Figure 4
Application of MS-MIMIC to remnant, archival materials from the HIT-SIOP-PNET4 clinical trial cohort enables the first subgroup-specific characterisation of standard-risk medulloblastoma. (a) The HIT-SIOP-PNET4 clinical trial ran prior to the discovery of medulloblastoma molecular subgroups (2000–2007) and remnant tumour sample materials available were old, scant and unsuitable for array-based DNA methylomic subgrouping. (b,c) First and second replicates of individual samples show concordance both at the level of estimate of methylation (β-value; R2 = 0.59; n = 55 replicates, 17 CpG loci) and subgroup assigment calls (55/55 replicates; 100%). (d) Bar plot of MS-MIMIC assay input dsDNA amounts. The proportions of samples that were successfully subgrouped are shown grey and were consistent across different dsDNA input quantities. Lilac represents samples for which subgrouping was not possible due to QC failure in vitro or in classification. (e) Proportion of attempted samples (153/161; 95%) successfully subgrouped (107/153; 70%) or failing the assay due to failure to 1) bisulfite convert (9/153; 6%) 2) meet QC citerion for CpG-locus specific failure (>6/17 CpG-locus fails, 24/153 samples; 16%), 3) meet probablility threshold in classification confidence (13/153; 8%). (f) Subgroup assignment and probability estimates (dot) along with 95% confidence intervals (boxplots) for 120 standard risk HIT-SIOP-PNET4 tumours after applying the confidence probability threshold (red-line). (g) All samples which were CTNNB1 mutated (CTNNB1 mut; a well-established marker of WNT medulloblastoma) were assigned as WNT by MS-MIMIC. No non-WNT tumours were CTNNB1 mut. (h) Progression free survival (PFS) Kaplan-Meier curves for MS-MIMIC derived subgroups reveal that standard-risk Grp4 medulloblastomas show a significantly worse disease outcome compared to other subgroups (p = 0.038, log-rank test). Numbers below x-axis represent patients at risk of event.
