Integrated molecular analysis of adult sonic hedgehog (SHH)-activated medulloblastomas reveals two clinically relevant tumor subsets with VEGFA as potent prognostic indicator

Abstract

Background: Up to now, adult medulloblastoma (MB) patients are treated according to the protocols elaborated for pediatric MB although these tumors are different in terms of clinical outcomes and biology. Approximately 70% of adult MB disclose a sonic hedgehog (SHH) molecular signature in contrast to about 30% in pediatric cohorts. In addition, adult SHH-MB (aSHH-MB) are clinically heterogeneous but there is consensus neither on their optimal treatment nor on risk stratification. Thus, the identification of clinically relevant molecular subsets of aSHH-MB and identification of potential treatment targets remains inconclusive.

Methods: We analyzed 96 samples of institutionally diagnosed aSHH-MB through genome-wide DNA methylation profiling, targeted DNA sequencing, and RNA sequencing to identify molecular subcategories of these tumors and assess their prognostic significance.

Results: We defined two aSHH-MB numerically comparable epigenetic subsets with clinical and molecular variability. The subset "aSHH-MBI" (46%/48%) was associated with PTCH1/SMO (54%/46%) mutations, "neuronal" transcriptional signatures, and favorable outcomes after combined radio-chemotherapy (5-year PFS = 80% and OS = 92%). The clinically unfavorable "aSHH-MBII" subset (50%/52%; 5-year PFS = 24% and OS = 45%) disclosed GLI2 amplifications (8%), loss of 10q (22%), and gene expression signatures associated with angiogenesis and embryonal development. aSHH-MBII tumors revealed strong and ubiquitous expression of VEGFA both at transcript and protein levels that was correlated with unfavorable outcome.

Conclusions: (1) The histologically uniform aSHH-MB cohort exhibits clear molecular heterogeneity separating these tumors into two molecular subsets (aSHH-MBI and aSHH-MBII), which are associated with different cytogenetics, mutational landscapes, gene expression signatures, and clinical course. (2) VEGFA appears to be a promising biomarker to predict clinical course, which needs further prospective validation as its potential role in the pathogenesis of this subset.

Keywords: SHH; VEGFA; adult; medulloblastoma; prognosis; transcriptome.

Fig.1

(A) Both molecular aSHH-MB variants (see B, C) revealed similar histological appearance designated as desmoplastic/nodular MB variant (the scale bars: 50 µm). (B) Unsupervised hierarchical clustering (UCL) of DNA methylation of the full adult SHH MB cohort (n = 96) based on the 20 000 most variable methylation probes (SD > 0.30). Two main epigenetic clusters were identified (aSHH-MBI—red line and aSHH-MBII—blue line). (C) Two-dimensional t-distributed stochastic neighbor embedding (t-SNE) analysis of adult SHH MB also revealed two tumor methylation subgroups—aSHH-MBI (blue spots) and aSHH-MBII (green spots). (D) Oncoprint depicting clinical characteristics, chromosomal aberrations, and driver SHH-activated gene mutations which are differentially distributed in two aSHH-MB subgroups, derived from UCL and t-SNE clustering. (E) Survival analysis for two epigenetic aSHH-MB variants shows that progression-free and overall survival for aSHH-MBII patients (green line) were significantly worse than those for aSHH-MBI patients (black line; log-rank test; P < .01).

Fig. 2

(A) A set of 20 top most-confident genes differentially overexpressed in aSHH-MBI (violet top line) and aSHH-MBII (blue top line), respectively, allows one to discriminate clearly between these tumor cohorts. (B) Gene Ontology analysis disclosed that aSHH-MBI RNA profiles (blue circles) were associated with neuronal metabolism, synaptic transmission, and phagocytosis. In contrast, expression profiling signatures of aSHH-MBII (red circles) were characterized by genes involved in angiogenesis, tissue morphogenesis, and embryonal development. (C) VEGFA expression level is significantly higher (Mann-Whitney U test, P < .01) in aSHH-MBII (blue boxplot; n = 38) in comparison to aSHH-MBI (violet boxplot; n = 36), children TP53 wt SHH MB (orange boxplot; n = 16), infant SHH-MB (red boxplot; n = 73). However, VEGFA expression was also high in TP53 mutant SHH MB (pink boxplot; n = 11). (D) Survival analysis for VEGFA expression in aSHH-MB shows that both overall survival (OS) and progression-free survival (PFS) were significantly worse for those with gene overexpression harboring log2 Read per Million Kilobase (RPKM) > 2.5 (green line; log-rank test; P < .01). (E) Diffuse VEGFA immunoexpression in aSHH-MB allocated to MBII cohort (the scale bars: 50 µm). (F) Completely VEGFA-negative tumor sample with aSHH-MBI molecular signature (the scale bars: 50 µm). (G) Survival analysis for 96 aSHH-MB shows that both OS and PFS were significantly worse for patients harboring VEGFA immunopositive tumor samples (green line; log-rank test; P < .01).