MethPed: a DNA methylation classifier tool for the identification of pediatric brain tumor subtypes

Abstract

Background: Classification of pediatric tumors into biologically defined subtypes is challenging, and multifaceted approaches are needed. For this aim, we developed a diagnostic classifier based on DNA methylation profiles.

Results: Methylation data generated by the Illumina Infinium HumanMethylation 450 BeadChip arrays were downloaded from the Gene Expression Omnibus (n = 472). Using the data, we built MethPed, which is a multiclass random forest algorithm, based on DNA methylation profiles from nine subgroups of pediatric brain tumors. DNA from 18 regional samples was used to validate MethPed. MethPed was additionally applied to a set of 28 publically available tumors with the heterogeneous diagnosis PNET. MethPed could successfully separate individual histology tumor types at a very high accuracy (κ = 0.98). Analysis of a regional cohort demonstrated the clinical benefit of MethPed, as confirmation of diagnosis of tumors with clear histology but also identified possible differential diagnoses in tumors with complicated and mixed type morphology.

Conclusions: We demonstrate the utility of methylation profiling of pediatric brain tumors and offer MethPed as an easy-to-use toolbox that allows researchers and clinical diagnosticians to test single samples as well as large cohorts for subclass prediction of pediatric brain tumors. This will immediately aid clinical practice and importantly increase our molecular knowledge of these tumors for further therapeutic development.

Keywords: 450 K; Astrocytoma; Classifier (classification tool); DNA methylation; Ependymoma; GBM; Medulloblastoma; MethPed; PNET; Random forest.

Fig. 1

Accuracy of the MethPed classifier. a Classification accuracy of individual methylation probes in one vs all other logistic regression analyses. The boxplots displays the classification accuracy as measured by the area under the curve (AUC values or c-statistics) for the 100 probes per tumor subtype that provided the highest predictive power; b Confusion matrix showing an extremely high predictive capacity of MethPed, illustrated by the high percentage of correct classification of randomly drawn pairs; and c Decision boundaries for five tumor types exemplifying the possibility to delimitate a certain tumor type from the rest based on the two probes that proved to be the best for each tumor in one vs all other regression analyses

Fig. 2

Histopathological and molecular analyses of two patients in the regional cohort. a Four-year-old child (BPC A7) diagnosed with a PNET in the right hemisphere. MethPed classification (upper panel). H&E shows polymorphic, anaplastic cells and regions with necrotic areas; synaptophysin shows clonal positivity; GFAP mostly negative areas but also individual tumor cells with very strong expression and Ki-67 variable positivity (middle section, original magnification of the objective in all cases ×40). Magnetic resonance imaging (MRI) shows the location of the tumor, and Sanger sequencing chromatogram shows a HIST1H3B Lys27Met mutation in the tumor. Red arrow indicates the site of the mutation (lower panel). b Twelve-year-old child (BPC B5) diagnosed with a PNET in the brain stem. MethPed classification (upper panel). H&E shows cells variable in morphology with areas of rosette formation similar to Homer-Wright type; synaptophysin areas with granular cytoplasmic pattern and other areas with diffuse positivity as well as negative cells; GFAP positivity in a high number of cells indicates an unusual high incidence of astrocytic differentiation and high positivity of Ki-67 (middle section, objective original magnification ×40). Magnetic resonance imaging (MRI) shows the location of the tumor, and Sanger sequencing chromatogram shows a H3F3A Lys27Met mutation in the tumor. red Arrow indicates the site of the mutation (lower panel)

Fig. 3

Immunohistochemical analyses of two patients with challenging diagnoses. a Infant (case BPC C1) diagnosed according to the WHO criteria with a large-cell medulloblastoma, located in the vermis. H&E shows predominantly large cells with a high frequency of apoptotic bodies, clonal positivity of GFAP, and low positivity for synaptophysin and clonal areas with high Ki-67 positivity (objective original magnification × 40). b Four-year-old child (BPC B7) diagnosed with an intra- and periventricular PNET tumor. H&E shows high frequency of necrosis and vessels, very strong, clonal positivity of GFAP in tumor cells as well as positivity in reactive gliosis, high positivity for synaptophysin, and high Ki-67 positivity (objective original magnification ×40)