoclonal antibodies and polyclonal sera to detect or exclude tumor related protein markers.

The rationale for a molecular classification was a better correlation with biological behavior and to implement new therapies targeting actionable mutations. Although all medulloblastomas are diagnosed as highly malignant grade IV tumors, the four major molecular groups better reflect their development and clinical outcome. Mutations in the TP53 gene indicate a poor outcome when present in one of these four groups, but not in another, where they can also be observed less frequently. This molecular classification system will be used to guide the patients for an improved personalized treatment in precision oncology. Some patients should be identified, who need a more aggressive radio- and chemotherapy after surgical removal of the tumor ^[6]. In contrast, identification of patients, who may not need the most aggressive treatment, may benefit from the prevention of dangerous and long-term side-effects affecting the developing central nervous system (CNS). Patients may also benefit from combining the new molecular profiling from solid tumor samples with another, also recently emerging milestone in medicine: liquid biopsy, usually from blood or cerebrospinal fluid (CSF), but also urine can serve as a low-risk tool to monitor tumor development, as well as treatment response significantly earlier than standard medical imaging or CSF cytology.

Table 1. Milestones in medulloblastoma diagnostics, research and treatment.

2. Diagnosis - A Century of Debates: Does Medulloblastoma per se Exist?

Unspecified neurological symptoms, including morning headaches, vomiting and ataxia are related to the rapidly growing tumor in the cerebellum or brain stem mainly in children, which often leads to a blockage of the fourth ventricle, augmenting intracranial pressure. Computed tomography (CT) or magnetic resonance imaging (MRI) can reveal a suspicious mass in the posterior fossa region which needs to be finally confirmed and graded by a pathologist, or a neuropathologist to guide the clinicians to treatment options. Medulloblastoma can regularly form metastasis into the spinal cord, which generally is explained as “drop metastasis” into the ventricle system, followed by transport via the CSF down to the cauda equina, base of the spinal cord. However, a recent discovery in a parabiotic xenograft model showed an unexpected hematogenous spread of medulloblastoma to the leptomeninges ^[16]. This was comparable to lymphocyte homing and also involving a chemokine and its receptor. Although metastasis formation of medulloblastoma outside the brain is extremely rare, the cells (”seed”) may frequently enter the bloodstream, but don’t seem to be competent to survive or grow out except in a CNS microenvironment (“soil”). This finding of hematogenous spread supports the application of searching for CTCs in the blood of medulloblastoma patients (see Liquid Biopsy paragraph). Medical imaging and CSF cytology allow the detection of common metastasis to the spinal cord independent of the pathway of metastasis (but also without the sensitivity and specificity of liquid biopsy). A pathologist or neuropathologist needs to confirm the diagnosis of the primary tumor in the brain by microscopic analysis (histology). Ideally, the completely new WHO molecular diagnostics (2016, with an update in 2021) is already applied and integrated to the standard histology.

Before the 2021 update of the WHO classification four major morphological types of medulloblastoma were distinguished by histology: 1) classic, 2) desmoplastic/nodular, 3) medulloblastoma with extensive nodularity (MBEN) and 4) large cell/anaplastic. They are now combined to just one section: “Medulloblastoma, histologically defined”. In contrast, the new and more reliable, molecular classification identifies four major groups as shown in table 2 (Table 2).

3. The New WHO Diagnostic Classification: Activated Oncogenic Signaling Pathways

Two activated signaling pathways have been identified for the first two groups: wingless/Integration-1 (WNT)-activated and sonic hedgehog (SHH)-activated. WNT is a portmanteau for the Drosophila gene “wingless” (Wg), detected in mutants lacking wings, and the homologous mouse gene, integration 1 (Int-1), which was found earlier to cause tumors by insertional mutagenesis with a retrovirus; SHH refers to the hedgehog gene (hh) found in Drosophila mutants with spikes, reminiscent of a hedgehog (SHH is a vertebrate homolog and named after a character in a video game, Sonic the Hedgehog). Groups 3 and 4 can also be summarized as non-WNT/non-SHH. Recently, in 2022, the cell of origin for both was identified by multi-omics in the rhombic lip nodulus for cerebellar development in humans ^[17]. Mutually exclusive mutations in group 4 medulloblastomas can be attributed to affect the core binding factor (CBFA) and include CBFA2T2, CBFA2T3, PRDM6, UTX and OTX2 genes ^[18]. The expression profiles of group 4 medulloblastoma reflect those of progenitor cells of the subventricular rhombic lip, a specific part of the developing human cerebellum. In contrast to normal cells, which are able to progress from progenitor cells to more differentiated lineages, the tumor cell appears to be stuck at an earlier, embryonal stage and produces more, or too many, progenitor-like tumor cells. Although these four groups reproducibly allow prognostic evaluations, with group 1 showing only a low tendency to metastasize and the highest survival rate, it became clear that there is still heterogeneity within each of the major groups. Therefore, numerous subgroups were introduced since the WHO classification 2016 (not shown) – and may continue to increase in coming years. Interestingly, 20-30% of the medulloblastomas in the SHH group show TP53 mutations, which confer a poor prognosis. TP53 mutations in the SHH group represent the most important risk factor. In contrast, TP53 mutations can also occur in the WNT group (16%), but then they are not associated with increased risks of poor outcome and treatment failure ^[19].

Table 2. Molecular classification of medulloblastoma ^[2][3].

SHH – sonic hedgehog; WHO – world health organisation; WNT – wingless/Integration-1

Despite the new and independent molecular definition of medulloblastomas, overlapping associations with histology exist: desmoplastic/nodular medulloblastomas and MBEN belong to the SHH-group. Almost all WNT medulloblastomas show classic histology, whereas most large cell/anaplastic medulloblastoma can be found in a specific SHH-subgroup or group 3/group 4-subgroup ^[2]. The inclusion of distinguishable, but biologically different tumors in just one term as medulloblastoma led to the idea that medulloblastoma per se doesn’t exist: although these tumors share a common microscopic appearance, they differ in decisive aspects of biological behavior, clinical outcome and molecular pathways, and therefore need different treatment. Different cells of origin for each group of tumor can explain these discrepancies. Until recently, and due to their identical histological appearance, medulloblastomas were also counted as a member of primitive neuroectodermal tumors (PNETs), a concept which also included neuroblastoma and retinoblastoma. In addition, the identification of SMARCB1-INI1 mutations allowed diagnosing some atypical teratoid/rhabdoid tumors (AT/RTs) in the cerebellum, which were formerly considered to be medulloblastomas ^[20]. Currently, it is accepted that not only very different tumor entities were formerly summarized as medulloblastoma. Medulloblastomas also arise from different cells of origin and differ from the supratentorial PNETs as well as from AT/RTs.

4. Cell of Origin

Different lines of evidence – as well as a controversial debate of over a century - lead to current understanding that medulloblastoma include independent groups of tumors, sharing basic morphology but not the same cell of origin. Even in 1910, before Bailey and Cushing introduced the term medulloblastoma, the pathologist J. Homer Wright separated these tumors composed of primitive neuroepithelial cells from other CNS tumors. In his view, these neurocytomas and neuroblastomas were one entity, which he summarized from autopsies and case reports from the literature, but also neurosurgical biopsies, including one performed by the young neurosurgeon Cushing ^[7]. Later concepts on the cell of origin remained theoretical or controversial and included classification as sarcomas, neuroblastomas, spongioblastomas (or undifferentiated astrocytic, oligodendral or ependymal gliomas), or primitive tumors with multidirectional differentiation potential ^[21]. German pathologists, like Ribbert, already postulated a different cell of origin for each of these groups of brain tumors, which triggered Bailey for his major brain tumor classification with Cushing. Similar thoughts led to the current nomenclature of tumors, such as astrocytomas as deriving from astrocytes. On the other hand, one should not imply that any differentiated cell can transform into a tumor cell. As a result of asymmetric cell division, a less differentiated stem cell initiates tumor formation and produces the rare cancer stem cells, but also the more differentiated daughter cells, which further develop into the main tumor mass ^[14]. With the concept of the cell of origin, and to avoid a major confusion with the term spongioblastoma, Cushing and Bailey created the term medulloblast to postulate a hypothetical, embryonic neuroepithelial cell, and then introduced medulloblastoma as an own entity different from other CNS tumors. It remained under debate how unique and restricted to the cerebellum such a hypothesized cell needs to be, when similar looking tumors arose from outside the cerebellum, like neuroblastoma or retinoblastoma. Rubinstein assumed different unique primitive neuroectodermal stem cells in different regions of the CNS, like retinoblast, glioblast, neuroblast, pineoblast, as well as the medulloblast as the primitive neuroepithelial stem cell in in the cerebellum with a bipotential, glial or neuronal differentiation potential. Tumors arising from those hypothesized, regional stem cells should lead to retinoblastoma, glioblastoma, etc.

Within a century of controversial debates some major neuropathologists remained skeptical on how important the search for the cell of origin really was. Rorke challenged the theoretical debates in favour of focusing more on the practical approach to develop better treatments, which can be measured in clinical studies, but doesn’t need any hypothetical cell of origin. In fact, even the recent identification of such cells of origin has to prove its clinical value. It was also questioned, whether the cell of origin is important for understanding tumor development. Animal models using either the avian DNA virus SV40, or transgenic models using the SV40 large T-antigen (SV40 LT) gene helped to understand both entities better, primitive neuroectodermal tumors as well as medulloblastomas. Most of the animal models lead to similar medulloblastoma-like tumors in the brain. One exceptional model included a retrovirus-mediated gene transfer of the SV40 LT into transgenic neural transplants in rat brains. After long latency periods of several months more than half of the animals developed primitive neuroectodermal tumors, morphologically indistinguishable from human medulloblastomas, including a bipotential differentiation potential with neuronal and glial markers, formation of neuroblastic Homer-Wright rosettes and a striking migratory potential. The high resemblance to a human tumor is intriguing, since many animal tumor models look different to their human counterparts. Cell lines derived from these SV40LT expressing rat medulloblastoma-like tumors showed neuron-like processes and developed similar tumors after retransplantation. From immune precipitation studies studies it was known that SV40 LT was able to bind to p53 and form complexes ^[22]. The suggested mechanism of action was therefore the binding to p53 with inactivation of a tumor suppressor function, which leads to an oncogenic stimulus. Similar models of gene transfer into neural transplants, e.g. Ras, Myc and RasMyc as a highly oncogenic, cooperating combination of two oncogenes usually lead to very different tumors in much shorter time, often in just one or two weeks ^[23][24][25]. The beauty of the SV40LT system is the long latency period for tumor development and the suggested additional necessary hit to form these tumors.

5. First TP53 Mutations

SV40 LT was known to bind to TP53, forming complexes, suggesting its functional inactivation as a tumor suppressor. As suggested from his medulloblastoma-like rat tumor model, Eibl then tested a potential inactivation of TP53 by point mutations. With DNA extracts from frozen tumor samples, SSCP-PCR and direct sequencing of exons, Eibl found the first TP53 mutations in primary medulloblastoma tissue ^[1], whereas others were unable to detect such mutations in primary tumor tissue, but found one in a cultured cell line, which was assumed to have developed it in culture ^[26]. TP53 was also detected at that time in other brain tumors of different grades ^[27]. Eibl detected high frequency even in low-grade astrocytomas, which was fully reproduced with his colleague (von Deimling, Eibl et al., 1992) ^[28]. Eibl found no mutations in pilocytic astrocytomas (WHO I) and ependymomas ^[27]. The Li-Fraumeni syndrome is caused by TP53 mutations in the germline. Those families also develop medulloblastomas despite other tumors. This implies a major function of TP53 in human medulloblastoma development.

6. Summary

A century after medulloblastoma has entered the stage as sharing a common neuroepithelial morphology, but distinct from other CNS tumors, genetic and epigenetic analysis revealed that medulloblastoma per se doesn’t exist. Distinct tumor entities hid under the umbrella of a hypothetical and transformed medulloblast, postulated a century ago. Elucidation of mutually exclusive, activated oncogenic signaling pathways also explains differences in biological behavior and clinical outcome. The recent identification of the cell of origin for medulloblastoma group 3 and 4 supports the importance of scientific debates over a century. The rat tumor model for PNETs three decades ago triggered the first detection of TP53 mutations in human medulloblastomas. Interestingly, TP53 mutations in SHH medulloblastoma, but not in WNT are associated with a poor outcome. After 25 years these findings lead to the new, genetically defined WHO classification of brain tumors, but also influenced fundamental research on neuronal stem cells ^[29]. With four distinct groups and several subgroups, the new diagnostic system has clinical relevance and will further be developed for actionable target mutations. The identification of patients with high risks for a poor outcome supports clinical decision for an aggressive radio- and chemotherapy, whereas the average (or not high) risk patient may be prevented from major side-effects with a less aggressive, or a later start of a potentially harmful treatment. Liquid biopsy is entering clinical routine and offers to apply the current knowledge from transcriptomics and methylomics. New treatment options will have to be developed, including immune or vaccination therapies, to allow the new diagnostic achievements to direct the individual patient to the best outcome.