Genome sequencing of SHH medulloblastoma predicts genotype-related response to smoothened inhibition

Abstract

Smoothened (SMO) inhibitors recently entered clinical trials for sonic-hedgehog-driven medulloblastoma (SHH-MB). Clinical response is highly variable. To understand the mechanism(s) of primary resistance and identify pathways cooperating with aberrant SHH signaling, we sequenced and profiled a large cohort of SHH-MBs (n = 133). SHH pathway mutations involved PTCH1 (across all age groups), SUFU (infants, including germline), and SMO (adults). Children >3 years old harbored an excess of downstream MYCN and GLI2 amplifications and frequent TP53 mutations, often in the germline, all of which were rare in infants and adults. Functional assays in different SHH-MB xenograft models demonstrated that SHH-MBs harboring a PTCH1 mutation were responsive to SMO inhibition, whereas tumors harboring an SUFU mutation or MYCN amplification were primarily resistant.

Figure 1. Genetic and Epigenetic Differences between SHH-MBs from Infants, Children, and Adults

(A) Cluster analysis of DNA methylation and gene expression data of SHH-MB. Both methylation profiling (left; n = 129) and gene expression profiling (right; n = 103) reveal two SHH-MB subgroups identified by unsupervised k-means consensus clustering. Each row represents a methylation probe/expression probeset, each column represents a sample. The level of DNA methylation (b value) is represented with a color scale as depicted. For each sample patient age (blue, infants; yellow, children; and pink, adults) and clustering according to expression data or methylation data (when available) is shown. Grey indicates that no data were available. (B) GISTIC2 significance plots of amplifications (red) and deletions (blue) observed in SHH-MB infants, children, and adults. Candidate genes mapping significant regions have been indicated.

Figure 2. Number and Type of Somatic Mutations in Medulloblastoma Tumors in Relation to the Age of the Patient

(A) Total number of somatic mutations genome wide correlates with age of the patient. Plotted are the total number of somatic SNVs identified genome wide versus age of the patient for all cases for which we performed whole genome sequencing (WGS; n = 45). Red indicates patients harboring a TP53 mutation. (B) Same as in (A), but only the total number of coding SNVs is plotted versus age for all cases for which we performed either whole genome or whole exome sequencing (WGS and WES, n = 67). (C) Mutation signatures. Plotted are the total numbers of somatic mutations genome wide sorted by age of the patient. Coloring of bars represents the ratio of the six possible nucleotide changes (C > A, C > G, C > T, T > A, T > C, and T > G) for each sample. (D) Normalized mutation signatures sorted by age.

Figure 3. Genetic and Histological Differences between SHH-MBs from Infants, Children, and Adults

(A) SHH pathway mutations, gender, histology and 9q/10q/17p aberrations in all sequenced 133 SHH-MB. Cases have been split up in infants, children and adults, and are sorted based on type of mutation in the SHH-pathway. Potential response to SMO inhibition: cases with SHH amplifications, PTCH1 mutations, or SMO mutations will likely respond to SMO inhibition (indicated in green). Cases with SUFU mutations or MYCN or GLI2 amplifications will likely not respond to SMO inhibition (indicated in red). In cases for which no mutations in the SHH pathway were detected, it is not clear whether they will respond to SMO inhibitors (indicated in yellow). Percentages indicate fraction of infants, children, or adults, respectively, of each category. p Values indicate whether distributions are significantly different among infants, children, and adults. (B) Pie charts showing in infants, children, and adults with SHH the distribution of gender (male, blue; female, pink; unknown, gray), histology (classic, dark red; nodular/desmoplastic, green; large cell/anaplastic LCA, orange; MBEN, yellow; and unknown, gray), 9q loss (yes, black; no, gray), 10q loss (yes, black; no, gray), 17p loss (yes, black; no, gray), and type of SHH pathway mutation (SHH amp, purple; PTCH1 mut, red; SMO mut, green; SUFU mut, orange; GLI2/MYCN amp, blue; and unknown, gray). (C) Trimodal age distribution of patients with SHH-MB. Red line indicates age distribution of all patients with SHH-MB. Three subgroups make up this age distribution: young children with PTCH1 and SUFU mutations (blue line), older children with PTCH1 and TP53 mutations (purple line), and adults who mostly have PTCH1 or SMO mutations (green line). See also Figure S1.

Figure 4. Most Frequently Mutated Genes in SHH-MB and the Mutual Exclusivity of Mutations in Chromatin Modifier Genes

(A–C) Mutation frequencies of 33 genes that are mutated either in ≥5% of all SHH-MB cases or in ≥10% of SHH-MB cases in one of the age categories. Mutation frequencies for these 33 genes are shown in infants (A), children (B), and adults (C). Black indicates the fraction of mutations that is found in the germline. (D) Mutations in chromatin modifiers in infants, children, and adults with SHH-MB. The top line shows the mutations in the SHH pathway for each case. See also Figure S2.

Figure 5. Immunohistochemical Staining of MB Tissue Arrays for p-AKT and p-S6

(A) Example of positive p-AKT MB. (B) Example of negative p-AKT MB. (C) Example of positive p-S6 MB. (D) Example of negative p-S6 MB. (E) Overlap in staining results between p-AKT and p-S6. (F) Frequencies of p-AKT and p-S6 staining in infants, children, and adults. (G) Survival analysis for p-AKT and p-S6 in all SHH patients and in adults only. Numbers on the y-axis indicate the fraction of surviving patients. Numbers on the x-axis indicate the follow-up time in months. The number of patients per group is indicated next to the graphs plus the number of events within that group (between brackets). For infants and children, the number of patients staining positive was too low to draw conclusions from separate survival analyses.

Figure 6. SMO Antagonists Do Not Suppress Proliferation of All SHH-Associated MB Tumors

(A) Characteristics of SHH-MB models treated with LDE225. (B–D) Cells from patient-derived xenografts of SHH-associated MB were treated with DMSO (0.05% [hatched bars] or 0.25% [solid bars]) or LDE-225 (100 [hatched bars] or 500 nM [solid bars]). Cells were pulsed with [methyl-³H]thymidine (³H-Td) after 48 hr and harvested for analysis of ³H-Td incorporation at 66 hr. In DMB-012 (B), LDE-225 significantly inhibited ³H incorporation compared to DMSO control (p < 0.01 based on paired two-tailed t test). In RCMB-018 (C) and RCMB-025 (D), LDE-225 did not significantly inhibit ³H incorporation (p > 0.5 and p > 0.1, respectively). Data represent means of triplicate samples ± SD. (E and F) Cells from MB xenograft DMB-012 (E) or RCMB-018 (F) were infected with luciferase virus and transplanted into NSG mice. Bioluminescence images were taken pretreatment (day 0) and at different time points after daily treatment with vehicle or SHH antagonist (LDE-225, 5 or 20 mg/kg/day). Five mice per group were used. Representative examples from each group are shown. Other examples are shown in Figure S3. A red cross indicates when mice were sacrificed. (G and H) Kaplan-Meier survival plots for the mice harboring DMB-012 tumors (G) or RCMB-018 tumors (H) and treated with vehicle or LDE-225. (I) RCMB-018 cells were treated with DMSO (0.25%; gray bar), LDE-225 (500 nM; red bar), vehicle (PBS + 0.01 N NaOH; light blue bars), or increasing concentrations of ATO (dark blue bars). Cells were pulsed with [methyl-³H] thymidine (³H-Td) after 48 hr and harvested for analysis of ³H-Td incorporation at 66 hr. LDE-225 did not inhibit ³H incorporation compared to DMSO control, but ATO did at 5 and 10 μM concentrations. Data represent means of triplicate samples ± SD. See also Figure S3.

Figure 7. Schematic Overview of SHH-, PI3K/AKT/mTOR-, and PKA Pathways and How They Interact

Genes that were found in the genomic analyses of SHH-MBs to harbor activating mutations (green stars), inactivating mutations (red stars), or were found to be amplified (MYCN and GLI) are indicated. All these mutations lead to activation of GLI proteins and their downstream pathways. Options for targeted treatment are indicated. Patients harboring mutations in either PTCH1 or SMO should be responsive to SMO inhibitors, whereas patients harboring mutations more downstream in the SHH pathway (SUFU, MYCN, and GLI) or in the PI3K/AKT/mTOR and/or PKA-pathways may be treated using arsenic trioxide (ATO) or other more specific GLI-inhibitors or PI3K/AKT/mTOR inhibitors.