1. Clinical Characteristics of G34-DHGs
The clinical characteristics of the 10 patients with G34-DHGs from SYSUCC are summarized in Table 1. The patient group included four male patients and six female patients. The patient age at initial diagnosis ranged from 13 to 25 years, and the median patient age was 20.5 years. The symptoms were the classic clinical features of brain tumors, including dizziness, headache, nausea, vomiting, weakness of right limbs and generalized tonic-clonic seizures dependent on the site of the lesion. All lesions were primarily located in cerebral hemispheres; the lesions were in the right hemisphere in five patients, the left hemisphere in four patients and bilateral hemisphere in one patient. Tumor invasion in the frontal lobe was detected in eight patients, and invasion in the parietal lobe was detected in three patients; temporal lobe involvement was observed in two patients and insular lobe involvement was observed in two patients. In addition, corona radiata, basal ganglia or corpus callosum were infiltrated by tumors in four patients.
For initial treatment, gross total resection (GTR) was performed in five patients, whereas four patients received subtotal resection and one patient only received biopsy. For postoperative adjuvant treatment, nine patients received adjuvant concurrent chemotherapy and radiation therapy (CCRT), and one patient who was included in a clinical trial received 12 courses of dianhydrodulcitol. Among the nine patients receiving CCRT, five patients further received maintenance temozolomide (TMZ) treatment and one patient received four courses of TMZ plus cisplatin followed by maintenance TMZ treatment.
Seven patients showed recurrent disease after the initial treatment, and four patients died. Four patients who initially received subtotal resection chose conservative treatment after recurrence, and two died at 6 and 16 months after the initial operation. The remaining three patients with recurrent disease received a second operation, and two of them further received adjuvant therapy after the second operation. Patient 9 received TG02, a novel pyrimidine-based multi-kinase inhibitor of CDKs together with JAK2 and FLT3, after STR for recurrent disease. This patient died at 1 year after recurrence with an OS of 17 months. Patient 1, who was diagnosed with methylated MGMT promoter after initially receiving GTR followed by CCRT and 12 courses of maintenance TMZ treatment, then again underwent GTR followed by radiotherapy and 12 courses of maintenance TMZ treatment for her recurrent disease; this patient achieved the longest OS of 75 months. Moreover, three patients who were also diagnosed with methylated MGMT promoter after initially receiving GTR followed by CCRT and maintenance TMZ treatment were free of recurrence after at least 18 months since the initial operation. Intriguingly, two of the three patients without recurrent disease continuously received maintenance TMZ treatment (ongoing), including patient 4, who initially received only biopsy. These results indicated that only GTR and long-term maintenance TMZ treatment might benefit patients with G34-DHG in conventional treatment.
2. Pathological Characteristics of G34-DHGs
Histopathologic examination of the surgical specimens from our cohort showed high-grade glioma morphology (Figure 1). All cases showed high cell density, featuring nuclear atypia, mitotic activity and cellular pleomorphism with focal gemistocytic cells and multinuclear giant cells. Most cases were glioblastoma (GBM)-like with microvascular proliferating and/or palisade necrosis and two cases showed focal embryonal appearance. However, calcification, perivascular growth pattern and perineuronal satellitosis, which rarely appears in GBM, were also observed (Figure 1E, F). All cases showed GFAP expression and negative expression of OLIG2; most cases were negative for ATRX (9/10) and most showed diffuse strong p53 positivity (8/10). The Ki67 labeling index was high, ranging from 20% to 60%. Sanger sequencing revealed that all cases were IDH wild-type and TERT promoter wild-type (Table 1). Methylation PCR revealed that most cases showed MGMT promoter methylation (8/10) (Table 1).
3. Mutational Landscape of G34-DHGs
To investigate the mutational landscape of G34-DHGs, we performed WES on five fresh samples and matched peripheral blood leucocytes from our cohort. The average depths of targeted exome regions in tumors and matched blood samples were 400× and 100×, respectively. More than 98.69% of the targeted regions were covered sufficiently for confident variant calling (≥10× depth). We included three G34-DHG cases from the CGGA and seven G34-DHG cases from the HERBY trial (study BO25041; clinicaltrials.gov NCT01390948). The total somatic mutations of the 15 G34-DHG samples are listed in Figure 2.
The mutational frequencies, mutation types and clinical features of the 15 G34-DHGs are displayed in Figure 2A. TP53 (13/15 87.0%), PDGFRA (12/15, 80.0%) and ATRX (10/15, 67.0%) were the most three frequently mutated genes. Other frequently mutated genes including MUC17 (8/15, 53.0%), MUC16 (8/15, 53.0%), MUC5B (7/15, 47.0%), MUC3A (7/15, 47.0%), OBSCN (6/15, 40.0%), SSPO (6/15, 40.0%) and DOCK3 (6/15, 40.0%) were identified by WES.
Notably, MUC family gene mutations have not been reported in G34-DHGs. The human MUC16 gene is located on chromosome 19p13.2 and the MUC17 gene is located on chromosome 7q22.1. Out of the eight cases with MUC16 mutations, 87.5% (7/8) of the cases had missense mutations, and the remaining case had a frame_shift_ins mutation. Moreover, all eight cases with MUC17 mutations had missense mutations.
Somatic SNVs and Indels
A total of 8285 exonic mutations were identified in the 15 G34-DHG patients. Of these mutations, 7187 were missense mutations, 448 were nonsense mutations, 153 were frameshift deletions, 313 were frameshift insertions, 10 were in_frame_del mutations, 1 was an in_frame_ins mutation and 173 were splice site mutations. We removed 1561 silent variants with unknown function. The predominant types of nucleotide substitutions in SNVs in G34-DHGs were C > T/G > A transitions and C > A/G > T transversions.
Oncogenic Pathways Analysis
We identified multiple pathways of somatic mutated genes in G34-DHGs using the oncogenic pathways module of the R package maftools. Crucial signal transduction pathways included the RTK-RAS, NOTCH, WNT, Hippo, PI3K, TP53, MYC and Cell_Cycle pathways (Figure 2B). Exome sequencing revealed that 14 of 15 cases had changes in the receptor tyrosine kinase RTK-RAS pathway, involving 42/84 pathway genes (including PDGFRA, MET, BRAF, ERF, FGFR1 and NF1). KEGG pathway analysis revealed that many of the mutated genes were involved in cancer signal transduction. Mutant genes may promote tumor cell proliferation and escape apoptosis through cascade reaction.
Copy Number Alterations (CNAs)
We conducted somatic CNA analyses in the five SYSUCC cases and seven HEBRY cases. The results identified recurrent gains in chromosomes 3p26.33 (11/12), 4p12 (11/12), 9q21.13 (7/12) and 9q34.11 (4/12). Recurrent losses were identified in chromosomal regions 4p35.1 (10/12), 10q25.1 (10/12), 19p13.43 (9/12), 18q23 (8/12) and 9p21.3 (7/12). Loss of somatic CNAs on chromosome 10q25 affects the largest number of genes (1330 genes), including the MGMT locus. Gains of somatic CNAs on chromosome 3p26 affects 396 genes, including ABCC5. Studies have shown that ABCC5 is associated with chemoresistance of astrocytic tumors .
4. PDGFRA Mutation and G34-DHGs
PDGFRA is an important receptor tyrosine kinase in glial development and a recurrent driver in high-grade gliomas [7-9] PDGFRA mutation and the PDGFRA signaling pathway were reported to play potent oncogenic roles in G34-DHGs . Our sequencing results also showed that most G34-DHGs had PDGFRA mutation (12/15).
To explore the potential pathways and genes related to PDGFRA mutation in G34-DHGs, we analyzed differentially expressed genes (DEGs) using RNA-Seq data from two G34-DHG patients with wild-type PDGFRA and eight G34-DHG patients with mutated PDGFRA. The results identified 150 DEGs (|FC| ≥2.0 and P<0.05), including 95 down-regulated genes and 55 up-regulated genes. The DEGs between the two groups are shown in a heatmap in Figure 3A. We identified the top 10 hub genes ranked by degree, including FOS, CXCL8, CXCR1, IL1B, COL1A1, MMP9, FCGR3B, TNF, CCL4 and CCL3. We used MCODE in Cytoscape to identify gene modules in the PPI network and mapped the interaction network of 52 core genes in module 1 (Figure 3B). The results indicated that the genes were mainly involved with blood microparticles, the phospholipase C-activating G protein-coupled receptor signaling pathway, substrate-specific channel activity, complement and coagulation cascades, the neuroactive ligand-receptor interaction and aldosterone-regulated sodium reabsorption.
We used the DAVID database to analyze the GO and KEGG pathways of the DEGs. KEGG pathway analysis revealed that DEGs were significantly enriched in the extracellular matrix–receptor interaction, cytokine-cytokine receptor interaction, PIK3-AKT signaling pathway and chemokine signaling pathway (Figure 3C). The enriched GO-Biological Process terms included immune and inflammatory response (Figure 3D).
5. MUC16 Mutation and Immune Infiltration Characteristics of G34-DHGs
Previous studies indicated that pHGGs with a high mutation load have an elevated neoantigen load and immune response [11,12]. However, tumors with histone H3 gene mutations are considered as immune cold tumors, which are defined as a lack of CD8 immunoreactivity and lack of tumor-infiltrating lymphocytes . Using the RNA sequencing results, we next analyzed the expressions of immune-related genes in 10 patients with G34-DHGs, including 5 from SYSUCC, 3 from CGGA and 2 of the HERBY cases. The 67 differentially expressed immune-related genes were classified according to CD8+ T cell, T cell (general), B cells, monocyte, tumor-associated macrophage, M1 macrophage, M2 macrophage, neutrophil, natural killer cell, dendritic cell, Th1, Th2, Tfh, Th17, Treg and T cell exhaustion markers (Figure 4). In general, consistent with HERBY Phase II Randomized Trial, we also found G34-DHGs were immune cold tumors, with a few exceptions. For instance, the tumor specimen of patient 1 in our cohort showed significant immune infiltration with substantial amounts of CD4 and CD8 T cells (Figure S2). The OS of this patient reached 75 months, which was markedly longer than that of the nine patients with low immune infiltration (mean survival time: 12 months).
A previously published pan-cancer analysis of 30 solid tumor types showed that patients with MUC16 mutations showed higher tumor mutation burden and neoantigen burden, indicating increased tumor immunogenicity, which can predict immune checkpoint inhibitor treatment response. The study included 397 cases of glioblastoma and 61 of these cases (15.37%) showed MUC16 mutations . Therefore, we wondered whether the MUC16 mutation in G34-DHGs might also be associated with immune infiltration.
We used TIMER 2.0 to analyze the immune cell infiltration of G34-DHGs with MUC16 mutation and G34-DHGs with wild-type MUC16. However, no connection between MUC16 mutation and immune cell infiltration was found; this may be because of the small number of cases and relatively low immune infiltration of G34 glioma. We further analyzed the DEGs between G34-DHGs with MUC16 mutation and G34-DHGs with wild-type MUC16. G Protein-Coupled Receptors (GPCR) signaling relevant proteins MTNR1B, OXTR and PDYN were all low expressed in G34-DHGs with MUC16 mutation (Figure 5).
6. Survival Analysis of Patients with G34-DHGs
We performed survival analyses of the patients from our center (SYSUCC). We found that the OS of patients with H3G34-mutant DHGs was worse than that of patients with IDH-mutant high-grade gliomas, but better than that of patients with H3K27M-mutant DMGs. The mean survival times of patients with IDH-mutant high-grade gliomas, G34-DHGs and H3K27M DMGs were 58.4, 53.8 and 18.4 months, respectively (P<0.001).
We further performed analysis in the overall patient group (patients with G34-DHGs from SYSUCC, CGGA and HERBY Trial). Age (≥18 years old vs. <18 years old) and sex (male vs. female) did not influence patient prognosis. Although no statistical difference was achieved, Kaplan–Meier curve analyses showed some trends according to race, PDGFRA mutation, MUC16 mutation and MUC17 mutation. The median OS for Chinese patients was 18 months compared with 12 months for Caucasian patients (P = 0.105). Patients with PDGFRA mutation tended to show a shorter OS (median OS: 11.5 months vs. 16 months), and MUC16 and MUC17 mutations both seemed favorable for prognosis (median OS: 15 months vs. 12 months and 16 months vs. 11.5 months, respectively). Importantly, although immune infiltration varied in G34 glioma patients, patient 1 in our cohort who harbored MUC16 mutation had obviously high immune infiltration and achieved the longest OS of 75 months.