Alterations in EGFR and PDGFRA are associated with the localization of contrast-enhancing lesions in glioblastoma

Abstract Background Glioblastoma (GBM) is a malignant brain tumor, with radiological and genetic heterogeneity. We examined the association between radiological characteristics and driver gene alterations. Methods We analyzed the driver genes of 124 patients with IDH wild-type GBM with contrast enhancement using magnetic resonance imaging. We used a next-generation sequencing panel to identify mutations in driver genes and matched them with radiological information. Contrast-enhancing lesion localization of GBMs was classified into 4 groups based on their relationship with the subventricular zone (SVZ) and cortex (Ctx). Results The cohort included 69 men (55.6%) and 55 women (44.4%) with a mean age of 66.4 ± 13.3 years. EGFR and PDGFRA alterations were detected in 28.2% and 22.6% of the patients, respectively. Contrast-enhancing lesion touching both the SVZ and Ctx was excluded because it was difficult to determine whether it originated from the SVZ or Ctx. Contrast-enhancing lesions touching the SVZ but not the Ctx had significantly worse overall survival than non-SVZ lesions (441 days vs. 897 days, P = .002). GBM touching only the Ctx had a better prognosis (901 days vs. 473 days, P < .001) than non-Ctx lesions and was associated with EGFR alteration (39.4% vs. 13.2%, P = .015). Multiple contrast lesions were predominant in PDGFRA alteration and RB1-wild type (P = .036 and P = .031, respectively). Conclusions EGFR alteration was associated with cortical lesions. And PDGFRA alteration correlated with multiple lesions. Our results suggest that clarifying the association between driver genes and tumor localization may be useful in clinical practice, including prognosis prediction.


Materials and Methods
This retrospective study was approved by the Institutional Review Board of Kagoshima University Hospital (approval no.180104) and complied with the guiding principles of the Declaration of Helsinki.Furthermore, written informed consent was obtained from each patient.We collected 124 samples diagnosed with supratentorial, IDH wild type, and WHO grade 4 GBM from the central nervous system tumor tissue bank at Kagoshima University Hospital.The tumors were fixed with phosphate-buffered 10% formalin for 24 h and processed for paraffin embedding, followed by sectioning for hematoxylin and eosin staining.All tissues were histologically evaluated by board-certified pathologists to ensure an estimated tumor cell content of ≥30%.

Next-Generation Sequencing
NGS was performed using an amplicon-based gliomatailored 48-gene (version 2) or 50-gene panel (version 3) (QIAGEN) with 2244 primers for the regions of interest (161179 bp) and an average exon coverage of 99.95%. 20,21he amplicon sequences were aligned to the human reference genome GRCh37 (hg19) in the target region of the sequence.Data were analyzed using the QIAGEN Web Portal service (https://www.qiagen.com/ja-us)and Mitsubishi Space Software (Amagasakihttps://www.mesw.co.jp).

Methylation-Specific Polymerase Chain Reaction
Bisulfite conversion of the extracted genomic DNA was performed using the EpiTect Bisulfite Kit (Qiagen).The converted genomic DNA was amplified for the target O6-methylguanine-DNA methyltransferase promoter (MGMTp) region with primers specific for the methylated or unmethylated template using KOD One® PCR Master Mix (Toyobo).Two pairs of primers specific for methylated or unmethylated MGMTp regions were used for methylationspecific polymerase as previously reported. 22The amplification was performed with an initial denaturation at 98°C for 1 min, followed by 40 cycles of 98°C for 10 s and 64°C for 5 s.MCE-202 MultiNA (Shimadzu) was used for the analysis.

Radiological Analysis
Based on previous reports, CELs of GBM were classified into 4 groups: group I, the CEL extends from the atrium SVZ to the pia; group II, the CEL touches the SVZ but does not involve the cortex (Ctx); group III, the CEL invades the

Importance of the Study
Glioblastoma (GBM) is a malignant brain tumor with genetic heterogeneity.Receptor tyrosine kinases have been implicated in the pathogenesis of glioblastoma, and EGFR amplification has been added to the diagnostic criteria of the 2021 World Health Organization classification of tumors of the central nervous system.GBMs have radiologically heterogeneous features; however, their relationship with molecular genetics remains unclear.We analyzed IDH wild-type glioblastomas and contrast-enhancing lesions shown using magnetic resonance imaging and found that cortical lesions showed significantly higher EGFR alteration and PTEN loss, and periventricular lesions showed higher CDK4 amplification.In addition, multiple lesions were more common in PDGFRA alteration and RB1 wild type.The alterations in EGFR and PDGFRA expression showed contrasting characteristics regarding Ki-67 expression and resection rates.Our report is the first radiogenomic study to show that receptor tyrosine kinases are potentially involved in the localization, multiplicity, and proliferative capacity of GBMs.

Makino et al.: EGFR and PDGFRA alterations and localization in GBM
Ctx and reaches the pia but does not touch the SVZ; and group IV, the CEL does not touch either the Ctx or SVZ (Supplementary Figure 1). 6,8We first classified the entire cohort of 124 patients into the 4 categories.
GBM can be classified into 3 categories based on their multiplicity: solitary, multifocal, and multicentric.Multifocal GBM is defined as a CEL that appears remote to some extent but is microscopically connected and contained within a common hyperintense area on a T2-weighted image (T2WI).Multicentric GBM refers to CELs in different lobes or bilateral brains that do not have potential contacts, such that they have a common highintensity area on a T2WI. 15,17,23All image evaluations were performed by 2 board-certified neurosurgeons (N.H. and H.Y.) using a PACS system (Synapse PACS 4.1.3;Fuji Film Medical Systems).The interval between imaging and tissue collection was within 1 month in all cases.

Data Analysis
Statistical analyses were performed using the GraphPad Prism9 software (MDF Co., Ltd.).The relationship among location, multiplicity, and genetic information was analyzed using χ 2 tests and t-test.The log-rank test was used to analyze overall survival (OS).Differences with P < .05were considered statistically significant.

Differences in Clinical Background and Genetic Status Depending on Location Within the SVZ and Cortex
Comparisons between the SVZ (CEL Groups I and II) and non-SVZ groups (CEL Groups III and IV) are presented in Table 2. Patients with GBM in contact with the SVZ were older than those with GBM not in contact with the SVZ (P = .070,unpaired t-test).This group showed poor preoperative KPS (P < .001)and had difficulties achieving more than subtotal resection (P < .001).Moreover, there was no significant difference in the driver gene status between the SVZ and non-SVZ groups.
CEL group I touched the SVZ and Ctx, making it difficult to determine whether it originated from the SVZ or Ctx.Therefore, we excluded group I and re-examined the tumor localization and genetic characteristics.We compared group II, which was thought to have originated from the SVZ, with the non-SVZ groups (groups III and IV).Similarly, we compared group III, which was thought to have originated from the Ctx, with the non-Ctx groups (groups II and IV).Group II had fewer EGFR alterations (12.1% vs. 36.8%,P = 0.027) than the non-SVZ groups (groups III and IV; Table 3).CDK4 amplification was more common in group II (P = .051)than in the non-SVZ groups; however, the difference was not statistically significant (Table 3).Group III showed significantly higher EGFR alteration (39.4% vs. 13.2%,P = .015)and PTEN loss (66.7% vs. 34.2%,P = .009)than the non-Ctx groups (Table 3).PDGFRA, TP53, RB1, CDK4/6, CDKN2A/B, and TERT promoters showed no significant differences in localization.

Clinical Features and Genetic Alternations in Multiple GBM
We identified solitary, multifocal, and multicentric GBM lesions in 87 (70%), 28 (23%), and 9 (7%) patients, respectively.First, we compared solitary and multiple lesions; the latter were a combination of multifocal and multicenter lesions.Age, sex, preoperative KPS score, and Ki-67 labeling index were not significantly different between patients with solitary and multiple GBMs.PDGFRA alteration (P = .036)and RB1 wild type (P = .031)were more common in multiple GBMs than in solitary GBM.TERTp mutation was more common in multiple GBMs than in solitary GBM; however, the difference was not statistically significant (P = .07;Table 4).

OS Differences in Imaging Findings
The SVZ group (CEL groups I and II) had a significantly worse OS than the non-SVZ group (median survival: 567 vs. 897 days, P = .037;Figure 2A).Among the 4 groups, group II had a poor prognosis than the other groups (groups I, III, and IV) (441 vs. 768 days, P = .004),and group III had a Abbreviations: CEL, contrast-enhanced lesion; OS, overall survival.(A) Patients with PDGFRA alteration were significantly older than those with intact PDGFRA (P = .020,unpaired t-test).EGFR alteration did not show a significant difference in age compared with that of EGFR intact.(B) Ki-67 labeling index was classified into higher (≥30) and lower (<30) groups.EGFR and PDGFRA alterations showed lower and higher Ki-67 labeling index scores, respectively (P < .0001,P = .021:Fisher's exact test).(C) We classified the patients into 2 groups: subtotal or greater resection and partial or less resection.PDGFRA alteration had a significantly poorer resection rate (P = .019).EGFR alteration had better removal rates; however, the difference was not statistically significant (P = .075).

Makino et al.: EGFR and PDGFRA alterations and localization in GBM
more favorable prognosis than the other groups (groups I, II, and IV) (901 vs. 566 days, P = .008;Supplementary Figure 2A and B].Subsequently, group I was excluded considering its heterogeneous characteristics.CEL group II had a significantly worse OS (441 vs. 897 days, P = .002)and group III had a better prognosis (901 vs. 473 days, P < .001)than the other groups (Figures 2B and C).A comparison between solitary and multiple GBMs showed no significant difference in OS (688 vs. 488 days, P = .239;Figure 2D).

Location of CELs and Gene Alterations
GBM is extremely heterogeneous and highly treatment resistant.It has the worst prognosis among central nervous system tumors.The possibility that tumor localization affects the prognosis of patients with GBM has been discussed.Previous reports have indicated that periventricular GBM and ventricular entry during surgery are associated with distant metastasis and intrathecal dissemination, 24,25 and there was a rebuttal to this conclusion. 26The association of SVZ-GBM with high invasiveness and multiple lesions was first reported in 2007, and SVZ-GBM has since been associated with poor progression-free survival and OS. 6,9,279][30] The inhibition of PI3K/Akt signaling downstream of RTK has also been reported to inhibit glioma growth in the SVZ, 31 suggesting that RTK is an important therapeutic target for GBM tumorigenesis.A previous study showed no difference in EGFR amplification or PTEN loss between SVZ and non-SVZ GBMs in 26 newly diagnosed IDH wild-type GBM cases. 32This The frequency of EGFR amplification in Japanese patients has been reported to be lower than that in other patients, which is consistent with the results of the present study. 33,34The present study included a larger cohort than   The subventricular zone (SVZ) group had a significantly worse OS than the non-SVZ group.(B) Group II had a significantly worse OS than the other groups (groups III and IV).(C) Group III had a better prognosis than the other groups (groups II and IV).(D) A comparison between solitary and multiple GBMs showed no significant difference in OS.

Makino et al.: EGFR and PDGFRA alterations and localization in GBM
did previous reports and showed an association between EGFR and PTEN status and cortical lesions.Using our NGS panel, we showed that several driver genes were associated with the localization and multiplicity of CELs in GBM.The present study differs from previous reports because in addition to comparing the SVZ and non-SVZ groups, CEL group I was excluded, and cases clearly localized to the cortex or SVZ were compared.In addition to EGFR amplification, mutations were also included to demonstrate their relationship with tumor localization.Our study suggests for the first time that PDGFRA alteration is associated with multiple lesions.There was no significant relationship between PDGFRA alteration and CEL localization; however, the relationship between PDGFRA alteration and poor removal and survival rates was clinically meaningful.
EGFR alterations are more common in cortical lesions with a good removal rate, suggesting that this is one of the reasons why EGFR alteration has a good prognosis.Alternatively, the Ki-67 level was clearly lower in the EGFR alteration group, and there may be a significant difference in surgical outcomes and tumor aggressiveness.In addition to EGFR alteration, PTEN loss was significantly associated with cortical lesions.In a mouse model of SVZ NSCs transfected with mutated p53, PTEN, and EGFR, mutant cells migrated from the SVZ to remote areas of the brain and formed high-grade gliomas. 35To the best of our knowledge, this is the first radiogenomic study to show consistent results in humans.EGFR has been shown to the ability of cells to migrate remotely from the SVZ, which explains the high incidence of EGFR alteration in group III.
The origin of GBM is assumed to be the NSCs/neural progenitor cells (NPCs) and oligodendrocyte progenitor cells (OPCs) of the SVZ, [36][37][38][39] and GBM is presumed to originate from cells years before diagnosis. 40NSCs in the SVZ have a genetically distinct feature from other localizations that may influence the development and progression of GBM. 11,12SCs have the ability to differentiate into neurons and glial cells and are abundant in fetal tissues.Owing to their high proliferative potential, NSCs have been implicated in the development of GBM.Neftel et al. focused on the different states of tumor cells in GBM and found that 4 cell types exist: NPC-like, OPC-like, astrocyte-like, and mesenchymal-like cell morphologies. 41The marker of NPC-like GBM is CDK4 amplification.In our study, CDK4 amplification was more common in periventricular lesions (group II), which may reflect the characteristics of NPC-like GBM and tumor localization.
RTK status affects the proliferative potential. 42,43We showed that EGFR and PDGFRA alterations had contrasting characteristics regarding Ki-67 scores.Ki-67 is a known predictor of proliferative potential in various cancer types; however, its expression is heterogeneous, and the details of its association with specific genetic mutations in GBM remain unclear.PDGFRA alteration is a marker of OPC-like cells, which are less differentiated and may indicate high proliferative potential.

Multiple CELs and Gene Alterations
Multiple GBM lesions are associated with secondary malignancies, a family history of cancer, germline p53 mutations, and a poor prognosis. 17,44,45There have been several reports on the genetic differences between solitary GBM and multiple GBM; however, these are limited to several major genes, and only a few studies have used comprehensive gene panels.A previous report showed that the frequency of EGFR mutation and EGFR/PTEN concurrent mutation is higher in multiple GBM and that multicentric GBM has fewer CDK4 mutations and more CDKN2A/B mutations than multifocal GBM. 19Furthermore, a previous study showed that multifocal GBM arises from PTEN loss or TERTp mutations and develops through the RTK/PI3K, p53, and RB1 pathways. 18Other reports have shown that TERTp mutation is strongly correlated with multiple GBM lesions. 34In our study, TERTp mutation was more common in the multiple GBM group than in the solitary GBM group, although the difference was not statistically significant.In addition, our results showed that multiple GBMs were strongly correlated with RB1 wild type, but we have not found similar reports for other cancers.A recent report indicated that RB1 alterations affect the prognosis of GBM 46 ; however, there are no studies on multiple lesions, and future research is needed.
The present study has some limitations.First, it was a retrospective study that is susceptible to selection bias.Second, because GBM is a histologically and genetically heterogeneous tumor, the analysis of some specimens removed from multifocal and multicentric GBM may not reflect the entire tumor.Additionally, in this study, bulk sequencing was performed on the portion of the tumor mass that was sampled, and the cell density was secured.Detailed data on where the collected tumor corresponds to on MRI have not been recorded.In the future, it is desirable to use neuronavigation to record the location of samples taken intraoperatively, create a database, and conduct prospective studies.Third, our cohort had a small number of patients in certain groups, such as CEL group IV and multicentric gliomas.Therefore, the clinical characteristics of each group and detailed subgroup analyses were unavailable.More cases are needed for future studies.Finally, this study was conducted only in Japanese patients.Recent reports suggest that the frequency of EGFR amplification in Asian cohorts is lower than that in Western cohorts, 47,48 and genetic differences between races are controversial.
In conclusion, EGFR alteration was associated with cortical lesions, favorable removal rates, and low Ki-67 scores.In contrast, PDGFRA alteration correlated with high proliferative potential and multiple lesions.These results might have influenced the differences in prognosis among alterations in the RTKs.We considered EGFR and PDGFRA status as important markers of different localizations and clinical courses in GBM.Our results suggest that clarifying the association between driver genes and tumor localization may be useful in clinical practice, including prognosis prediction.

Figure 1 .
Figure 1.Clinical features of GBM depending on EGFR and PDGFRA status.(A) Patients with PDGFRA alteration were significantly older than those with intact PDGFRA (P = .020,unpaired t-test).EGFR alteration did not show a significant difference in age compared with that of EGFR intact.(B) Ki-67 labeling index was classified into higher (≥30) and lower (<30) groups.EGFR and PDGFRA alterations showed lower and higher Ki-67 labeling index scores, respectively (P < .0001,P = .021:Fisher's exact test).(C) We classified the patients into 2 groups: subtotal or greater resection and partial or less resection.PDGFRA alteration had a significantly poorer resection rate (P = .019).EGFR alteration had better removal rates; however, the difference was not statistically significant (P = .075).

Figure 2 .
Figure 2. Differences in overall survival (OS) in imaging findings.(A)The subventricular zone (SVZ) group had a significantly worse OS than the non-SVZ group.(B) Group II had a significantly worse OS than the other groups (groups III and IV).(C) Group III had a better prognosis than the other groups (groups II and IV).(D) A comparison between solitary and multiple GBMs showed no significant difference in OS.