The role of RB1 alteration and 4q12 amplification in IDH-WT glioblastoma

Abstract Background Recent studies have identified that glioblastoma IDH-wildtype (GBM IDH-WT) might be comprised of molecular subgroups with distinct prognoses. Therefore, we investigated the correlation between genetic alterations and survival in 282 GBM IDH-WT patients, to identify subgroups with distinct outcomes. Methods We reviewed characteristics of GBM IDH-WT (2009–2019) patients analyzed by next-generation sequencing interrogating 205 genes and 26 rearrangements. Progression-free survival (PFS) and overall survival (OS) were evaluated with the log-rank test and Cox regression models. We validated our results utilizing data from cBioPortal (MSK-IMPACT dataset). Results Multivariable analysis of GBM IDH-WT revealed that treatment with chemoradiation and RB1-mutant status correlated with improved PFS (hazard ratio [HR] 0.25, P < .001 and HR 0.47, P = .002) and OS (HR 0.24, P < .001 and HR 0.49, P = .016). In addition, younger age (<55 years) was associated with improved OS. Karnofsky performance status less than 80 (HR 1.44, P = .024) and KDR amplification (HR 2.51, P = .008) were predictors of worse OS. KDR-amplified patients harbored coexisting PDGFRA and KIT amplification (P < .001) and TP53 mutations (P = .04). RB1-mutant patients had less frequent CDKN2A/B and EGFR alterations (P < .001). Conversely, RB1-mutant patients had more frequent TP53 (P < .001) and SETD2 (P = .006) mutations. Analysis of the MSK-IMPACT dataset (n = 551) validated the association between RB1 mutations and improved PFS (11.0 vs 8.7 months, P = .009) and OS (34.7 vs 21.7 months, P = .016). Conclusions RB1-mutant GBM IDH-WT is a molecular subgroup with improved PFS and OS. Meanwhile, 4q12 amplification (KDR/PDGFRA/KIT) denoted patients with worse OS. Identifying subgroups of GBM IDH-WT with distinct survival is important for optimal clinical trial design, incorporation of targeted therapies, and personalized neuro-oncological care.

Glioblastoma (GBM) is the most common and aggressive central nervous system (CNS) primary malignancy. 1 Despite aggressive treatment with maximal safe resection and chemoradiotherapy, 2 and multimodal therapy upon recurrence, GBM is associated with a dismal prognosis. 1 Over the past decade, molecular characterization of gliomas has revealed the heterogeneous nature of this group of tumors. [3][4][5][6][7] These findings led to a reclassification of infiltrating gliomas, in which both tumor histology and genetic alterations are considered. 8 Infiltrating gliomas are classified by the presence or absence of mutations in the isocitrate dehydrogenase (IDH) 1 or 2 genes, as IDH-wildtype (WT) or IDH-mutant, with different demographic, clinical, and prognostic characteristics. 8 Recent studies have identified that GBM IDH-WT consists of different molecular subgroups, which might have a distinct prognosis. 6,7,9,10 However, more studies are needed to understand molecular subgroups of GBM IDH-WT as survival differences between these have not been thoroughly investigated.
Therefore, we examined the correlation between genetic alterations and survival in a cohort of GBM IDH-WT patients, to identify potential subgroups with different behavior, who may potentially benefit from targeted therapies. Our findings were validated using a large publicly available dataset from Memorial Sloan Kettering Cancer Center (MSK-IMPACT). 11

Patients and Tumor Samples
We performed a retrospective review of GBMs in an institutional glioma registry of patients diagnosed between 2009 and 2019. The inclusion criteria for this study were confirmed diagnosis of GBM IDH-WT according to the cIMPACT-NOW Update 3 12 and availability of sequencing data from a comprehensive next-generation sequencing (NGS) assay. A flow diagram selection of the study population is depicted in Supplementary Figure S1.
Data for this study were collected from Memorial Hermann Hospital's electronic medical records. Data were managed with REDCap electronic data capture tools hosted at the University of Texas Health Science Center at Houston (UTHealth). 13 These included age, sex, race, Karnofsky performance status (KPS), diagnosis, radiographic extent of resection, treatment strategy, and survival. Tumors were classified by a board-certified neuropathologist following the 2016 WHO Classification of Tumors of the CNS 8 and cIMPACT-NOW updates. Radiographic extent of resection was classified as gross total resection (GTR), near-total resection (NTR), or subtotal resection as previously described. 14 Recurrence and therapeutic strategy were determined by individual revision of cases by a multidisciplinary tumor board as previously described. 15

Ethical Statement
This study was approved by the Institutional Review Board (ID: HSC-MS-17-0917) of UTHealth and Memorial Hermann Hospital, Houston, TX.

Targeted Sequencing
Tumor samples were analyzed for genomic alterations by a targeted NGS panel interrogating 205 genes and 26 gene rearrangements including telomerase reverse transcriptase promoter (TERTp) mutations (FoundationOne; Foundation Medicine, Inc.). The FoundationOne assay was performed in a Clinical Laboratory Improvement Amendments certified laboratory, as previously described. 16,17 TERTp status was not available for 61 patients.

Validation Cohorts
To validate our findings, we utilized the dataset from the MSK-IMPACT available at cBioPortal (https://www. cbioportal.org/), accessed on August 2020. 11,18,19 This dataset provided clinical and genetic information including IDH, MGMT, and TERTp status. Additionally, we used a GBM cohort with IDH1 p.R132H status from a publicly available study evaluating the 4q12 amplicon in GBM. 20 Tumors classified as GBM IDH-WT according to the cIMPACT-NOW Update 3 criteria were analyzed. 12

Importance of the Study
Glioblastoma IDH-wildtype (GBM IDH-WT) comprises different molecular subgroups; however, the prognostic significance of these subgroups has not been defined. We demonstrated in a large molecularly characterized GBM IDH-WT cohort 2 genetically distinct subgroups with different prognoses. Our findings were validated with large external GBM IDH-WT datasets. GBM IDH-WT with RB1 mutations is a molecular subgroup with improved PFS and OS and decreased frequency of CDKN2A/B loss and EGFR alterations. On the other hand, 4q12 (KDR/ PDGFRA/KIT) amplified patients have worse survival. Our data revealed the importance of genetic profiling of GBM IDH-WT to identify subgroups with distinct survival. This is crucial for optimal clinical trial design, targeted therapies, and personalized neuro-oncological care.

Statistical Analyses
Descriptive analyses were performed by the Mann-Whitney U test or Fisher's exact test for continuous or categorical variables, respectively. The endpoints of the study were overall survival (OS) and progression-free survival (PFS). OS was calculated as the time in months from diagnosis to death or the last available follow-up. PFS was calculated as the time in months from diagnosis to progression of the disease. The univariable two-sided log-rank test was used to examine statistical significance in survival, while the Kaplan-Meier method was employed to plot visual survival curves. Univariable and multivariable Cox proportional hazard regression models were utilized to calculate the hazard ratio (HR) estimates with a 95% confidence interval (CI) adjusted for possible confounders. Multivariable Cox proportional hazard regression model analysis for PFS and OS was adjusted for the variables with a P value of .05 or less in univariable analysis, as these might affect survival. Demographic, clinical, and genetic characteristics were evaluated by the genes of interest to identify differences in such traits. The genes of interest were defined as the genes that correlate with survival. P value of .05 or less was considered statistically significant and was two-sided. The differences in genetic characteristics were adjusted for multiple comparisons utilizing the Benjamini-Hochberg FDR correction procedure (q value). Statistical analyses were performed using EZR (1.40) 17,21 and Prism v.8.4.3 (GraphPad). The oncoplots were created using cBioPortal OncoPrinter Tool. 18

Outcomes in GBM IDH-WT
Univariable analysis of PFS showed that patients who received chemoradiotherapy with TMZ according to the Stupp protocol (HR 0.25, P < .001) had a significantly lower risk of death. This finding was further confirmed by multivariable analysis (HR 0.25, P < .001; Table 1). Additionally, it was observed that GBM IDH-WT patients harboring RB1 mutations (n = 28) had an improved PFS compared to RB1-WT (n = 254) patients (11.9 vs 7.5 months, P = .0001, log-rank test; Supplementary Figure S2A). This finding was confirmed by multivariable analysis (HR 0.47, P = .002; Table 1). The PFS of GBM IDH-WT patients harboring a KDR amplification was not significantly different compared to KDR-WT patients (5.8 vs 8.3 months, Supplementary Figure S2B).
Multivariable analysis of OS demonstrated that patients younger than 55 years of age (HR 0.63, P = 0.010), who received chemoradiotherapy with TMZ (HR 0.24, P < .001), received salvage Bevacizumab after progression (HR 0.54, P < .001), and harbored an RB1 mutation (HR 0.49, P = .016) had a significantly lower risk of death. Conversely, patients who had a KPS less than 80 (HR 1.44, P = .024) and who harbored a KDR amplification (HR 2.51, P = .008) had a significantly higher risk of death (Table 1). Overall, our institutional cohort shows that GBM IDH-WT can be divided into 3 subgroups with different PFS and OS, by KDR and RB1 status ( Figure 2).
We also evaluated the KDR amplification in the MSK-IMPACT GBM IDH-WT dataset (n = 551). Twenty-six (5%) patients harbored KDR amplification in this dataset. However, KDR was not significantly associated with PFS

Neuro-Oncology Advances
vs 28%), it was not significant after multiple comparison adjustments. Additionally, we did not observe differences in MGMT status between KDR-amplified and -WT patients (P = .2180).
Furthermore, we evaluated the relationship of KDR amplification and survival in GBM IDH-WT, utilizing the data of a published study that evaluated KDR amplification through fluorescence in situ hybridization (FISH). 20 In this dataset, 19 of 142 (13%) GBM IDH-WT tumors had KDR amplification. Consistent with our study, KDR-amplified patients also had worse survival compared to KDR-WT patients

Discussion
In the present study, we sought to identify molecular subgroups of GBM IDH-WT with prognostic significance.

Outcomes in GBM IDH-WT
Several studies have demonstrated that the survival of patients with GBM is affected by age, KPS, extent of resection, and chemoradiotherapy with TMZ. 1,2,22,23 The relationship of age in GBM IDH-WT has been confirmed by studies with comprehensive genetic characterization. 6,10 Our results further confirmed that younger age and chemoradiotherapy with TMZ improved OS in GBM IDH-WT. In line with prior studies, patients with low preoperative functional status (KPS <80) had worse survival. 23 In addition, we observed that patients treated with salvage bevacizumab had an improved outcome. However, this finding deserves further study, as patients treated with salvage bevacizumab would inevitably have a lead-time bias to improve survival compared to patients who died without a documented recurrence and who were unable to benefit from this therapy. Despite the lack of survival benefit of bevacizumab in randomized clinical trials, 24,25 these trials were performed prior to the molecular classification of gliomas. Additional studies are needed to evaluate salvage bevacizumab therapy in molecular subgroups of GBM IDH-WT. A recent report has shown that EGFR-amplified and classical GBM subgroups are associated with poor response to bevacizumab in recurrent GBM. 26

GBM IDH-WT, RB1 Mutant
The RB1 gene was the first tumor-suppressor gene to be molecularly defined. 27 Dysregulations of the RB pathway signaling are a critical event in gliomagenesis. 28 In our study, we identified that RB1 loss-of-function mutations are present in 10% of our institutional cohort and 14.6% of MSK-IMPACT GBM IDH-WT. Interestingly, we observed that RB1-mutant GBM IDH-WT patients less frequently harbored EGFR alterations and CDKN2A/B loss. In addition, RB1-mutant cases had a higher frequency of mutations in TP53 in both the UTHealth and MSK-IMPACT cohorts. Our findings demonstrated that RB1-mutant tumors are a subgroup of GBM IDH-WT with a distinct prognosis. TP53 and RB1 mutation co-occurrence has been previously reported in several cancers. 7,27 Also, RB1 exclusivity with EGFR amplification has been demonstrated in GBM xenograft models and glioma patients. 7,28 Importantly, we identified for the first time that RB1-mutant GBM IDH-WT had improved PFS and OS than RB1-WT patients after multivariable adjustment ( Figure 2 and Table  1). Remarkably, our findings were validated by the MSK-IMPACT cohort. Importantly, these results were not explained by differences in MGMT status in the MSK-IMPACT cohort, as RB1-mutant patients' MGMT status did not differ from RB1-WT patients. Loss of the RB1 gene coupled with a loss in homologous recombination DNA repair pathway genes in other cancers, particularly high-grade ovarian carcinoma, has been associated with increased CD8 + tumorinfiltrating lymphocytes (TILs), PFS, and OS. 29 In GBM, an increase in TILs has been associated with RB1 mutations. 30 Additionally, it has been suggested that RB1 mutations provoke replication stress in tumor cells, leading to DNA damage and activation of the innate immune system. This has been hypothesized to enhance immune checkpoint blockade in ovarian carcinoma. 31 The DNA damage and increased TILs in RB1-mutant patients are plausible explanations for the increased survival in this subtype of GBM IDH-WT (Figure 4). These findings should provoke further studies to identify if RB1-mutant GBM IDH-WT might respond more favorably to immune checkpoint inhibitors, a previously failed therapy in this dismal disease. 32 GBM IDH-WT, KDR Amplified KDR (VEGFR2) is a vascular endothelial growth factor (VEGF) receptor located on the chromosomal 4q11-12 locus, along with KIT and PDGFRA. 33,34 KDR activation by binding of VEGF ligands leads to activation of several downstream oncogenic signaling pathways such as PI3K/ AKT, focal adhesion kinase, and mitogen-activated kinase, all resulting in increased cell survival, migration, and angiogenesis. 35 KDR alterations are frequently observed in various cancers, including GBM. 33 KDR is usually expressed within the tumor endothelium and is the primary VEGF signal transducer, which results in increased cell survival, proliferation, and angiogenesis 20 ( Figure 5). The critical role of VEGF in tumor angiogenesis has been demonstrated in multiple studies. This led to the development of bevacizumab, a monoclonal antibody against the VEGF-A ligand that binds to KDR. 36 KDR status and its relationship with outcomes have been investigated in GBM. 20,[37][38][39] However, these studies were performed in a heterogeneous group of patients (IDH-WT and IDHmutant), prior to the TMZ era, or in relatively small cohorts (Supplementary Table S2). Moreover, recent studies have demonstrated that KDR activation through sustained VEGF-C promotes GBM maintenance and growth even under bevacizumab therapy, meaning that activation of KDR through VEGF-C is an escape mechanism of GBM to overcome bevacizumab therapy. 40 The proliferative effects of KDR activation in GBM occur with the binding of both VEGF-A and VEGF-C ligands, though it should be noted  that the effects of these 2 VEGF ligands have been demonstrated to be non-overlapping ( Figure 5). In our study, we identified that KDR amplification, which would cause an activation of the KDR signaling pathway, correlated to worse outcomes. These findings were further validated with the results of a published GBM IDH-WT study in patients evaluated for KDR amplification using FISH. 20 In contrast to the UTHealth cohort and the published study by Burford et al., 20 both demonstrating a statistically significant association with survival, the MSK-IMPACT cohort demonstrated a trend toward shorter survival (16.6 vs 22.8 months) in KDR-amplified cases which was not statistically significant.
In addition, autocrine VEGF-C/KDR signaling has been shown to regulate cell viability, cell cycle, and in vivo tumor growth in GBM, while autocrine VEGF-A/KDR signaling plays a similarly critical role in the proliferation and self-renewal of GBM stem-like cells. 40 Figure 5. Proposed mechanism for the proliferative effects of KDR/VEGFR2 amplification in GBM IDH-WT based on prior studies. As seen on the left, binding of VEGF-A and VEGF-C ligands results in increased proliferation of GBM cells. On the right, it is hypothesized that KDR amplification would increase the number of available receptors, thus magnifying these effects. In addition, with KDR amplification, there may also be an additional elevation of certain protein kinase levels, such as that of c-Met and p38, which serve to increase the invasiveness of the tumor.

Neuro-Oncology Advances
Along with this, KDR amplification has been previously shown in non-small cell lung carcinoma to be associated with VEGF-induced elevated expression of KDR, p38, and mTOR pathway components, promoting a more invasive phenotype. 41 Specifically, it was noted that there was phosphorylation of p38, elevated levels of HIF1α, and increased c-Met activation, correlating with increased angiogenesis and tumorigenesis. Considering these effects together, it would seem to reasonably explain the decreased survival observed in our cohort ( Figure 5). Multi-institutional large GBM IDH-WT cohorts are required to further validate the deleterious effect of KDR amplification and help identify targeted therapies for this molecular subgroup of GBM IDH-WT. More importantly, survival differences between these molecular subgroups should be considered for clinical trial enrollment to avoid unintended bias.

Limitations
The limitations of our study include its retrospective design and potential selection bias, as not all GBM IDH-WT patients in our institution underwent NGS. MGMT promoter status for most of the UTHealth cohort was unavailable. However, we did not identify differences in MGMT status by RB1 or KDR mutational status in the MSK-IMPACT cohort. UTHealth and MSK-IMPACT datasets represent different geographic, socioeconomic, and ethnic populations. Moreover, there might be variations in practice patterns between institutions. Additionally, this study did not assess for KDR protein expression levels or TILs. However, we hope that the hypotheses generated by the study results motivate neuro-oncology researchers to identify the mechanisms underlying the survival differences between GBM IDH-WT subgroups. Despite these limitations, our study confirms the association of age, KPS, and chemoradiotherapy (TMZ) with survival in GBM IDH-WT patients. Moreover, we identified molecular subgroups of GBM IDH-WT with prognostic significance in 2 large independent cohorts.

Conclusions
The current study demonstrates that GBM IDH-WT, RB1mutant represents a different molecular subgroup of GBM with improved PFS and OS. Additionally, we identified that KDR-amplified patients had worse survival, which might be related to increased angiogenesis and potential resistance to bevacizumab. Further studies are needed to determine the best treatment strategy for various GBM IDH-WT subgroups, allowing the incorporation of targeted therapies and personalized neurooncological care.

Supplementary Material
Supplementary material is available at Neuro-Oncology Advances online.