The whole-genome landscape of medulloblastoma subtypes

Authors

Northcott PA

Buchhalter I

Morrissy AS

Hovestadt V

Weischenfeldt J

Ehrenberger T

Gröbner S

Segura-Wang M

Zichner T

Rudneva VA

Warnatz HJ

Sidiropoulos N

Jones DTW

Kool M

Korbel JO

Brors B

Schlesner M

Eils R

Marra MA

Pfister SM

Taylor MD

Lichter P

Doi

10.1038/nature22973

PMID: 28726821 · DOI: 10.1038/nature22973 · Journal: Nature (2017)

TL;DR

The largest integrated genomic and epigenomic survey of medulloblastoma to date — 491 sequenced tumours plus 1,256 methylation-profiled cases — by the ICGC PedBrain Tumour and MAGIC consortia. Whole-genome sequencing across all four consensus subgroups (WNT, SHH, Group 3, Group 4) confidently assigns driver events to 76% of Group 3 and 82% of Group 4 cases, more than doubling the previously explained fraction. The study nominates new subgroup-specific drivers including hotspot in-frame insertions in KBTBD4 and an “enhancer-hijacking” mechanism in which Group 4-restricted SNCAIP tandem duplications activate the neighbouring PRDM6 gene. DNA methylation t-SNE on Group 3 and Group 4 (n=740) refines these subgroups into eight subtypes, each enriched for distinct driver alterations.

Cohort & data

Key findings

  • Driver events were confidently assigned to 76% of Group 3 (n=131) and 82% of Group 4 (n=193) patients in this study, compared to ~30% in previous NGS series of n=173 — a more-than-doubling of explanatory power per subgroup.
  • All 36 sequenced WNT tumours were explained by at least one driver: 86% had somatic CTNNB1 hotspot mutations; three CTNNB1-wild-type WNT tumours carried pathogenic germline APC variants (Turcot syndrome). Monosomy 6 was present in 83% of WNT cases — confirming that ~10–15% of bona fide WNT MBs lack both routinely-used clinical features.
  • New WNT-associated alterations: CSNK2B (14%), EPHA7 (8%), and SWI/SNF subunits SMARCA4 / ARID1A / ARID2 collectively in 33% of WNT tumours.
  • SHH MB: ≥1 driver assigned to >95% of patients. Six IDH1 R132C mutations identified (5 SHH, 1 WNT); IDH1-mutant SHH tumours were CIMP+. HAT-complex genes (CREBBP, KAT6B, EP300) and HAT-regulators (BRPF1, KANSL1) were recurrently mutated (19% of SHH; q=2.2×10⁻³ for HAT-complex enrichment).
  • 24 mutational signatures detected with ≥5% contribution. Signature 3 (BRCA1/2-associated) was unexpectedly seen in most Group 3 and Group 4 patients; signature 18 enriched in Group 3 (P=4.7×10⁻⁵) and signature 5 in Group 4 (P=1.0×10⁻¹¹, age-correlated).
  • Two hypermutator outliers explained by mutator-gene hits: ICGC_MB62 (>25,000 mutations) carried a somatic POLE mutation and signature-10 substitutions; ICGC_MB265 carried a somatic MLH1 mutation with signature 6 (mismatch-repair deficiency).
  • Hotspot in-frame insertions in KBTBD4: 18/20 (90%) of somatic KBTBD4 variants were in-frame indels clustered in a 6-amino-acid stretch (residues 308–313) of the Kelch domain. Subgroup-specific predominant insertions: Group 3 R313delinsPRR; Group 4 P311delinsPP. KBTBD4 mutants were tightly enriched in methylation subtypes II (21%) and VII (14%) — the most prevalent driver in those subtypes.
  • “Enhancer-hijacking” activates PRDM6: Group 4-restricted SNCAIP tandem duplications (~600 kb upstream of PRDM6) reposition the SNCAIP super-enhancer to drive PRDM6 (P=6.07×10⁻²⁴ for PRDM6 expression in SV+ vs SV-; ≥20-fold upregulation). PRDM6 is now the most prevalent driver alteration in Group 4 (17% of n=193).
  • t-SNE on the 12,454 most variable methylation probes (s.d.>0.25) across 740 Group 3/4 cases resolves eight molecular subtypes (I–VIII), each with distinctive driver-event enrichment: GFI1B in subtype I, MYC in subtype II, KBTBD4 in II and VII, KDM6A and ZMYM3 in subtype VIII. Iterative down-sampling showed subtype structure stabilises at ≥500 samples.
  • Group 3 (n=131) recurrent drivers (% of subgroup): MYC amplification 17%, GFI1B SVs 11%, SMARCA4 9%, CTDNEP1 / KMT2D / KBTBD4 5%, MYCN 5%, GFI1 4%, OTX2 3%, BRCA2 germline 2%.
  • Group 4 (n=193) recurrent drivers: PRDM6 17%, GFI1B 9%, KDM6A 7%, ZMYM3 6%, OTX2 6%, KMT2C 6%, KBTBD4 6%, MYCN 6%, ZIC1 4%, CDK6 4%, TERT 3%, FLG 3%, KMT2D 3%, MED12 2%, SMARCA4 2%.
  • Pathway analysis: significant overrepresentation of Notch and TGFβ signalling alterations in Group 3 (first report of Notch pathway mutations in MB) and chromatin-modifier alterations in Group 4.
  • Recurrent fusion transcripts integrating SV breakpoints with RNA-seq targeted GLI2, PTEN and PVT1.
  • TERT promoter hotspots were the only confidently called noncoding driver region across the cohort.

Genes & alterations

  • KBTBD4: Recurrent in-frame insertions in the Kelch domain (residues 308–313); 18/20 KBTBD4 variants were in-frame indels. Predominant Group 3 insertion R313delinsPRR; predominant Group 4 insertion P311delinsPP. Highly enriched in methylation subtypes II (21%) and VII (14%); homology modelling places insertions on the substrate-binding interface, suggesting altered substrate recognition rather than loss of fold.
  • PRDM6: SET-domain protein activated by SNCAIP-locus tandem duplications in 17% of Group 4 (most prevalent Group 4 driver). PRDM6 expression upregulated ≥20-fold in SV+ vs SV- cases (P=6.07×10⁻²⁴).
  • SNCAIP: Group 4-restricted tandem duplications; locus overlaps a Group 4-specific super-enhancer that the duplications reposition into the neighbouring TAD containing PRDM6. SV breakpoints cluster near CTCF sites at the TAD boundary separating SNCAIP and PRDM6.
  • CTNNB1: Hotspot mutations in 86% of WNT MBs; sometimes subclonal alongside subclonal DDX3X variants.
  • APC: Pathogenic germline variants in 3 CTNNB1-wild-type WNT MBs — Turcot syndrome predisposition.
  • PTCH1, SUFU, TP53: Germline variants predominantly restricted to SHH MB.
  • SMO: Mutated in subset of SHH MBs.
  • CSNK2B: Clinically actionable, mutated in 14% of WNT MBs.
  • EPHA7: WNT-associated, 8%.
  • SMARCA4, ARID1A, ARID2: SWI/SNF subunits — 33% of WNT MBs collectively; rationale for PRC2 inhibitors in trial NCT02601937.
  • CREBBP, KAT6B, EP300, BRPF1, KANSL1: Recurrent mostly-SHH HAT-complex alterations (19% of SHH).
  • IDH1: R132C in 6 cases (5 SHH, 1 WNT); IDH1-mutant SHH tumours are CIMP+, mirroring G-CIMP in glioma.
  • KMT2D, KMT2C, KDM6A: Recurrent histone-methylation-modifier mutations across subgroups; KDM6A and ZMYM3 especially enriched in methylation subtype VIII.
  • MYC: Amplifications restricted to Group 3 (17%).
  • MYCN: Amplifications in Group 3 (5%) and Group 4 (6%) at comparable frequencies.
  • GFI1, GFI1B: SV-driven aberrant induction (mutually exclusive); distributed in both Group 3 and Group 4. GFI1B was a top CESAM-confirmed enhancer-hijacking target.
  • OTX2: Recurrent in Group 3 (3%) and Group 4 (6%).
  • TERT: Promoter hotspots — only confident noncoding driver region across the cohort.
  • POLE, MLH1: Somatic mutations explaining the two cohort hypermutators (ICGC_MB62 and ICGC_MB265 respectively).
  • BCOR, FBXW7, TCF4, ZIC1, GSE1, CTDNEP1, CHD7, CDK6, MED12, STAG2, BRCA2, FLG: Additional recurrently altered candidate drivers stratified across subgroups in Figs 2–3 oncoprints.
  • Fusion transcripts: recurrent rearrangements targeting GLI2, PTEN and PVT1.

Clinical implications

  • The reliance on CTNNB1 mutation or chromosome-6 monosomy to clinically assign WNT MB will miss ~10–15% of bona fide WNT cases — APC germline testing is warranted when WNT MB is suspected without CTNNB1 mutation (Turcot syndrome).
  • SWI/SNF mutations in WNT MB (33% of WNT) provide rationale for PRC2 inhibitors already being evaluated in paediatric SWI/SNF-mutant trials (e.g. NCT02601937).
  • HAT-complex alterations across 19% of SHH MBs suggest HAT-pathway-targeted epigenetic therapy as an avenue meriting further study.
  • IDH1-mutant SHH MBs (CIMP+) phenocopy IDH-mutant glioma, opening the door to repurposing IDH1 inhibitors in this small subset.
  • KBTBD4 hotspot insertions and PRDM6 enhancer-hijacking are nominated as new actionable targets in Group 3/4 MB, but require functional validation before clinical translation.
  • Eight methylation-defined Group 3/4 subtypes provide a substrate for subtype-tailored risk-stratification and trial design.
  • The single MLH1-mutated mismatch-repair-deficient case (signature 6) is the kind of MB patient who could be considered for immune-checkpoint blockade; the POLE hypermutator likewise.

Limitations & open questions

  • Functional validation is missing for the central new candidates: KBTBD4 hotspot insertions and PRDM6 enhancer-hijacking are nominated by integrative genomics, but their oncogenic mechanisms (KBTBD4 substrate altered? PRDM6 methyltransferase activity essential?) and “actionability” remain to be experimentally established.
  • Sample size, although the largest to date, is still underpowered for the long tail of <5% recurrently mutated genes that the authors call “biologically crucial” but cannot yet statistically distinguish from passengers.
  • The eight methylation subtypes within Group 3/4 are defined by the cohort; subtype boundaries and the precise developmental cell-of-origin still require subtype-relevant model systems.
  • Whether SWI/SNF mutations confer the same PRC2-inhibitor sensitivity in WNT MB as in other paediatric SWI/SNF-mutant tumours is untested.
  • The authors observe that signature 3 (BRCA1/2-associated) is unexpectedly common in Group 3 and Group 4, but the underlying somatic or germline lesion(s) responsible are not resolved.
  • Notch and TGFβ signalling enrichment in Group 3 are reported as the first formal genomic evidence — functional studies needed to determine whether they are druggable in MB.
  • Cohort skews paediatric (median age 8 years); generalisability to the rare adult MB population is limited despite an age range up to 50 years.

Citations from this paper used in the wiki

  • “Here we analyse the somatic landscape across 491 sequenced medulloblastoma samples and the molecular heterogeneity among 1,256 epigenetically analysed cases” (Abstract).
  • “we confidently assigned potential driver events to 76% and 82% of Group 3 and Group 4, respectively, more than doubling the proportion of explained cases per subgroup” (Group 3 and Group 4 subgroup MB).
  • “Of 20 somatic KBTBD4 variants we identified, 18 (90%) were determined to be in-frame insertions clustered across just six amino acids within the KBTBD4 Kelch domain” (Hotspot insertions target KBTBD4).
  • “the predominant insertion inferred in Group 3, of proline and arginine at Arg313 (R313>PRR) differed from that observed in Group 4, an insertion of proline at Pro311 (P311>PP)” (Hotspot insertions target KBTBD4).
  • “the structural variants observed in PRDM6-activated Group 4 converge on the SNCAIP super-enhancer, consistent with enhancer hijacking” (Enhancer hijacking activates PRDM6).
  • “PRDM6 expression was markedly upregulated (at least 20-fold), considerably more than any other gene mapping within the proximal TADs including SNCAIP” (Enhancer hijacking activates PRDM6).
  • “All 36 WNT MBs sequenced in this study were confidently explained by mutations in at least one or more driver genes… Somatic CTNNB1 mutations, the hallmark feature of WNT-driven MB, were found in 86% of patients.” (WNT subgroup MB).
  • “Three CTNNB1 wild-type MBs harboured pathogenic APC germline variants, explaining the WNT pathway activation seen in these patients and underscoring the need to perform genetic testing for APC carrier status (that is, Turcot syndrome) when WNT MB is suspected despite failure to detect mutant CTNNB1.” (WNT subgroup MB).
  • “We identified six IDH1 R132C mutations (five SHH, one WNT)… IDH1-mutant SHH MBs were determined to be CIMP+” (SHH subgroup MB).
  • “ICGC_MB62 (SHH) harboured over 25,000 mutations and was the only MB we considered a bona fide hypermutator among the cohort” (Mutational signatures operative in MB).

This page was processed by crosslinker on 2026-05-15.