Novel EGFR mutations in diffuse midline gliomas using cost-effective strategies: A report of 2 cases

Abstract

Background

Diffuse midline gliomas (DMGs) are malignant tumors predominantly affecting children, often leading to poor outcomes. The 2021 World Health Organization classification identifies 3 subtypes of DMGs, all characterized by the loss of H3K27 trimethylation. Here, we report 2 cases of DMG with Epidermal Growth Factor Receptor (EGFR) mutations within exon 20, contributing to the understanding of the molecular complexity of these pediatric brain tumors.

Methods

An economical immunohistochemical panel was designed to aid in the diagnosis of most DMGs in resource-constrained regions. Sanger sequencing was employed to identify rare EGFR mutations in exon 20 of 2 cases.

Results

Molecular analyses of 2 cases of DMG revealed novel EGFR mutations within exon 20. These mutations were identified using cost-effective diagnostic approaches. The presence of EGFR mutations expands the molecular landscape of DMGs and highlights the genetic heterogeneity within this tumor entity.

Conclusions

These findings underscore the molecular heterogeneity of DMGs and the significance of identifying novel mutations, such as EGFR mutations in exon 20. Further research into the molecular mechanisms underlying DMGs is warranted to advance therapeutic strategies and improve outcomes for pediatric patients.

Pediatric high-grade glioma (pHGG) is the primary cause of cancer-related mortality in children, resulting in a median overall survival of less than a year.¹ The first identified mutations in histone genes associated with cancer, referred to as “on histones,” were found in pHGG. These mutations encompass H3 p.K28M (K27M) (H3.3 and H3.1/H3.2) and H3.3 p.G35R/V (G34R/V) delineating distinct subgroups.^2–6 In 2021, the World Health Organization (WHO) revised the classification of these tumors, differentiating between H3-K27M-altered gliomas and H3-G34-mutant gliomas. These are now denoted as diffuse midline glioma (DMG), H3 K27-altered, and diffuse hemispheric glioma, H3 G34-mutant, respectively. Both fall under the category of pHGGs WHO grade 4.⁷ Recent studies have unveiled midline glial tumors that lack the H3 p.K28M (K27M) mutation but exhibit the loss of H3K27me3, coupled with either EZHIP overexpression or Epidermal Growth Factor Receptor (EGFR) mutation. This expansion broadens the spectrum of DMGs beyond the H3 p.K28M (K27M) mutation.^8,9 Consequently, the term “DMG, H3K27-altered” has been introduced, comprising 3 molecular subtypes: DMG-H3K27M mutant, DMG-EZHIP overexpression, and DMG-EGFR mutant.⁷ H3K27M-mutant tumors remain the most common subtype, with the other 2 subclasses being rare exceptions.¹⁰ DMG-EGFR mutant tumors exhibit EGFR oncogene mutations without concurrent amplification of the EGFR gene, infrequent histone H3 mutations, and distinct genome-wide DNA methylation profiles compared to other glioma subtypes.⁸ Additionally, all cases show loss of H3K27me3 and inactivating mutations of the TP53 gene are also prevalent in the majority of cases.^8,9 Unfortunately, these cases are not amenable to surgical resection and have uniformly poor outcomes despite radiation therapy and traditional cytotoxic chemotherapy.^8,11

While most cases of DMG with EGFR mutations are typically identified through high-throughput techniques like next-generation sequencing (NGS) and methylation profiling, it is worth noting that in resource-constrained settings, the presence or absence of the EGFR mutation, particularly the frequently observed exon 20 insertion/duplication, can still be determined using more cost-effective methods like Sanger sequencing. In this study, we present an integrated diagnostic strategy for DMGs in resource-limited environments. Additionally, we report 2 novel EGFR mutations located in the loop following the C-helix domain of exon 20.

Clinical Summary

Case 1

A 5-year-old female presented with altered sensorium along with clinical signs indicative of increased intracranial pressure. She had hydrocephalus on non-contrast computed tomography (NCCT), for which a ventriculoperitoneal (VP) shunt was placed. Post-VP shunt, an MRI scan revealed a bithalamic lesion, more on the left side. The lesion consisted of 2 components. The predominant ventral component appeared isodense on NCCT (Figure 1A), iso-hypointense on T1-weighted image (WI) (Figure 1A), and iso-hyperintense on T2-WIs (Figure 1C and 1E) and fluid-attenuated inversion recovery (FLAIR) image (Figure 1D). The smaller dorsal component appeared hypodense on NCCT, hyperintense on T1-WI (Figure 1A), and hyperintense on T2-WIs (Figure 1C and 1E) and FLAIR Image (Figure 1D). On the coronal T2-WI, the mass was observed to extend to the upper midbrain. After administering gadolinium, axial T1-WI (Figure 1F) showed no significant enhancement.

Figure 1:

Case 1—The predominant ventral component appeared isodense on non-contrast computed tomography (NCCT) (A), iso-hypointense on T1-weighted image (WI) (A), and iso-hyperintense on T2-WIs (C, E) and fluid-attenuated inversion recovery (FLAIR) image (D). The smaller dorsal component appeared hypodense on NCCT, hyperintense on T1-weighted image (A), and hyperintense on T2-WIs (C, E) and FLAIR image (D). On the coronal T2-WI, the mass was observed to extend to the upper midbrain. After administering gadolinium, axial T1-WI (F) showed no significant enhancement. Case 2—The lesion was predominantly isodense on NCCT (arrow in G), iso-hypodense on T1-WI (arrow in H), and iso-hyperintense on T2IWI (arrow in I). The focal central area of hypodensity in NCCT (dotted arrow in G), hypointensity in T1-WI (dotted arrow in H), and hyperintensity in T2-WI (dotted arrow in I). The lesion also exhibited restricted diffusion on DWI (arrow in J) except in the central portion (dotted arrow in J). Interestingly, the susceptibility-weighted (SW) images did not display any foci of hypointensities, suggesting the absence of hemorrhages or calcification (K). Post-contrast T1-WI (L) images didn’t reveal any significant enhancement.

A 2-stage decompressive surgical procedure was performed. During the surgery, the tumor was found to be soft in consistency, amenable to suction with a CUSA device, and exhibited a high degree of vascularity.

Case 2

An 11-year-old male presented with persistent, bifrontal headaches and recurrent episodes of projectile vomiting. There was no history of limb weakness, seizures, or loss of consciousness. Imaging revealed a large, poorly defined solid mass, measuring 4.5 cm  × 4.2 cm  × 5.0 cm in size. The tumor was primarily located in the left thalamus and caused a midline shift of 9 mm and mild hydrocephalus. The lesion was predominantly isodense on NCCT (arrow in Figure 1G), iso-hypodense on T1-WI (arrow in Figure 1H), and iso-hyperintense on T2IWI (arrow in Figure 1I). The focal central area of hypodensity in NCCT (dotted arrow in Figure 1G), hypointensity in T1-WI (dotted arrow in Figure 1H), and hyperintensity in T2-WI (dotted arrow in Figure 1I). The lesion also exhibited restricted diffusion on diffusion-weighted images (DWI) (arrow in J) except in the central portion (dotted arrow in Figure 1J). Interestingly, the susceptibility-weighted image did not display any foci of hypointensities, suggesting the absence of hemorrhages or calcification (Figure 1K). Post-contrast T1-WI (Figure 1L) images didn’t reveal any significant enhancement. The patient underwent a left parietal craniotomy, and a subtotal decompression procedure was performed on the intraventricular tumor. During the surgery, the tumor was found to be soft, grayish-white in color, amenable to suction using a CUSA device, and moderately vascular.

Methods

Sample Collection and Processing

Tumor tissue was fixed in 10% neutral buffered formalin, routinely processed, and paraffin embedded. Hematoxylin and eosin-stained slides were reviewed by 2 neuropathologists (V.S., M.C.S.), and a consensus diagnosis was arrived at. Immunostaining was performed on 5-micron-thick formalin-fixed, paraffin-embedded tumor sections using an automated immunostainer (Benchmark XT, Ventana, Tucson, AZ) and standard protocols including pretreatment using cell conditioning 1 buffer (Ventana) for 52 minutes and standard Ventana signal amplification. Diaminobenzidine was used as a chromogen.

Immunohistochemistry (IHC) was performed using the following monoclonal antibodies: glial fibrillary acidic protein (GFAP) (1:800; Dako, Denmark), OLIG2 (IgG1 goat polyclonal antibody H11 clone; R&D systems, 1:150), Isocitrate dehydrogenase 1 (IDH1) (IDH-1R32H, Dianova, mouse monoclonal, 1:50), alpha thalassemia/mental retardation syndrome X-linked (ATRX), Sigma Aldrich, St. Louis, MO; dilution 1:400), P53 (Santa Cruz Biotechnology, Inc., CA; 1:200), MIB-1 ((DAKO, Denmark, 1:200), a mutation in 27th amino acid of Histone H3 (H3K27M, Millipore, Billerica, MA; dilution 1: 1000)), H3G34R (RM-240, RevMab Biosciences; 1:200), and H3G34V (RM-307, RevMab Biosciences; 1:500), EZHIP (Sigma Aldrich, 1:100).

Fluorescence In Situ Hybridization

Dual-probe Fluorescence In Situ Hybridization (FISH) assay was performed on paraffin-embedded sections for assessment of EGFR amplification and homozygous CDKN2A deletion. Signals were scored in at least 200 nonoverlapping, intact nuclei. Sections from non-neoplastic cortical tissue obtained from epilepsy surgery specimens were used as a control for each probe pair. Locus-specific probes paired with centromere probes for chromosome 7 (CEP7, Vysis, Downers Grove, IL) were used for EGFR assay. EGFR amplification was considered when >10% of tumor cells showed innumerable tight clusters of signals of the locus probe. For the CDKN2A assay, locus-specific probes paired with centromere probes for chromosome 9 (Vysis LSI CDKN2A Spectrum orange/CEP9 Spectrum green probes, Downers Grove, IL) were used. Homozygous deletion of the CDKN2A was considered when loss of both red signals in each nucleus was seen in a minimum of 30% of nuclei. The cutoff value was determined by calculating the mean +3 SD of deletion seen in non-neoplastic brain tissue.

Mutation Analysis

DNA from the formalin-fixed paraffin-embedded blocks was extracted using the Maxwell DNA FFPE kit (Promega, Hilden, Germany) according to the manufacturer’s protocol. Sanger sequencing was performed to detect mutation in exon 20 of EGFR. The primer sequences used for EGFR exon 20 sequencing were forward-5’-CCCTCCTTCTGGCCACC-3’ and reverse 5’- CTGCGTGATGAGCTGCAC-3’. PCR amplification was conducted in a total of 10 µL reaction mixture in 1 µL of DNA with approximately the concentration of 50 ng, 5 µL of master mix (Promega), and 0.5 µL of each forward and reverse primer per reaction. The initial denaturation was performed at 95°C for 5 minutes, followed by 40 cycles of amplification including denaturation at 95°C for 30 seconds, annealing at 57°C for 35 seconds, and extension at 72°C for 10 minutes. The bidirectional sequencing was performed using ABI 3500XL Genetic Analyzer (Applied Biosystems).

Results

Histopathology

In both cases, histopathological examination unveiled a glial tumor characterized by moderate-to-high cellularity and moderate cytological pleomorphism with brisk mitotic activity. Case 1 displayed endothelial cell proliferation and foci of necrosis, whereas Case 2 did not exhibit endothelial cell proliferation or necrotic areas. These findings suggest a shared underlying astrocytic nature of the tumors with distinct variations in histological characteristics. (Supplementary Figure 1A and B)

Immunohistochemistry

In both cases, IHC revealed similar patterns. Tumor cells in both cases were positive for GFAP and Olig2, with immunonegativity for IDHR132H, H3K27M, H3G34R, H3G34V, EZHIP, and H3G34V. H3K27Me3 immuno-expression was lost while ATRX protein expression was retained. Case 2 was immunopositive for P53. An MMR proficient profile was observed in both cases. MIB-1 labeling index was 40% in Case 1 and 20%–25% in Case 2 (Supplementary Figure 1C–I).

Fluorescent In Situ Hybridization (FISH)

FISH analysis for both cases did not reveal EGFR amplification or CDKN2A homozygous deletion.

Sanger Sequencing

Case 1: Sanger sequencing detected a small in-frame duplication of 9 base pairs, c.2311_2319dupAACCCCCAC (p.Asn771_His773dupAsn_His) in the loop following C-helix (amino acids Ala767 to Cys775) of exon 20 (Supplementary Figure 2A).

Case 2: Sanger sequencing showed a small in-frame duplication of 9 base pairs, c.2300_2308dupCCAGCGTGG (p.Ala767_Val769dupAla_Val) in the loop following C-helix (amino acids Ala767 to Cys775) of exon 20 (Supplementary Figure 2B).

The final diagnosis of both cases was rendered as DMG, EGFR mutant CNS WHO Grade 4.

Discussion

Primary thalamic gliomas represent 1%–1.5% of all intracranial tumors, with about 25% occurring in children aged 15 or younger^.12 There is no gender predilection, with more than 50% of cases occurring in pediatric patients aged between 3 months and 16 years, while they are relatively rare in adults, particularly those older than 60 years.⁸ Despite exhibiting a low-grade morphology in certain instances, these tumors are associated with unfavorable prognoses and typically symmetrically affect both thalamic nuclei, with occasional extensions into the temporal lobe or brain stem.^12,13

Symptoms of thalamic tumors include increased intracranial pressure, movement disorders, and mass effect, presenting with hemiparesis, sensory issues, dysmetria, unsteady gait, and nystagmus, while adult cases may exhibit personality changes.¹³ Complete surgical excision is challenging due to extensive thalamic nuclei involvement and only partial tumor debulking is feasible.¹⁴ CT scans show bilateral isodense lesions without enhancement, while MRI reveals hypointensity to isointensity on T1-weighted and hyperintensity on T2-WIs, without contrast enhancement.^14,15 Thalamic tumors may exhibit varying contrast enhancement. DWIs show no restriction. MR spectroscopy indicates elevated creatinine and choline peaks (creatinine > choline), and reduced NAA peak, suggesting heightened cellular metabolism distinct from low-grade gliomas. Neuronal loss is marked by a decreased NAA peak.^16,17

In 2020, Mondal and colleagues performed comprehensive genomic and epigenomic analysis on a cohort of 13 children with bithalamic and identified that these tumors harbor frequent mutations in the EGFR oncogene, only rare histone H3 mutation (in contrast to their unilateral counterparts), and a distinct genome-wide DNA methylation profile. These cases exhibited distinctive EGFR mutations, notably exon 20 insertions/duplications, and shared commonalities with specific DNA methylation classes that overlapped with the pedRTK2 subgroup.⁸ Sievers et al. provided further evidence of pediatric-type malignant glioma that is preferentially located in the thalamic region and shows frequent alteration of EGFR and DNA methylation profiling revealed EGFR gene amplification in 16/58 cases (27%). By genome sequencing, they also reported a missense mutation in EGFR in 15/30 cases and an in-frame insertion within exon 20 of EGFR in 5 tumors, including 3 of the tumors initially described by Mondal et al.⁹ Seminal papers have reported that EGFR exon 20 mutation activates signaling pathways like Ras/MAPK, PI3K/AKT, and STAT, influencing crucial cellular functions like survival, proliferation, and migration.¹⁸ Consequently, these mutations are frequently observed in different cancer types, including lung adenocarcinomas, fibrous hamartomas, and midline gliomas.^19–21

IHC offers reliable, cost-effective, and timely diagnostics with high sensitivity and specificity, reported to be higher than that of sequencing, but the latter technique is useful for identifying rarer mutations. If IHC is inconclusive in certain tumors or suspected, direct sequencing using PCR or NGS is recommended.²² However, in resource-limited settings, we propose employing a primary IHC panel that includes Olig2, H3K27me3, and H3K27M, along with MIB-LI, as a comprehensive approach for diagnosing midline gliomas. In H3K27M-negative cases with H3K27me3 loss, EZHIP immunostaining should be performed. Further, in thalamic gliomas EGFR mutation by Sanger sequencing should be performed if the above panel is noncontributary. Additionally, IHC markers such as GFAP, IDHR132H, ATRX, P53, H3G34R, H3G34V, and FISH analysis for EGFR amplification and CDKN2A deletion can be further utilized in conjunction with clinical data in the diagnostically challenging cases. The choice between methods hinges on lab expertise, available equipment, and the preferences of clinicians and pathologists.

We have identified 2 novel mutations of EGFR in exon 20 in the loop following the C-helix domain (amino acids Ala767 to Cys775).²³ Case 1 shows a pathogenic variant consisting of a duplication of nine nucleotides in exon 20 (c.2311_2319dupAACCCCCAC p.Asn771_His773dup). This variant is reported in other tumors like fibrous hamartoma of infancy and nonsmall-cell lung cancer.^20,24 Case 2 harbored another unique pathogenic variant consisting of a duplication of nine nucleotides in the exon 20 (c.2300_2308dupCCAGCGTGG p.Ala767_Val769dup). To the best of our knowledge, this particular variant is not reported in databases, but in the same nucleotide position, different insertions have been observed in DMG and lung tumors.^21,25,26 Although this diagnostic distinction has limited implications for patient outcomes at present, treatment with targeted kinase inhibitors (EGFR inhibitors) in bithalamic gliomas manifested encouraging results.^9,23

In conclusion, our center has made a significant contribution by uncovering 2 novel EGFR exon 20 mutations situated in the loop following the C-helix domain within DMG cases. While high-throughput platforms have been the primary method for identifying such cases in the past, our study suggests that employing a comprehensive combination of IHC markers, FISH assays, and Sanger sequencing results can significantly enhance diagnostic accuracy, particularly in resource-constrained settings. Given the limited surgical options available, it highlights the importance of exploring alternative treatments, such as targeted EGFR inhibitor therapy.

Conflict of interest statement

The authors declare no conflict of interest.

Funding

None declared.