Profiling the molecular and clinical landscape of glioblastoma utilizing the Oncology Research Information Exchange Network brain cancer database

Abstract Background Glioblastoma exhibits aggressive growth and poor outcomes despite treatment, and its marked variability renders therapeutic design and prognostication challenging. The Oncology Research Information Exchange Network (ORIEN) database contains complementary clinical, genomic, and transcriptomic profiling of 206 glioblastoma patients, providing opportunities to identify novel associations between molecular features and clinical outcomes. Methods Survival analyses were performed using the Logrank test, and clinical features were evaluated using Wilcoxon and chi-squared tests with q-values derived via Benjamini-Hochberg correction. Mutational analyses utilized sample-level enrichments from whole exome sequencing data, and statistical tests were performed using the one-sided Fisher Exact test with Benjamini-Hochberg correction. Transcriptomic analyses utilized a student’s t-test with Benjamini-Hochberg correction. Expression fold changes were processed with Ingenuity Pathway Analysis to determine pathway-level alterations between groups. Results Key findings include an association of MUC17, SYNE1, and TENM1 mutations with prolonged overall survival (OS); decreased OS associated with higher epithelial growth factor receptor (EGFR) mRNA expression, but not with EGFR amplification or mutation; a 14-transcript signature associated with OS > 2 years; and 2 transcripts associated with OS < 1 year. Conclusions Herein, we report the first clinical, genomic, and transcriptomic analysis of ORIEN glioblastoma cases, incorporating sample reclassification under updated 2021 diagnostic criteria. These findings create multiple avenues for further investigation and reinforce the value of multi-institutional consortia such as ORIEN in deepening our knowledge of intractable diseases such as glioblastoma.


Genomic and Transcriptomic Analyses
Sample-level enrichments were compared on the basis of alteration frequencies per gene and resultant log ratios of alterations across groups.One-sided Fisher Exact test was used for analysis of mutation frequencies between groups and q-values were derived via Benjamini-Hochberg correction.Transcriptional data were evaluated based on log ratios of mean expression levels per gene, with statistical analysis performed via student's t-test.CIBERSORT 7 was used to evaluate the presence of tumor-associated macrophages (TAMs) in male versus female samples; mean percentages of TAMs were compared using unpaired t-tests.

EGFRvIII Expression
Following the identification of EGFRvIII cases (Supplementary Material), samples with and without the variant were compared on the metrics of overall survival (OS),

Importance of the Study
This study represents the first analysis of 206 glioblastoma cases documented in the Oncology Research Information Exchange Network brain cancer database.All cases were reclassified under new 2021 World Health Organization diagnostic criteria prior to analysis to evaluate hallmarks of glioblastoma through an updated lens and identify novel trends when histologically diagnosed cases are viewed in combination with molecularly diagnosed glioblastoma.This study identifies 3 genes (MUC17, SYNE1, and TENM1) in which mutations are associated with prolonged overall survival (OS), 14 transcripts (NAJB5, PHTF2, TIPRL, CDC23, PGRMC2, CDKN2A, EXOSC9, MIS18BP1, RFC4, CNOT6, IQGAP2, AP3M1, ZNF521, and EPC1) associated with OS > 2 years, and two transcripts (TMBIM1 and CLSTN1) associated with OS < 1 year.Furthermore, the analysis characterizes the landscape of epithelial growth factor receptor alterations commonly seen in glioblastoma and suggests that increased EGFR mRNA expression, rather than amplification or mutation, correlates more strongly with OS. recorded clinical features, genomic alterations, and transcriptional profiles.Transcriptomic data were analyzed utilizing the omics data processing software Ingenuity Pathway Analysis (IPA), which produced a table of drugs and experimental molecules targeting significantly altered genes.

Survival Analysis
A Logrank test was used to compare survival across subgroups stratified by clinical or molecular features.Cases associated with both a CNS and a non-CNS diagnostic code were excluded for survival analyses and for determining median age at diagnosis, given that non-CNS tumor diagnoses always occurred prior to CNS diagnoses and OS data for those cases therefore reflected survival from earlier, non-CNS primaries.Cases were filtered to only include those with a listed vital status of "deceased" to avoid confounding effects of recently diagnosed, living patients.
For analysis of long-term survivors (LTS) versus shortterm survivors (STS), patients with a non-CNS diagnostic code in addition to their CNS diagnostic code were first excluded to avoid confounding of OS data.One case lacking survival data was additionally excluded.STS cases included those with OS < 12 months and a vital status of "deceased" with cause of death "due to cancer" or "probably due to cancer" (n = 13).The LTS group included patients surviving past 24 months, irrespective of vital status (n = 69).Clinical, genomic, and transcriptomic data for LTS versus STS patients were analyzed as described above.

Glioblastoma Cases Exhibit Frequent Mutations in a Set of 28 Genes
In total, 348 samples from 206 patients qualified as glioblastoma via histologic or molecular characteristics.Twenty-eight genes were mutated at significantly higher rates in tumor samples compared to germline samples (Figure 2).There were no significant mutational differences between molecularly and histologically diagnosed cohorts.

Specific Mutations Correlate With Differences in Overall Mutation Burden and Microsatellite Instability
Of the 28 significant genes, TERT mutations were associated with lower median mutation count (101.5 in mutant samples vs. 242 in wild type; q < 0.05) and lower median mutational burden (3.15 in mutant samples vs. 4.53 in wild type; q < 0.05), while EGFR or PTEN mutations exhibited no such differences.All other 25 genes were associated with increased mutational counts and burdens compared to wild-type counterparts.Mutations in PCLO, FSIP2, DST, MKI67, RYR3, DNAH8, and LRRQ1 were significantly associated with increased microsatellite instability (q < 0.05).

Five Transcripts Exhibit Increased Expression in Younger Patients
No mutations occurred at a significantly higher rate in any age quartile.However, 5 mRNAs were expressed at a significantly higher rate in patients aged 23.27-50.02years: ECD, DDX50, HIF1AN, CHUK, and IDE (q < 0.05).

Male Glioblastoma Patients Overexpress CD68
Six genes were differentially expressed between males and females (q < 0.05).Only one, CD68, was a somatic gene associated with TAMs while the rest (NLGN4Y, ZFX, KDM5C, DDX3X, and KDM6A) were Germline samples Glioblastoma samples

Demetriou et al.: Clinical and molecular landscape of glioblastoma
sex-specific.CIBERSORT analysis revealed no differences in percentage of activated TAMs in male (n = 26 samples) versus female tumors (n = 27 samples; Supplementary Figure 1).

Neither EGFR Mutation Nor Gene Amplification is Associated With OS
Of 321 samples profiled, 59 (18.4%) harbored EGFR mutations (Table 1).EGFR mutations were not independently associated with OS.Ninety-three tumor samples from 92 patients exhibited EGFR amplification, representing 56% of a cohort of 164 patients with EGFR copy number alteration data.Among patients with EGFR amplification, 35 (38%) simultaneously exhibited complete or partial gain of chromosome 7, indicating that most EGFR amplifications occurred independently of chromosomal copy number alteration.Survival analysis was performed on 116 patients profiled for EGFR copy number alteration with a vital status of "deceased" and OS data calculated from a confirmed CNS primary tumor.Patients with EGFR amplification (n = 63) exhibited nonsignificantly increased median OS compared to those without EGFR amplification (n = 53), (18.71 vs. 16.04 months, P = .67).
Of the 93 samples with EGFR amplifications, 37 additionally harbored mutations within the amplified EGFR gene, and 18 harbored amplifications of genes downstream in the EGFR RTK-Ras canonical signaling pathway.To analyze the survival effect of EGFR amplification in isolation from other EGFR alterations, samples with EGFR amplification but no mutations or downstream pathway amplifications (13 samples from 13 patients) were compared with those harboring none of the aforementioned features (36 samples from 31 patients).No significant difference in performance status or median OS was observed (12.52 months OS with EGFR amplification versus 15.55 months without, P = .69).To evaluate the added effect of EGFR pathway downstream amplification on OS, samples with both EGFR amplifications and downstream amplifications (11 samples from 11 patients) were compared with samples harboring EGFR amplification but lacking downstream amplification (34 samples from 34 patients).There were no significant differences in performance status or median OS between groups (23.21 months with downstream amplification vs. 18.71 months without, P = .61).Similarly, transcriptomic analysis comparing cases with both EGFR amplification and downstream pathway amplification to those with EGFR amplification but no downstream amplifications revealed no significant differences in EGFR RTK-Ras pathway mRNA expression.

EGFRvIII Expression is Associated With a Distinct Transcriptomic Profile
Of the 51 patients with RNA sequencing data available for EGFRvIII analysis, 8 samples from 8 patients expressed the variant (15.69% of cases).Although a comparison of samples harboring versus lacking the variant revealed no significant differences in clinical or genomic features, transcriptomic analysis revealed 240 transcripts differentially expressed between groups (q ≤ 0.05; Figure 4).Of the 240 transcripts, 67 exhibited a q-value of ≤0.01 (Supplementary Table 1).
Processing of the 240 transcripts using IPA revealed that EGFRvIII samples exhibited significantly higher expression of transcripts involved in the following pathways (ordered by statistical significance): DNA methylation and transcriptional repression signaling, epithelial adherens junction signaling, coordinated lysosomal expression and regulation signaling, microRNA biogenesis signaling, PPARα/RXRα activation, cAMP-mediated signaling, PTEN signaling, and ribonucleotide reductase signaling.Fiftytwo canonical pathways were expressed at a significantly higher level in samples lacking EGFRvIII expression (Table 2).Transcripts associated with approved or experimental drugs are shown in Supplementary Table 2.

EGFRvIII Expression Does Not Significantly Influence OS
Within the subset of glioblastoma patients with RNA sequencing data available, 40 met criteria for survival analysis based on vital status and OS data, including 7 who expressed the EGFRvIII variant and 33 who did not.Patients expressing the variant exhibited a modest but nonsignificant increase in median survival (21.50 vs. 15.85 months; P = .93).

Discussion
The present study represents the first analysis of 206 glioblastoma cases within the ORIEN brain cancer database, Analysis of the ORIEN glioblastoma cohort identified 28 genes mutated significantly more frequently in tumor samples; some included genes with known roles in glioblastoma including TERT, PTEN, and EGFR. 3 TERT mutations were mostly promoter mutations and were significantly associated with older age and decreased survival, corroborating findings from prior reports. 3Despite PTEN mutations also being implicated in reduced OS, our study did not identify an association between PTEN mutation and shorter survival. 8UC17 (mucin 17) mutations were associated with increased median OS, which represents, to the best of our knowledge, the first report of this association.MUC17 encodes a membrane-bound protein involved in cell structure, and its protein family may play a role in extravasation of metastatic cancer cells. 9Contrary to our findings, a recent multi-database glioblastoma analysis found that MUC17 mutations were associated with poorer prognosis. 10In breast cancer, MUC17 knockdown improved chemosensitivity in vitro, while in lung cancer, decreased MUC17 expression was observed in tyrosine kinase inhibitor (TKI) resistance. 11,12Given that EGFR TKIs have been trialed in glioblastoma with little success, the possibility that MUC17 activity relates to TKI resistance may warrant further exploration.
SYNE1 (synaptic nuclear envelope protein 1) mutations were also associated with prolonged survival.SYNE1 is involved in cell cycle progression and is implicated in numerous malignancies including lung, ovarian, and colon cancers. 13One analysis of The Cancer Genome Atlas glioblastoma samples reported that SYNE1 mutation was associated with higher expression of the oncogene RAF1 and decreased expression of tumor suppressors MTUS1, ZFHX3, and SPINT2. 14The study additionally reported that SYNE1 mutations were associated with mutations in mismatch repair genes MSH6 and MLH1, the dysfunction of which contributes to a mutator phenotype.As in our study, another analysis of The Cancer Genome Atlas glioblastoma cases revealed an association between increased SYNE1 expression and longer survival. 15tations in TENM1 (tenurin transmembrane protein 1) were also associated with prolonged survival.Tenurins represent a subfamily of proteins within the tenascin family involved in CNS development, neurite outgrowth, transcriptional regulation, and interactions between cells and extracellular matrices. 167][18] One study reported that ODZ1 enabled glioblastoma invasion via Myc-dependent upregulation of RhoA, and ODZ1 overexpression in glioblastoma cells in xenografted mice reduced survival. 17Knockdown of ODZ1 decreased glioblastoma cell invasiveness, and analysis of human glioblastoma samples revealed that ODZ1 expression was inversely correlated with survival.Another study investigated the role of glioma-associated macrophages in promoting the invasive phenotype of glioblastoma and found that monocytic cells releasing IL-6 induced ODZ1 expression and contributed to invasiveness via a Stat3-mediated mechanism; Blockade of this signaling pathway decreased ODZ1 expression and glioblastoma invasiveness. 18These findings suggest that further investigation of TENM1 signaling would be valuable to determine the prognostic implications of alterations in this gene in glioblastoma.
Genomic analysis between sexes revealed that male glioblastoma patients exhibited increased CD68 mRNA expression.CD68 is a marker used to identify TAMs in cancer tissue samples, and the presence of CD68-positive cells is correlated with an immunosuppressive tumor microenvironment and worse prognosis. 19CIBERSORT analysis revealed no significant differences in percentages of TAMs in male versus female tumors, suggesting that differences in CD68 expression may reflect glioma rather than immune cell CD68 expression.Human U87 glioblastoma spheroids can express CD68 even when cultured without macrophages, and increased CD68 expression is correlated with higher tumor grade. 20,21Although our data did not show a significant survival difference between males and females, males exhibited slightly decreased median OS.This is consistent with previous reports demonstrating increased OS for female glioblastoma patients. 22One study investigating sex differences in glioblastoma linked treatment response with transcriptomic data; Female patients exhibited better overall response to standard therapies and their responses were influenced by integrin signaling pathway activity, whereas male response to therapy was influenced by cell cycle regulator expression. 23Sex-specific differences in gene expression may thus prove useful for prognostication in glioblastoma treatment.
This study also investigated the influence of EGFR alterations on glioblastoma survival.EGFR mutations, amplifications, and expression of the EGFRvIII variant occur frequently in glioblastoma; previous studies have reported varying results regarding the prognostic value of EGFR alterations. 24Although EGFR has received much attention as a potential therapeutic target, little success has been achieved via targeting EGFR signaling in glioblastoma.Despite EGFR being one of the more frequently mutated genes in our cohort, EGFR mutations were not associated with performance status or OS.Similarly, EGFR amplification was not associated with survival differences, nor was   expression of the EGFRvIII variant.However, comparison of patients with higher versus lower EGFR mRNA expression z-scores revealed significantly shorter overall survival in those with high expression, which aligns with existing literature on EGFR expression in glioblastoma. 25Clinical evaluation of EGFR mRNA expression levels may potentially play a useful role in glioblastoma prognostication.EGFRvIII expression was associated with significantly different expression of 240 transcripts compared to non-EGFRvIII samples.EGFRvIII samples were more likely to exhibit higher expression of genes involved in DNA methylation, transcriptional repression signaling, and several additional pathways.Non-EGFRvIII samples exhibited increased expression of S100A1, a marker of mesenchymal glioblastoma, whereas EGFRvIII samples exhibited higher expression of the classic/proneural subtype marker NOTCH1 and the proneural-associated gene IDH1. 26,27ther notable transcriptional differences in EGFRvIIIcontaining samples include increased HDAC6 expression, reported to promote glioma proliferation; increased expression of the stem cell marker SOX9; and increased expression of tumor suppressor TSC2. 26,28Processing the full set of 240 significant transcripts using IPA revealed numerous drugs associated with the identified transcripts, creating opportunities to hone precision medicine in glioblastoma.
EGFRvIII samples exhibited increased expression of CDK6 and ADAMTS9, which are both implicated in  29 This finding is interesting given that EGFRvIII expression may be related to temozolomide sensitivity. 30To our knowledge, this represents the first study demonstrating a significant association between EGFRvIII expression and upregulation of resistance molecules such as CDK6 and ADAMTS9, which creates an avenue for further investigation of combined influences of EGFRvIII, CDK6 upregulation, and ADAMTS9 upregulation related to temozolomide responsiveness.Given that inhibitors of CDK6 have shown promise in cell line studies and CDK6 inhibition is undergoing investigation in clinical trials, further elucidation of the relationships between EGFRvIII, resistance gene expression, and therapeutic response is recommended. 31nalysis of long-versus short-term glioblastoma survivors revealed 14 transcripts upregulated in the LTS cohort, representing the first report of this transcriptional signature associated with prolonged glioblastoma survival.A handful of these transcripts have been studied in various cancer models with positive prognostic value.Deletion of CDKN2A, a cell cycle regulator and tumor suppressor, is a known prognostic marker of poor survival in glioblastoma. 32It therefore follows that CDKN2A upregulation as observed in our study was associated with longer OS.Similarly, CDC23 is a cell cycle regulator that contributes to breakdown of mitotic proteins, and in glioblastoma models, CDC23 knockdown is associated with increased mitotic activity. 33Higher CDC23 expression was observed in the LTS cohort, consistent with existing literature characterizing the role of its protein product.
Several additional LTS genes are correlated with more favorable disease progression based on studies in other cancers, including DNAJB5 (prostate cancer 34 ), EPC1 (head and neck squamous cell carcinoma 35 ), and AP3M1 (cervical cancer 36 ).Conversely, a handful of genes in the LTS transcriptomic signature have been implicated in poorer prognosis in other cancers; these include EXOSC9 (breast cancer 37 ), CNOT6 (osteosarcoma 38 ), and RFC4 and PGRMC2 (numerous cancers [39][40][41][42] ).MIS18BP1 may be a microsatellite instability target gene in colorectal cancer. 43he exact roles of these genes in glioblastoma are not yet well-elucidated.
For other genes upregulated in the LTS cohort, the literature conflicts as to whether they are more strongly associated with poor or improved prognosis in various cancers.For example, PHTF2 expression in esophageal squamous cell carcinoma is associated with immune infiltration and prolonged survival, but in gastric cancer, PHTF2 contributes to tumorigenesis. 44,45TIPRL appears to play a tumorsuppressive role in gastric cancer, but promotes cancer progression in lung and hepatocellular cancers. 46ZNF521 expression in hepatocellular carcinoma suppresses tumor growth, whereas in medulloblastoma, leukemias, and gastric cancer, it appears to increase tumorigenicity. 47ncreased IQGAP2 expression is implicated in colon cancer, while decreased expression is thought to contribute to gastric cancer and hormone-refractory prostate cancer. 48iven these mixed findings in other cancer types, further investigation into the roles of the aforementioned genes in glioblastoma will be useful to better clarify their prognostic potential.
Finally, our analysis additionally identified 2 transcripts, TMBIM1 (transmembrane BAX inhibitor motif-containing 1) and CLSTN1 (calsyntenin 1), that were more highly expressed in the STS cohort.Upregulation of TMBIM1 is reported to contribute to glioblastoma proliferation and attenuated apoptosis via the p38/MAPK pathway, as well as resistance to temozolomide, and knockdown of TMBIM1 has been demonstrated to prolong survival in animal models. 49The other upregulated gene in our STS cohort, CLSTN1, belongs to the cadherin family and is a cell adhesion molecule and postsynaptic membrane protein highly expressed in lung and ovarian cancer; its role in glioblastoma is not established. 50Further studies will be needed to elucidate the role of TMBIM1 and CLSTN1 upregulation in glioblastoma.
This study benefitted from the ability to compare a sizeable cohort of glioblastoma patients using detailed clinical metrics, genomic data, and transcriptomic profiling available via ORIEN.Further, this study involved a case reclassification process per updated 2021 WHO CNS guidelines, ensuring that conclusions drawn from these data are optimally relevant.Finally, for survival analyses, cases with confounding variables (eg death from other causes or short survival datapoints reflective of recently established diagnoses) were excluded to better determine the influences of particular genomic and transcriptomic alterations on patient outcomes.
One key limitation of this study was a lack of epigenetic data available for analysis.Epigenetic changes may influence the clinical course of glioblastoma, most notably in the case of O 6 -methylguanine-DNA methyltransferase promoter methylation influencing response to temozolomide treatment. 6Additionally, data were collected from patients receiving care at leading cancer centers and therefore may not be representative of the broader population.Patients treated at highly resourced centers benefit from access to leading-edge technologies, multidisciplinary treatment teams, clinical trials, and other factors that contribute to improved outcomes.Finally, validation studies using larger patient cohorts, as well as functional investigations of the molecular patterns identified herein, will serve as important next steps to add context to our present findings.
In conclusion, this study represents the first comprehensive clinical and molecular analysis of glioblastoma cases in the ORIEN brain cancer database.We identify an association of MUC17, SYNE1, and TENM1 mutations with prolonged survival; increased expression of CD68 in male glioblastoma patients; decreased OS with increased EGFR mRNA expression z-score, but not EGFR amplification or mutation; 14 transcripts upregulated in LTS patients; and 2 transcripts (TMBIM1 and CLSTN1) upregulated in STS patients.Further studies of the molecular alterations identified herein may advance the prediction of clinical outcomes and design of targeted therapeutics to enhance OS in glioblastoma.

Figure 2 .
Figure 2. Significant mutations in glioblastoma cohort.Graph depicting genes mutated at a significantly higher frequency in glioblastoma samples in comparison to their germline counterparts, ordered by statistical significance.

Figure 3 .
Figure 3. Relationship between EGFR mRNA expression and overall survival.Kaplan-Meier curve depicting survival differences between patients with increased versus decreased EGFR expression (mRNA expression z-score > 1 vs. < 1).

Figure 4 .
Figure 4. Transcriptional alterations associated with EGFRvIII expression.(A) MA plot depicting the distribution of intensity ratio over average intensity.Highlighted points signify q < 0.05.(B) Heatmap comparing gene expression across EGFRvIII and nonvariant samples; all genes selected for heatmap exhibited q < 0.05.

Table 1 .
EGFR Gene Mutations and Protein Changes: EGFR mutations and associated protein changes identified in our cohort, ordered by frequency of protein changes (VUS: Variant of Unknown Significance) Demetriou et al.: Clinical and molecular landscape of glioblastoma reclassified under updated 2021 WHO diagnostic criteria, and provides numerous novel insights into the ways genomic and transcriptomic features of glioblastoma influence clinical outcomes.