A CTNNB1-altered medulloblastoma shows the immunophenotypic, DNA methylation, and transcriptomic profiles of SHH-activated, and not WNT-activated, medulloblastoma

Recent advancement in molecular characterisation has identified four principal molecular groups of medulloblastoma (MB), namely WNT, SHH, group 3, and group 4, and each has its characteristic clinical features, predilection for specific anatomic sites, signature genetic alterations, and distinct DNA methylome profiles [1]. Immunophenotypically, MB can be divided into WNT, SHH, and non-WNT/non-SHH groups by their expression of YAP1 and GAB1—WNT MB expresses YAP1, in addition to its characteristic nuclear β-catenin positivity, and SHH MB expresses both. In contrast, non-WNT/non-SHH MB is negative for both [2]. Thus far, CTNNB1 mutations have been considered pathognomonic of WNT MB. Furthermore, it has been shown that CTNNB1 mutations dominantly drive the WNT-activated phenotype in MB, even in the presence of alterations in the SHH pathway [3, 4].

We herein report an illustrative case that challenges this belief. The patient was a 2-year-old, previously healthy male who presented with ataxia, emesis, and regression of developmental milestones. Imaging revealed a large posterior fossa mass with associated obstructive hydrocephalus (Fig. 1a). The epicentre of the tumour was slightly lateral within the fourth ventricle. No metastases were identified, and gross total resection was achieved. The histologic sections showed a biphasic embryonal tumour with alternating hypercellular areas and areas showing a neuropil-like matrix with decreased cell densities (Fig. 1b). Anaplasia was only appreciated focally. Ki-67 labelling further highlighted the biphasic nature of the tumour, showing the proliferative hypercellular areas and pale nodules of low proliferative indices (Fig. 1c). However, unlike MB with extensive nodularity or desmoplastic/nodular MB prevalent in this age group, there was no stromal desmoplasia between the pale nodules revealed by reticulin stain (Fig. 1d). Tumour cells in the hypercellular areas were strongly positive for YAP1 (Fig. 1e). There was patchy GAB1 immunoreactivity (Fig. 1f). No strong nuclear β-catenin immunoreactivity was observed (Fig. 1g). Immunostaining for p53 variably marks a small subset of tumour cell nuclei. Homozygous deletion of PTCH1 was detected by interphase fluorescence in situ hybridisation, while copy number gain or amplification of MYC and MYCN and monosomy 6, a signature finding in WNT MB, were not identified. A diagnosis of SHH-activated and TP53-wildtype MB with classic, biphasic histology was made based on the tumour’s histopathologic, immunophenotypic, and molecular findings.

Fig. 1. SHH-activated medulloblastoma with pathogenic CTNNB1 mutation.

(a) T1-weighted post-contrast magnetic resonance imaging shows a large posterior fossa tumour filling the fourth ventricle. (b) The biphasic tumour shows pale nodules of neurocytic differentiation in between hypercellular areas with embryonal cytology. (c) A Ki-67 stain further highlights the biphasic pattern. (d) There is no internodular desmoplasia. The tumour cells are positive for YAP1 (e) and GAB1 (f) and show no strong nuclear β-catenin immunoreactivity (g). (h) Pathogenic CTNNB1 S45Y variant was identified by whole exome sequencing and RNA sequencing. Methylation (i) and transcriptomic (j) profiles of this tumour cluster with SHH MB and are distinct from WNT MB. (k) Copy number variation plot shows the homozygous deletion of PTCH1, a frequent finding in SHH MB, and lack of monosomy 6, a signature finding of WNT MB.

Whole exome sequencing (WES) and transcriptome (RNA-seq) analysis of this tumour, however, revealed pathogenic CTNNB1 S45Y (NM_001904.3) (Fig. 1h) and MAX R60Q (NM_002382.4) variants, without additional alterations in the SHH pathway. The allele frequency of the CTNNB1 S45Y variant (59% WES, 52% RNA) supported its clonal presence in the tumour cells (Supplementary Fig. 1). PTCH1 alterations have been described in a small subset of WNT MB [3–5], but WNT pathway activating mutations have not been reported in SHH MB. To resolve the genotype-phenotype discrepancy as to whether the tumour reported herein should be classified as SHH or WNT MB, which has a significant impact on therapeutic strategies and prognosis, an in-house developed neural network classifier (St. Jude MLPnet version 1.0) based on DNA methylation data from the Infinium MethylationEPIC BeadChip array (Illumina, Inc., San Diego, CA) was utilised. The analysis revealed a high calibrated score for infant SHH MB (0.93/1.00). This finding was further supported by the t-SNE plots using the tumour’s DNA methylation (Fig. 1i) and transcriptomic profiles (Fig. 1j) and concurred with the result from the DKFZ’s brain tumour classifier v12.3 (https://www.molecularneuropathology.org/mnp), which revealed methylation class family medulloblastoma, SHH, and methylation class medulloblastoma, subclass SHH B (infant) with calibrated scores 0.95 and 0.90, respectively. CNV plot derived from the methylation array supports the homozygous deletion of PTCH1 without a loss of chromosome 6 (Fig. 1k). As expected for a SHH MB, RNA expression of OTX2 in the tumour was very low (Supplementary Fig. 2). These findings guided the therapeutic approach for this patient, and a radiation-sparing strategy was adopted.

The CTNNB1 S45Y mutation detected in this tumour has been identified in several cancer types of hepatic, renal, dermal, soft tissue, and adrenocortical origins (http://cancer.sanger.ac.uk/cosmic). This variant falls within a mutational hotspot region in exon 3 of CTNNB1 that plays a critical role in β-catenin degradation. It is predicted to have a deleterious effect on protein function by five of seven in silico algorithms [6–12]. In vitro and in vivo studies suggest that this variant has weak WNT/β-catenin activating potential and is insufficient to drive tumorigenesis by itself [13, 14]. The lack of β-catenin nuclear staining in the tumour presented herein is consistent with this notion, supporting that different CTNNB1 variants have diverse biological characteristics and oncogenic ability [15]. And yet, the clonal nature of the somatic CTNNB1 mutation supports its role in this tumour. Of note, none of the reported WNT MBs with SHH pathway alterations harbour CTNNB1 S45 mutations [3, 5, 16]. In the report by Koch et al., a desmoplastic MB in an adult patient was found to have a PTCH1 variant and an activating S37C CTNNB1 variant [17]. However, the tumour was not molecularly classified as a WNT or as an SHH MB. In addition, the tumour reportedly showed a strong nuclear accumulation of β-catenin by immunohistochemistry, unlike the MB reported by our group, which supports a stronger CTNNB1 activating effect and WNT pathway signalling activation of codon 37 variants, compared with codon S45 alterations. Thus, the immunophenotype and transcriptional and epigenetic characteristics of the MB reported herein could be explained by the weakly activating effect of the CTNNB1 S45Y variant in conjunction with the homozygous PTCH1 deletion in this tumour.

The MAX R60Q mutation has been recurrently identified in many cancers, including SHH MB [16]. MAX is a transcription factor that serves as a partner for MYC. The R60 residue is a critical DNA and protein binding site and regulates the stability of the MAX homodimer complex [18, 19]. The R60Q variant disrupts MAX homodimerisation without affecting MYC/MAX heterodimerisation, resulting in upregulation of MYC [20].

In summary, we have presented an MB with a pathogenic CTNNB1 mutation that otherwise showed the histopathology, immunophenotype, and methylation and transcriptomic profiles of an SHH MB. Our illustrative example emphasises the diagnostic value of the immunohistochemistry panel with YAP1, GAB1, and β-catenin and DNA methylation profiling, combined with exome sequencing, in the characterisation of MB. CTNNB1 mutations are not specific for WNT MB, and different CTNNB1 mutations have diverse oncogenic potential.

Supplementary Material

fS1

Supplementary Fig. 1. Sequencing reads supporting the presence of CTNNB1 S45Y variant and demonstrating its allele frequency.

fS2

Supplementary Fig. 2. OTX2 Expression in the SHH-activated medulloblastoma with pathogenic CTNNB1 mutation is low and similar to other SHH-activated medulloblastomas (a) Normalised counts of mRNA expression in the tumour (Case) as well as WNT-activated, SHH-activated, group 3 (G3), and group 4 (G4) reference medulloblastomas (b) Integrative Genomics Viewer plot showing the low expression of OTX2 in the tumour.

Key Points.

1. CTNNB1 mutations are not specific for WNT-activated medulloblastoma.

2. Different CTNNB1 mutations have diverse oncogenic potential.

3. Our illustrative example emphasises the diagnostic value of the immunohistochemistry panel with YAP1, GAB1, and β-catenin and DNA methylation profiling, combined with exome sequencing, in the characterisation of medulloblastoma.