Genomic profiling of Indian gliomas reveals mutually exclusive IDH1 mutations and EGFR amplifications with distinct grade and age association and clinical implications

INTRODUCTION

Gliomas are a heterogeneous group of primary central nervous system (CNS) tumours that arise from glial cells, believed to originate from neural stem or progenitor cells harbouring tumour-initiating genes driving proliferation and invasive growth.^[1] The World Health Organisation (WHO) classifies gliomas based on histopathological and molecular characteristics, grading them from CNS grade 1 to 4, reflecting increasing malignancy.^[2] Low-grade gliomas (LGG) include grade 1 tumours, such as pilocytic astrocytomas, and grade 2 tumour, including diffuse astrocytomas and oligodendrogliomas (often characterised by 1p/19q co-deletion), as well as oligoastrocytomas, which exhibit both astrocytic and oligodendroglial components. High-grade gliomas (HGG, WHO grades 3–4), including glioblastoma multiforme (GBM, WHO grade 4), are the most aggressive, accounting for 57% of gliomas and 48% of primary CNS malignancies, with a high recurrence rate.^[3,4]

The 2021 WHO CNS5 classification incorporates molecular biomarkers for tumour stratification and clinical relevance.^[2,3] Key prognostic factors such as IDH1 mutations, 1p/19q co-deletion, ATRX loss, TP53 mutations, CDKN2A/B homozygous deletions, TERT promoter mutations, EGFR amplification, and MGMT promoter methylation aid in molecular classification and prognosis.^[3] ISNO guidelines adopt the WHO 2016 glioma classification for resource-limited Indian settings using IHC-based molecular surrogates.^[5] Adoption of the WHO 2021 framework could further refine molecular stratification and prognostic assessment. Systemic treatment for gliomas remains poorly defined and primarily based on tumour grade. For LGG, early surgical resection often combined with radiotherapy and chemotherapy (procarbazine, lomustine, and vincristine) prevents malignant progression.^[6] GBM management involves maximal resection, radiotherapy, and temozolomide (Stupp protocol).^[7,8]

Next-generation sequencing (NGS) has expanded treatment options in glioma by identifying novel clinically actionable biomarkers and targeted therapies tailored to individual patients.^[9,10] IDH1-mutant gliomas have shown improved clinical outcomes with inhibitors like vorasidenib (Phase 3 INDIGO trial) and ivosidenib (Phase I).^[11,12] EGFR-targeted therapies, such as depatuxizumab mafodotin and zotiraciclib, a CDK9 inhibitor, also show promise in recurrent or EGFR-mutant gliomas.^[13,14] However, further studies are needed to evaluate their effectiveness as standard of care for gliomas.

This study contributes to our ongoing effort to generate Indian cancer genomic data to address the underrepresentation of Asian populations in global databases. Our previous work provided a comprehensive view of pathogenic variants and potentially actionable alterations in Indian lung adenocarcinomas.^[15] Here, we investigate the landscape of actionable alterations in Indian gliomas using a targeted NGS panel and annotating variants with OncoKB to inform precision oncology strategies for glioma patients.

MATERIAL AND METHODS

RESULTS

Mutation spectrum

Oncogenic or likely oncogenic variants were identified in 35 of 37 cases (94.6%) using a variant allele frequency (VAF) cut-off of 2%, spanning 157 genetic alterations in 24 out of 46 cancer-relevant genes in the ODxET panel [Figure 1a, Supplementary Data]. The frequency distribution of distinct gene alterations ranged from samples with no variants detected (n = 2, 5.4%) to 7 gene variants (n = 3, 8.1%), which included multiple alterations in IDH1, EGFR, RET, and FGFR4 [Figure 1b].

Close

Figure 1: Gene distribution (a) Left: Pie-chart distribution of different types of genetic alterations detected in 37 gliomas. Right: Histogram of gene-wise distribution of different types of genetic alterations in frequently mutated genes (>10% frequency distribution used as cut-off for this figure), (b) Frequency of samples with the number of distinct genes altered in each sample. SNV: Single nucleotide variants, CNV: Copy number variants.

Export to PPT

Commonly mutated genes, along with their clinical parameters, were visualised using OncoPrinter analysis [Figure 2]. The most frequently altered genes included TP53 and IDH1 (both 48.6%), EGFR (40.5%), followed by RET (29.7%), PIK3CA (27.0%), ALK (21.6%), FGFR3 (16.2%), KEAP1 (13.5%), FGFR4, and KIT (both 10.8%).

Close

Figure 2: Mutational landscape of 37 gliomas visualized using OncoPrint analysis in cBioPortal, sorted by diagnosis from Grade I to IV, showing the gene alterations detected, frequencies of individual genes, and other clinical-histopathological features (#-Total no. of cases with respective gene alteration observed with at least one mutation, #Mut-Total no. of mutations per gene across all cases, Freq-Frequency of gene alteration with at least one mutation (n = 37). (Altered in 35 of 37 cases (94.6%) using a VAF cutoff of 2%).

Export to PPT

Most TP53 mutations were loss-of-function missense mutations (n = 27), 5 truncating nonsense mutations, and 1 indel primarily located in exons 5 to 8 of the TP53 gene, with hotspots at R273C/H and R248Q/W. IDH1 mutations, specifically the p.R132H/C variants located at the catalytic site, were detected in 48.6% (n = 18) of cases, predominantly in Grade 2 and Grade 3 gliomas. Two of the that had recurred gliomas (1 case each of Grade 2 astrocytoma and Grade 2 oligodendroglioma) were IDH1-mutated. Of the 37 gliomas, 8 had previously been tested for IDH1 mutation by immunohistochemistry; the IHC results were 100% concordant with NGS [Table 2].^[19]

Table 2: Concordance of IDH1 status across NGS and IHC methods

SBID	Devi S 2018 data (IDH1 status by IHC)	ODxET (IDH1 status by NGS)
SB00010686	IDH1 mut	IDH1 p.R132H
SB00014552	IDH1 wt	IDH1 wt
SB00000295	IDH1 mut	IDH1 p.R132H
SB00008812	IDH1 mut	IDH1 p.R132H
SB00033818	IDH1 wt	IDH1 wt
SB00008528	IDH1 mut	IDH1 p.R132H
SB00030649	IDH1 mut	IDH1 p.R132H
SB00026548	IDH1 wt	IDH1 wt

IDH1: Isocitrate dehydrogenase, NGS: Next-generation sequencing, IHC: Immunohistochemistry.

EGFR amplifications were identified in 14.9% (n = 7) of cases, with copy number gains ranging from 17.4 to 39.3. Notably, three of these cases also harboured the EGFRvIII in-frame deletion (exons 2–7), commonly found in Grade 4 primary glioblastomas. Additional EGFR missense mutations were identified in the tyrosine kinase domain (exons 18–21). RET mutations, particularly G810S, were observed to be predominantly gain-of-function (GOF) missense mutations located within the cytoplasmic kinase domain. PIK3CA activating mutations were observed in 27% of cases, with 12 missense mutations across 10 samples. Most mutations were located in the catalytic subunit of PI3-kinase, with hotspot alterations such as R108H/C in the p85 binding domain. ALK mutations were identified in 21.6% of cases, including 7 putative driver missense mutations and 3 variants of unknown significance.

The KIAA1549-BRAF fusion protein, commonly reported in paediatric brain tumours, particularly pilocytic astrocytomas, was detected in 1 case of Grade 1 pilocytic astrocytoma, resulting in a fusion breakpoint at exon 11 of KIAA1549 and exon 15 of BRAF. Additionally, BRAF GOF missense mutations were identified in 2 GBMs affecting the kinase domain. BRAF V600E was detected in a recurrent GBM, where the patient was initially diagnosed with oligoastrocytoma. Other rare alterations included FGFR4 p.V550M, IDH2 R140W/Q, NTRK1 fusion, and a MET fusion.

Clinical correlations with genetic alterations

EGFR amplifications were significantly associated with older patients (median age 62 years vs 38.5 years for EGFR diploid; p = 0.0024), whereas IDH1 mutations were enriched in significantly younger patients (median age 36 years vs. 61 years for unmutated IDH1, p = 0.0041) [Figure 3a]. Similarly, EGFR amplifications were frequently observed in Grade 4 primary glioblastomas (50%, p = 0.0099) while IDH1 mutations were common in Grades 2-3 (50% in anaplastic oligodendrogliomas to 100% of oligoastrocytomas), and rarely in GBMs (8.3%) and absent in pilocytic astrocytomas (p <0.0001) [Figures 3b and c]. IDH1 mutations were mutually exclusive with EGFR amplifications (p = 0.008) [Figure 3d].

Close

Figure 3: Clinical correlations of EGFR amplification and IDH1 mutation in gliomas. (a) Box plot comparing age at diagnosis between EGFR-amplified vs. EGFR-diploid wild-type tumors (left), and IDH1-mutant vs. IDH1 wild-type tumors (right), (b) Bar chart showing the frequency of EGFR:AMP and IDH1:MUT across glioma histological subtypes, (c) Stacked bar plot depicting the distribution of EGFR amplification and IDH1 mutations across WHO Grades II–IV, (d) Co-occurrence matrix illustrating mutual exclusivity of EGFR amplification and IDH1 mutations. MUT: Mutation, WHO: World Health Organization.

Export to PPT

Our observations were validated using publicly available glioma sequencing datasets in cBioPortal (MSK-IMPACT and FMI panels, n = 923) [Supplementary S1].^[20] In these datasets also, EGFR: AMP gliomas were significantly older age (median 59 years) and predominantly Grade 4 (85.9%), while IDH1:MUT gliomas were younger (median 36 years, p < 10^-10) and enriched in Grade 2 (36.4%) and Grade 3 gliomas (51.0%) [Supplementary S2]. Gene set enrichment analysis (GSEA) revealed distinct molecular characteristics and pathway enrichments that underscore the divergent biology of these two glioma subtypes. EGFR: AMP gliomas had upregulated pathways involved in cell proliferation and aggressive behaviour of tumours (TERT, CDKN2A, TP53), whereas IDH1:MUT gliomas displayed enrichment for genes and pathways related to metabolic reprogramming and cellular differentiation (ATRX, CIC), consistent with a less aggressive clinical course [Supplementary S3, Figure 4]. These results were supported by scatter and volcano plots shown in Supplementary S3.

Close

Figure 4: Gene enrichment patterns in molecular glioma subtypes. (a) Gene set enrichment analysis (GSEA) of MSK-IMPACT and FMI glioma datasets (n = 923) showing gene alterations enriched in EGFR-amplified (EGFR: AMP) and IDH1-mutated (IDH1:MUT) subgroups, (b) Table of significantly enriched genes in each subgroup with corresponding cytoband location, alteration frequency, log₂ ratio, cooccurrence profile, and adjusted p- and q-values, analysed via cBioPortal.

Export to PPT

Actionable alterations

Somatic alterations with potential clinical actionability were analysed using a higher mutation sensitivity threshold of 5% VAF, a commonly applied standard in clinical reporting, and the OncoKB Therapeutic Levels of Evidence V2 framework.

Actionable mutations were identified in 81.1% cases (n = 30) eligible for adjuvant targeted therapy as illustrated in Figure 5a; Level 1 (n = 2, 5.4%), Level 2 (n =1, 2.7%), Level 3A (n = 17, 45.9%), Level 3B, and Level 4 (n = 5 each, 13.5%) as the highest level of actionability. Level 3A alterations were mostly detected in Grade 2 and Grade 3 tumours, primarily associated with IDH1:MUT, while Level 4 alterations were exclusively in Grade 4 tumours, associated with EGFR: AMP [Figure 5b]. Notably, 39.6% of tumours (n = 11) harboured multiple actionable alterations, enabling the selection of combination or sequential lines of targeted therapy. Additionally, biomarkers predictive of resistance or lack of efficacy to FDA-approved therapies, Level R1, were observed in 2 cases.

Close

Figure 5: Clinically actionable alterations. (a) Distribution of OncoKB level of evidence of clinically actionable alterations (≥5% VAF), annotated using OncoKB Levels of Evidence V2: Level 1 (FDA-approved in glioma), Level 2 (strong clinical evidence), Level 3A (compelling clinical evidence), Level 3B (evidence in other tumour types), Level 4 (preclinical evidence), R1 (resistance biomarkers), and R2 (preclinical resistance), (b) Distribution of actionable alterations across tumour grades (I–IV), categorised by the highest OncoKB evidence per case, (c) Frequency of actionable alterations per case with associated targeted therapies.

Export to PPT

IDH1 R132 mutations, Level 3A, predictive of response to ivosidenib, were the most frequent predictive biomarker (n = 18, 48.6%) [Figure 5c]. Actionable EGFR alterations, including amplifications, R108K, and A289V, potentially targetable by the EGFR tyrosine kinase inhibitor lapatinib, were identified in 21.6% of the cases. Other actionable alterations included BRAF V600E, NTRK1, and KIAA1549-BRAF fusions, targetable by tovorafenib. FGFR1/3 mutations targetable by erdafitinib and fexagratinib, and a few mutations that are not yet included in the NCCN guidelines, including PIK3CA (n = 4), RET (n = 2), EGFR G719D (n = 2), ERBB2 (n = 1), and HRAS G13R (n = 1) [Figure 5c]. These findings highlight the clinical utility of NGS-based profiling for targeted therapy options in gliomas, with actionable variants identified in most gliomas across all grades.

DISCUSSION

Our study provides a molecular characterisation of Indian gliomas using an ODxET targeted NGS panel, identifying key genetic alterations and their clinical significance. As one of the few studies in Indian glioma patients, our findings offer novel insights into their molecular heterogeneity and clinical actionability, while also revealing a high degree of overlap in gene variants reported across other ethnicities. Our findings contribute to the growing evidence supporting the prognostic value of glioma genomics and its potential to guide targeted therapies.

Consistent with prior research, TP53 and IDH1 mutations were more common in lower-grade gliomas, mirroring TCGA and other genomic studies.^[21] TP53 mutations, primarily affecting exons 5-8, are early drivers in gliomagenesis, impairing DNA damage response and proliferation.^[2] Similarly, IDH1 mutations, predominantly R132H, were enriched in Grade 2-3 gliomas, associated with better prognosis due to their role in metabolic reprogramming^[22], leading to widespread epigenetic dysregulation, histone modifications, and glioma CpG island methylator phenotype (G-CIMP). This phenotype defines a distinct molecular subclass characterised by an indolent clinical course and better response to therapy.^[22]

In contrast, EGFR amplifications were enriched in high-grade gliomas, particularly GBMs, and associated with aggressive tumour behaviour and poor outcomes.^[23] Previous Indian studies have reported EGFR amplifications in 35–50% of adult GBMs using FISH, consistent with our findings of 50% prevalence of EGFR amplifications detected by NGS.^[24,25] EGFR: AMP frequently co-occurs with EGFRvIII mutation, which drives oncogenic signalling through the PI3K/AKT and RAS/RAF/MEK pathways, promoting tumour proliferation, angiogenesis, and therapy resistance.^[26] Alterations identified in RET, particularly G810S, PIK3CA, FGFR3, and ALK, are implicated in gliomagenesis, progression, and drug resistance.^[26,27] The detection of BRAF V600E mutations in GBMs and KIAA1549-BRAF fusions in pilocytic astrocytomas also aligns with their established roles in recurrent and paediatric gliomas.^[28]

Comparison with publicly available glioma sequencing datasets confirmed common findings, including high frequency of EGFR amplifications and IDH1 mutations, but also revealed differences, such as lower frequency of PTEN mutations in our cohort, likely due to the limitations in the NGS panel design and its detection sensitivity [Table 3].^[20,25] Specifically, the ODxET panel used in our study has limited coverage for PTEN, focusing primarily on SNVs and not CNVs, which are a common mechanism for the loss of PTEN in gliomas, particularly GBMs.

A key finding from our data was the mutual exclusivity of IDH1 mutations and EGFR amplifications, which we corroborated using another dataset.^[20] This exclusivity suggests two distinct molecular trajectories: IDH1-mutant gliomas with epigenetic reprogramming and metabolic alterations exhibit better outcomes as compared to EGFR-amplified GBMs that are highly proliferative and treatment-resistant.^[29] The clinical efficacy of EGFR-targeted therapies remains limited due to tumour heterogeneity, drug resistance mechanisms, and poor blood-brain barrier permeability.^[30] Consistent with Indian cohort data, IDH1 and ATRX alterations are predominantly observed in diffuse and anaplastic astrocytomas, are uncommon in pilocytic astrocytomas and glioblastomas, and provide useful markers for classifying diffuse gliomas into histomolecular subgroups.^[31,32]

Further supporting these two molecular subtypes, IDH1-mutant gliomas frequently harbour ATRX, CIC, and FUBP1 mutations that are strongly associated with chromatin remodelling and the co-deletion of 1p19q.^[33] Conversely, EGFR:AMP GBMs exhibit enrichment of TERT promoter mutations and CDKN2A/B deletions, both associated with poorer outcomes.^[34]

Importantly, we identified actionable mutations in 81.1% of gliomas as per OncoKB. with lower-grade gliomas harbouring the highest proportion of such alterations.^[35] IDH1 R132 mutations, especially R132C (seen in 77% of LGG samples and reclassified as a level 3A biomarker), predict sensitivity to ivosidenib, an oral brain-penetrant IDH1/2 inhibitor. Additionally, vorasidenib has shown promising results in significantly prolonging progression-free survival and time to next intervention in patients with Grade II IDH-mutant glioma.^[11,12] Similarly, EGFR alterations suggest therapeutic avenues, though EGFR inhibitors such as lapatinib that cross the BBB have had limited success in GBMs, likely due to drug resistance mechanisms.^[36] Rare but actionable fusions in NTRK1 and MET, and FGFR1/3 mutations highlight opportunities for targeted kinase inhibitors such as larotrectinib, crizotinib, erdafitinib, and fexagratinib, respectively.^[37]

In this context, our previous work on patient tumour-derived cells (PDCs), including some of these NGS-profiled gliomas, provides a valuable model for phenotypic or targeted drug testing.^[38] Testing in 2D and 3D spheroid cultures, to assess drug penetration, resistance, and immune modulation, singly or in combination with drugs targeting different signalling pathways. Our PDC panel spans Grade 1 DIG tumours to Grade 4 GBMs and different histotypes.^[38] Such models are essential for translating genomic findings into actionable therapies and for identifying effective treatments in the context of glioma heterogeneity.

Future studies should integrate molecular profiling with clinical outcomes to refine glioma classification and optimise biomarker-driven trials. In India, ISNO guidelines currently use the WHO 2016 classification with predominantly IHC-based surrogates in resource-limited settings, while recent work demonstrates that the WHO 2021 molecular criteria can be applied using a combination of IHC and targeted molecular assays, making it feasible to align local practice with global standards.^[5,39] Combination regimens that simultaneously target multiple pathways, along with immunotherapies such as checkpoint inhibitors and adaptive trial designs, are promising strategies, particularly in aggressive subtypes such as EGFR-amplified GBMs.^[40]

In summary, our findings reinforce the molecular heterogeneity of gliomas and highlight the clinical relevance of genomic profiling in personalised therapies. Continued efforts in genomic characterisation, targeted therapy development, and biomarker-driven clinical trials hold tremendous promise for improving outcomes for glioma patients.

TAKE HOME MESSAGE

Our findings reinforce the molecular heterogeneity of gliomas and highlight the clinical relevance of genomic profiling in personalized therapies. Continued efforts in genomic characterization, targeted therapy development, and biomarker-driven clinical trials hold tremendous promise for improving outcomes for glioma patients.