Venous Thromboembolism in Patients with Glioblastoma: Molecular Mechanisms and Clinical Implications

Patients with glioblastoma are among the cancer patients with the highest risk of developing venous thromboembolism (VTE). Long-term thromboprophylaxis is not generally prescribed because of the increased susceptibility of glioblastoma patients to intracranial hemorrhage. This review provides an overview of the current clinical standard for glioblastoma patients, as well as the molecular and genetic background which underlies the high incidence of VTE. The two main procoagulant proteins involved in glioblastoma-related VTE, podoplanin and tissue factor, are described, in addition to the genetic aberrations that can be linked to a hypercoagulable state in glioblastoma. Furthermore, possible novel biomarkers and future treatment strategies are discussed, along with the potential of sequencing approaches toward personalized risk prediction for VTE. A glioblastoma-speci ﬁ c VTE risk strati ﬁ cation model


Introduction
Glioblastoma, isocitrate dehydrogenase (IDH)-wild-type, is the most frequent and the most aggressive type of primary brain cancer in adults, accounting for 48.6% of all malignant tumors within the central nervous system. 1 The overall annual incidence is 3.23 per 100,000 persons, which further increases with age and is higher in males compared to females. 1,2The median expected survival time is 14.6 months despite extensive treatment, 3 with less than 5% of glioblastoma patients showing a survival rate of 5 years or more. 4,5ioblastoma is characterized by rapid proliferation, increased angiogenesis, hypoxia, and necrosis. 6Vascular pathology, reflected by aberrant microvasculature and vascular leakage, induces a procoagulant state, which results in a high number of local (micro)thrombi within the tumorhistologically observed in 90% of all glioblastoma samples. 7hese in turn propagate tumoral hypoxia and necrosis, which together may contribute to systemic hypercoagulability. 6ndeed, glioblastoma patients are at a high risk of developing venous thromboembolism (VTE), with an incidence of up to 10 to 30% per year. 8,9This is one of the highest incidences among all cancer types. 10The exact VTE risk may increase even further depending on patient-related, tumor-related, and therapy-related risk factors, such as age, tumor genetics, and surgery. 11,12he complex pathogenesis of glioblastoma is also illustrated by a high degree of inter-and intratumoral heterogeneity. 13umorigenesis is accompanied by a plethora of different mutations instead of a single-driver oncogene.This has great implications for glioblastoma prognosis and therapy, as treatment efficiency highly depends on tumoral gene expression.Moreover, the risk of developing VTE also relies on genetic signature and subsequent procoagulant gene and protein expression, the so-called coagulome. 14Tissue factor (TF), the primary initiator of the coagulation cascade, and podoplanin, involved in platelet aggregation, are often upregulated in glioblastoma and assumed to significantly increase the risk of VTE.Prescription of thromboprophylaxis needs to be carefully considered due to the increased susceptibility of glioblastoma patients to intracranial hemorrhage.Altogether, these insights warrant a personalized benefit-risk evaluation in order to offer appropriate treatment for all glioblastoma patients.This review provides an overview of the current clinical standard as well as the genetic and molecular background of glioblastoma-related VTE.Furthermore, potential novel biomarkers and future treatment strategies will be discussed in order to explore the prospects of personalized medicine for glioblastoma patients with an increased risk of VTE.

Clinical Overview Glioblastoma Disease Course and Treatment
Patients with glioblastoma are often diagnosed from age 40 or higher. 15Diagnosis requires histological and molecular characterization of tumor tissue, which also guides therapeutic management.Standard of care consists of surgery, pursuing maximum safe resection. 16Surgical resection alone extends survival with 6 months approximately. 17If resection is not feasible, a tumor biopsy is performed for diagnostic purposes. 18n addition, glioblastoma patients are treated with a 6-week course of radiotherapy in combination with concomitant and adjuvant chemotherapy (temozolomide), also known as the Stupp protocol. 19Elderly patients >70 years old have a worse prognosis and generally show lower tolerability of tumor-targeted treatment.For those patients, hypofractionated radiotherapy (3 weeks) with concomitant and adjuvant temozolomide and single modality treatment (monotherapy with either radiotherapy or temozolomide) are reasonable options. 20Surgery combined with concomitant chemoradiotherapy increases overall survival to 14.6 months after glioblastoma diagnosis. 19However, the response to temozolomide largely depends on promoter methylation of the gene MGMT, encoding O6-methylguanine-DNA methyltransferase (MGMT), since this DNA repair enzyme allows for reversal of temozolomide-induced DNA damage. 21MGMT promoter methylation is observed in approximately 30 to 60% of all glioblastoma patients. 22reatment of glioblastoma recurrence is less well defined, but includes re-resection if possible.This improves survival, especially in case of a subtotal primary resection. 23Additional treatment options are secondcourse chemotherapy, mostly lomustine, re-irradiation, and bevacizumab, a monoclonal antibody that inhibits angiogenesis by targeting the angiogenic protein vascular endothelial growth factor (VEGF).For patients with poor performance status, best supportive care is usually the most appropriate option. 24Supportive care throughout the disease trajectory may consist of the glucocorticoid dexamethasone for tumor-related oedema, although this is associated with many side effects and may be related to poor treatment outcome. 25,26For patients who develop epilepsy, antiseizure medication to reduce the risk of new seizures is standard of care. 27

Venous Thromboembolism in Patients with Glioblastoma
The incidence of VTE in patients with glioblastoma is generally acknowledged within a range of 10 to 30% throughout the disease trajectory, 28,29 although reports vary from 7.5 to 39% depending on VTE definition, VTE detection method, and the use of thromboprophylaxis. 8,9,30,31Most VTE events are observed within the postoperative period, but the risk remains higher over the course of the disease compared to other malignancies, with an incidence of 1.5 to 2.0% per month of survival. 32A large retrospective study with malignant glioma patients (n ¼ 9,489) reported a 30% increased 2-year mortality rate in patients who developed VTE as compared to non-VTE patients. 33][36] The development of glioblastoma-related VTE depends on both general risk factors such as age, history of VTE, and comorbidity, and glioblastoma-specific risk factors such as peri-operative immobility, tumor recurrence, and subtotal resection. 28Additional therapy-related factors comprise chemotherapy, bevacizumab, and dexamethasone.It is shown that chemotherapy results in a 3.4-fold increased VTE risk in cancer patients, 37 which may increase even further depending on the exact agent and protocol used. 38Limited data are available for temozolomide and lomustine, the most commonly used chemotherapeutic agents in glioblastoma.Nevertheless, Yust-Katz et al demonstrated that the majority of VTE events developed after the start of adjuvant chemotherapy with temozolomide in a cohort of glioblastoma patients. 8In line, our research group recently demonstrated increased TF-mediated procoagulant activity following treatment with temozolomide or lomustine in three well-established glioblastoma cell lines in vitro. 39Furthermore, treatment with bevacizumab may increase the risk of pulmonary embolism in glioblastoma patients, as a trend toward significance was observed in bevacizumab-treated glioblastoma patients compared to glioblastoma patients who did not receive bevacizumab (p ¼ 0.07). 40The risk of VTE may also be increased by highdose glucocorticoid therapy such as dexamethasone, which directly affects the vascular endothelium. 41

Intracranial Hemorrhage in Patients with Glioblastoma
In addition to VTE, tumor-related intracranial hemorrhage is also frequently observed in glioblastoma patients with reported incidences ranging from 2 to 12%, 9,42 either occurring as first manifestation or throughout the disease trajectory. 43,44s was already described in 1982, the presence of a primary brain tumor by itself may cause spontaneous (nontraumatic) intracranial hemorrhage. 45In glioblastoma, this is expected to be induced by increased expression of VEGF, which is involved in neovascularization and thereby contributes to vascular malformation and permeability. 46,47The increased risk of major bleeding events as imposed by the tumor hampers the prescription of long-term thromboprophylaxis despite the significant risk of VTE in glioblastoma patients.

Guidelines for Thromboprophylaxis
Primary thromboprophylaxis is recommended for ambulatory cancer patients on systemic anticancer therapy with a high risk of VTE, as assessed by VTE risk models. 48,49The currently recommended VTE risk assessment score for chemotherapytreated cancer patients with solid tumors (Khorana score) includes primary site of cancer (categorized into "very high risk" and "high risk"), body mass index, platelet count, leukocyte count, and hemoglobin level. 50However, patients with brain tumors were underrepresented in the cohorts used to develop the Khorana score.As a result, brain cancer was insufficiently powered and not included as high-risk cancer type in this model.A subsequent retrospective cohort study with glioblastoma patients demonstrated that the Khorana score is lacking specificity for risk prediction of glioblastomarelated VTE. 8 This may not only be the consequence of underrepresentation, but also due to the fact that the included risk factors are not specifically relevant for glioblastoma.Ay et al proposed to include high-grade glioma as "very high risk" within the model. 51Using this recommendation, an individual patient data meta-analysis by van Es et al reported an odds ratio (OR) of 3.5 (95% confidence interval [CI]: 0.89-14.0)for developing VTE in brain cancer patients with a high VTE risk (based on the Khorana score) compared to brain cancer patients with a Khorana score-based low-to-intermediate VTE risk. 52However, these data were based on a small population and not statistically significant.

Postoperative Thromboprophylaxis
Following the 2022 international clinical practice guidelines for the treatment and prophylaxis of VTE in patients with cancer, the use of low-molecular-weight heparin (LMWH) or unfractionated heparin is recommended postoperatively in cancer patients who are undergoing neurosurgery.The current standard for glioblastoma patients consists of postoperative treatment with LMWH for up to 10 days, which should be extended in case of prolonged immobilization. 53Additionally, graduated compression stockings and/or intermittent pneumatic compression may be used perioperatively. 54

Therapeutic Anticoagulation
For patients with brain tumors and established VTE, the use of LMWH or direct oral anticoagulants (DOACs) is recommended. 48Several retrospective cohort studies with primary brain cancer patients have been performed to evaluate the risk of bleeding following therapeutic anticoagulation.A significantly increased incidence of intracranial hemorrhage was observed in a retrospective cohort study with glioblastoma patients receiving LMWH, heparin, or warfarin following a VTE event. 55In line with this, retrospective data from high-grade glioma patients (of which 84.2% with glioblastoma) demonstrated a threefold increased risk of developing major intracranial hemorrhage in VTE patients receiving LMWH as compared to non-VTE patients (14.7 vs. 2.5%; hazard ratio [HR]: 3.37; 95% CI: 1.02-11.14;p ¼ 0.036). 56nother retrospective cohort study with high-grade glioma patients did not find an association between intracranial hemorrhage and the use of LMWH following VTE. 579][60] This has led to increased preference for DOACs due to the possibility of oral administration.However, large cohort studies with glioblastoma patients are required to determine the exact influence of therapeutic anticoagulation on bleeding risk in this specific population.

Long-Term Thromboprophylaxis
Long-term thromboprophylaxis for ambulatory cancer patients with LMWH is not generally prescribed because of the high bleeding risk, as recommended by the International Society on Thrombosis and Haemostasis. 49The only trial on prolonged prophylaxis in high-grade glioma patients, the PRODIGE trial (dalteparin vs. placebo), noticed a trend towards reduced VTE in the first 6 months, with 9 out of 99 LMWH patients (9.1%) developing VTE compared with 13 out of 87 placebo patients (14.9%;HR: 0.51; 95% CI: 0.19-1.4;p ¼ 0.29).Simultaneously, major bleeding occurred in 5 LMWH patients (5.1%) versus 1 placebo patient (1.1%; HR: 4.2; 95% CI: 0.48-36; p ¼ 0.22). 61However, this study was terminated prematurely due to expiration of study medica-tion, resulting in incomplete sample size and consequently, low statistical power.
Two recent prospective trials on the use of DOACs did not include patients with primary brain cancer at all (the Cassini trial on rivaroxaban 62 ) or only a small number (the AVERT trial on apixaban, n ¼ 24 63 ).Prospective data on long-term anticoagulation in larger cohorts of glioblastoma patients are unfortunately lacking.Due to potential advantages such as oral administration and relative safety, 60 prospective randomized clinical trials regarding the use of DOACs for long-term thromboprophylaxis in glioblastoma patients are warranted.
Altogether, the benefit-risk ratio of long-term thromboprophylaxis in glioblastoma patients is a delicate balance between the risk of developing life-threatening pulmonary embolism versus the risk of intracranial hemorrhagic events.Glioblastoma patients exhibit one of the highest risks of VTE in combination with increased susceptibility to intracranial hemorrhage, which warrants specialized prospective clinical trials and a glioblastoma-specific VTE risk assessment model.Risk stratification using novel procoagulant biomarkers may assist in identifying those patients with the highest risk of VTE, who may benefit from extended anticoagulation despite the increased risk of major bleeding.

Molecular Background of Glioblastoma-Related VTE
Hypercoagulability directly depends on the activity of procoagulant proteins, which are involved in thrombus formation.There are several proteins with a physiological role in hemostasis that are upregulated within the glioblastoma tumor depending on tumor-specific features (e.g., genetic aberrations, hypoxia, vascularization).This has consequences within the tumor, resulting in local microthrombi, as well as systemically, since these procoagulant proteins are also secreted by the tumor or present on circulating tumor cells and tumor cell-derived extracellular vesicles (EVs), allowing for procoagulant activity at distant sites in the circulation.In the next paragraphs, two key proteins involved in glioblastoma-related VTE, podoplanin and TF, will be described in more detail.

Podoplanin
Podoplanin is a transmembrane glycoprotein involved in platelet aggregation through the platelet-receptor CLEC-2. 64In healthy tissue, podoplanin is strongly expressed in lymphatic endothelial cells, being a widely used marker for lymphatic development. 65However, podoplanin is also frequently upregulated in several cancer types, such as skin cancer, lung cancer, germ cell cancer, and primary brain cancer, being involved in tumor progression, invasion, and metastasis. 66Indeed, overexpression has been associated with epithelial-to-mesenchymal transition (EMT) and lymphangiogenesis, which correlates with poor survival in cancer patients. 65Podoplanin is also expressed on circulating tumor cells and EVs in the blood, 67,68 which likely results in local platelet activation and, potentially, thrombus formation. 69Furthermore, the release of pro-angiogenic factors Thrombosis and Haemostasis © 2024.The Author(s).

VTE in Patients with Glioblastoma Kapteijn et al.
such as VEGF from platelet granules contributes to tumor growth, angiogenesis, and increased hypercoagulability. 70n glioblastoma, podoplanin overexpression strongly associates with intra-tumoral microthrombi and systemic VTE.Riedl et al described a sixfold increased risk of developing VTE in glioblastoma patients with high podoplanin expression levels compared to patients with low levels. 69hese patients also demonstrated low platelet counts and high D-dimer levels, presumably due to the consumption of platelets following podoplanin-induced platelet aggregation.Furthermore, a 2.6-fold increased mortality risk was observed in glioblastoma patients with high tumoral podoplanin expression.
Importantly, an association has been described between mutations within the gene IDH1 and decreased expression of podoplanin (see below). 71Since all glioblastoma tumors demonstrate IDH-wild-type expression following the most recent World Health Organization (WHO) classification of tumors of the central nervous system (2021), 72 glioblastoma patients inherently show high podoplanin levels and an increased VTE risk compared to patients with lower grade gliomas.This was confirmed in a recent cohort study with adult-type diffuse glioma (glioblastoma, IDH-wild-type vs. astrocytoma, IDH-mutant and oligodendroglioma, IDH-mutant and 1p/19q-codeleted), but no association was observed between the levels of circulating podoplanin and glioma subtype or cumulative VTE incidence. 73Furthermore, Tawil et al reported increased platelet activation following injection of glioma-derived podoplanin-positive EVs in mice, but no significant increase in plasma D-dimer levels.Thus, although a link has been described between VTE and high tumoral podoplanin levels in glioblastoma patients (as compared to patients with low levels), more research is required to determine the role of circulating podoplanin in glioblastoma-related VTE.

TF
TF is a transmembrane glycoprotein expressed on the surface of subendothelial cells.Upon vascular damage, TF may bind and activate its blood-borne ligand factor VII (FVII), resulting in the binary TF:FVIIa complex which activates FX into FXa. 74This subsequently leads to thrombin generation, platelet activation, and conversion of fibrinogen into fibrin.Aggregation of platelets together with fibrin at the site of injury ultimately results in clot formation.Additionally, TF is involved in intracellular signaling through G-protein coupled protease-activated receptors (PARs), present on the cell membrane of platelets, (sub)endothelial cells, and cancer cells. 74TF-mediated PAR2 signaling induces pro-angiogenic factors such as interleukin-8 (IL-8) and VEGF, 75,76 thereby contributing to several aspects of tumor progression such as migration, invasion, angiogenesis, and, potentially, hypercoagulability.
TF is upregulated in virtually all cancer types, resulting in TF expression within the tumor as well as in the circulation on circulating tumor cells and tumor-cell secreted EVs. 77This contributes to the hypercoagulable state in cancer patients, increasing the risk of VTE, but also induces tumor progression through TF-mediated PAR signaling. 78Because of being involved in both cancer and coagulation, TF is generally accepted as a protagonist that connects cancer and VTE. 79,80n brain cancer, TF upregulation is frequently observed, being associated with grade of malignancy and vascular density. 81,82Consequently, TF is highly expressed in glioblastoma.Furthermore, TF expression is induced following exposure to hypoxia, which is often seen in glioblastoma and was shown to enhance coagulation. 83Indeed, in a small cohort of brain cancer patients (n ¼ 96), Thaler et al observed widespread TF expression in glioblastoma tumor tissue using immunohistochemical staining. 84However, only 56 glioblastoma patients were included in this cohort (58.3%), and no correlation was found between TF expression and VTE.Nonetheless, as this study was clearly underpowered, an association between TF expression and risk of VTE in glioblastoma cannot be excluded.
The role of TF-positive EVs in cancer-associated thrombosis is still debated.Despite the fact that preclinical data emphasize the potential influence of circulating TF on VTE in cancer patients, 77 this could not readily be reproduced in clinical studies on ovarian cancer and non-small cell lung cancer. 85,86][89] Secretion of TF-bearing EVs by human glioblastoma cells was already described in 1984, being associated with platelet aggregation and thrombus formation. 90In a small cohort of glioblastoma patients (n ¼ 61), TF-EV levels were found to be particularly high prior to surgery, with a further increase up to 7 months afterwards. 91Increased TF-EV levels at 7 months after surgery were associated with subtotal tumor resection and radiological disease progression.Baseline mean TF-EV numbers were significantly higher in glioblastoma patients who developed VTE.In line with this, Unruh et al described a positive correlation between preoperative TF-EV activity and the risk of VTE in patients with IDH-wildtype glioma. 92Furthermore, a recent study with adult-type diffuse glioma demonstrated a trend toward increased TF-EV activity in IDH-wild-type glioblastoma, with a significant association between highest TF-EV activity and both the fastest time to VTE and the highest cumulative VTE incidence. 73However, in a study by Thaler et al, no significant association was found between TF-EV activity and development of VTE in brain cancer, 93 possibly due to differences in detection methods used as mentioned by Sartori et al. 91 Moreover, this cohort (n ¼ 119) consisted of high-grade glioma patients instead of specifically focusing on glioblastoma patients, and did not distinguish between IDH-wild-type and IDH-mutant glioma.
Since high TF-EV levels and activity can be directly linked to systemic hypercoagulability in selected cancer types including glioblastoma, these may be used as prognostic markers for the risk of glioblastoma-related VTE.Variation between studies may be explained by different detection methods assessing two different states of TF, being either encrypted or decrypted. 94Decrypted TF induces coagulation by activating FX.In contrast, when surface-expressed TF is in an encrypted state, it induces intracellular signaling rather than exerting direct procoagulant activity.TF decryption is required to Thrombosis and Haemostasis © 2024.The Author(s).
VTE in Patients with Glioblastoma Kapteijn et al.
become fully active, which can be achieved by various stimuli ultimately resulting in increased exposure of phosphatidylserine (PS).In the presence of surface-expressed TF, the negatively charged PS stimulates formation of the procoagulant tenase and prothrombinase complexes, thus accelerating coagulation. 95Additionally, TF decryption requires the formation of intracellular disulfide bonds between Cys 186 and Cys 209 as mediated by protein disulfide isomerase (PDI). 96Since PDI is present on EV surfaces, this may induce formation of procoagulant TF-positive EVs. 97However, the exact role of PDImediated TF decryption in cancer-associated thrombosis remains to be determined. 98,99hile decrypted TF directly contributes to VTE, cryptic TF induces tumor progression through TF-mediated PAR2 signaling, which indirectly may increase hypercoagulability through expression of VEGF and IL-8. 75,76Thus, both forms of TF may increase the risk of VTE, especially in highrisk cancers such as glioblastoma.Since TF decryption allows for increased procoagulant activity on (circulating) tumor cells and TF-EVs, further inquiry about this process will contribute to our knowledge of the underlying molecular mechanisms of VTE in glioblastoma.
Both podoplanin and TF are considered to play a central role in glioblastoma-related VTE, as summarized in ►Fig. 1.While high tumoral podoplanin expression has been shown to associate with VTE in glioblastoma, probably resulting in the secretion of podoplanin-bearing EVs, a positive correlation has also been described between the risk of glioblastoma-related VTE and both TF-EV levels and activity.Since systemic VTE in cancer presumably depends on increased hypercoagulability at distant sites from the tumor, as mediated by procoagulant EVs, TF may be a pivotal factor in glioblastoma-related VTE.This is in line with the most common genetic aberrations in glioblastoma, which are known to upregulate TF in several ways (see below).In addition, co-expression of TF and podoplanin within the glioblastoma tumor may have synergistic effects (see ►Fig. 1).In fact, in xenograft models it was shown that tumors expressing both TF and podoplanin demonstrated increased intravascular fibrin staining and vessel-occluding thrombi when compared to tumors expressing TF or podoplanin only. 68Furthermore, tumoral expression of procoagulant proteins in glioblastoma may be highly heterogeneous, representing a mosaic of different glioblastoma subtypes which requires further investigation, e.g., by single-cell RNA sequencing.Thus, the exact influence of TF and podoplanin may differ per glioblastoma patient, and a certain degree of cooperation is very likely.

Genetic Background of Glioblastoma-Related VTE
Inter-and intratumoral heterogeneity is a crucial and intrinsic hallmark of glioblastoma.This is captured by the different glioblastoma subtypes defined by Verhaak et al, who described a molecular classification system based on genetic signature. 100hile originally including four subtypes (i.e.classical, mesenchymal, proneural, and neural), more recent categorization resulted in a total of three (exclusion of the neural subtype 101 ), comprising glioblastoma, IDH-wild-type only following the most recent WHO classification of tumors of the central nervous system (2021). 72Classical glioblastoma harbors the most common genetic aberrations in glioblastoma (hence the name), such as amplification of the epidermal growth factor receptor (EGFR) and homozygous deletion of CDKN2A/B.At the same time, wellknown mutations in TP53, NF1, and PDGFRA are relatively underrepresented. 100The mesenchymal subtype is mostly characterized by deactivating mutations in the gene NF1.Furthermore, this subtype is known for its high expression of mesenchymal and astrocytic markers (e.g., CD44, MET), which leads to de-and transdifferentiated tumors due to the high degree of EMT. 100,102Finally, proneural glioblastoma shows distinct alterations in PDGFRA or TP53.Additionally, because of increased expression of some well-known proneural development genes and stem-cell markers such as SOX2 and Notch signaling proteins, the proneural signature is associated with cellular development and proliferation. 102his molecular classification has great implications for glioblastoma prognosis and therapy, as treatment efficiency highly depends on molecular subtype.In terms of survival, mesenchymal glioblastoma is associated with worse prognosis as compared to nonmesenchymal glioblastoma (i.e., classical or proneural). 101The risk of developing VTE also relies on glioblastoma subtype and subsequent procoagulant gene and protein expression, the so-called coagulome. 14nterestingly, the mesenchymal subtype shows the most procoagulant gene expression profile, 14 although this has not been validated by looking at actual VTE events in a cohort with glioblastoma patients.In a nested case-control study with 46 glioblastoma patients, of which 23 with and 23 without VTE, our research group used RNA-sequencing data to explore a potential link between molecular glioblastoma subtype and VTE.Here, proneural/neural glioblastoma was identified as a potential risk factor (OR: 3.05; 95% CI: 0.81-10.17;p ¼ 0.19).However, these data were not statistically significant. 103Below, the most common genetic aberrations in glioblastoma are discussed, as well as their potential link with a hypercoagulable state.

IDH1
Isocitrate dehydrogenase 1 (IDH1) is an enzyme of the citric acid cycle that normally converts isocitrate into α-ketoglutarate.Mutations in the IDH1 gene, of which R132H is observed most frequently, are highly prevalent in glioma. 104IDH1-mutant gliomas, currently classified as astrocytomas, IDH-mutant, are less aggressive and more sensitive to chemotherapy compared to glioblastoma, IDH-wild-type, resulting in better overall survival. 105IDH1 mutations are also associated with a decreased risk of VTE due to additional conversion of α-ketoglutarate into D-2-hydroxyglutarate, an oncometabolite that inhibits platelet aggregation. 92Moreover, IDH1 R132H mutation causes hypermethylation of the gene promoters of both F3, encoding TF, and Pdpn, encoding podoplanin, thus directly decreasing expression of the main procoagulant proteins involved in glioblastoma-related VTE. 71,92Nevertheless, following the most recent WHO classification, all glioblastomas are classified as IDH-wild-Thrombosis and Haemostasis © 2024.The Author(s).
VTE in Patients with Glioblastoma Kapteijn et al.
type. 72This may explain the increased risk of VTE in glioblastoma patients compared to lower grade gliomas, but does not allow for VTE risk stratification within the glioblastoma population.

EGFR
EGFR is a transmembrane tyrosine kinase receptor involved in cell proliferation, differentiation, and migration.EGFR overexpression contributes to tumor growth and EMT, being associated with invasion and metastasis. 106The EGFR gene is one of the most frequently mutated genes in glioblastoma, but less common in lower grade gliomas. 107,108It is estimated that up to 60% of glioblastoma tumors show EGFR upregulation, either caused by genomic amplification, rearrangement, and/or mutation. 109This is especially the case in classical glioblastoma, which is characterized by EGFR amplification. 100 The most common EGFR mutation in glioblastoma is EGFRvIII, which is caused by deletion of exon 2-7 and only found in cancer cells.EGFRvIII shows low constitutive activity and ligand-independent signaling, mainly through RAS and mTOR, resulting in increased proliferation, migration, and invasion. 106In 1994, EGFRvIII-expressing glioma cell lines were already reported to induce increased tumorigenesis in nude mice. 110Glioblastoma cells expressing EGFRvIII  demonstrate upregulation of TF as well as other procoagulant proteins, such as PAR1, PAR2, and FVII. 111,112Thus, constitutive signaling by EGFRvIII promotes tumor progression and a procoagulant microenvironment, suggesting a role in glioblastoma-related VTE.In line, expression of EGFR and TF was found to correlate in tumor specimens of patients with classical glioblastoma, 14 indicating an increased risk of VTE in glioblastoma patients with this specific subtype.However, to the best of our knowledge, a direct correlation between EGFR mutations, specifically EGFRvIII, and VTE in patients with glioblastoma has not been observed.
Intriguingly, high levels of EGFR and EGFRvIII expression in both glioblastoma cell lines and patient-derived glioblastoma stem cells correlated with low podoplanin expression. 68Thus, EGFR conversely regulates the two main procoagulant proteins in glioblastoma, podoplanin, and TF.

PTEN
PTEN is a tumor suppressor involved in cell-cycle control through inhibition of the PI3K-AKT signaling pathway.Loss of PTEN expression is common in all glioblastoma subtypes, with 20 to 40% of glioblastoma tumors harboring inactivating PTEN mutations and roughly 80% demonstrating loss of PTEN expression, e.g., by PTEN promoter methylation. 113,114PTEN inactivation leads to upregulation of RAS and mTOR signaling and subsequent overexpression of TF, similarly to activation of EGFR. 83Indeed, glioma cells that do not express PTEN show increased coagulation compared to cells with wildtype PTEN expression, especially during hypoxia, which was confirmed by increased TF levels in cell-culture media.This is in line with another study by Rong et al, in which EGFR overexpression led to TF upregulation in a PTEN-null glioblastoma cell line, which could be rescued by PTEN restoration. 115Here, TF expression was found to be controlled by PTEN-mediated transcriptional regulation.Moreover, in non-small cell lung cancer patients, combined presence of inactivating mutations in TP53 and PTEN was previously found to increase TF mRNA expression and decrease survival. 116n addition to TF, an inverse correlation has been described between PTEN expression and podoplanin levels in glioblastoma cell lines in vitro, an in vivo mouse model, and primary glioblastoma samples, 117 which could very well explain upregulation of podoplanin in glioblastoma.
As PTEN is often inactivated in glioblastoma patients, its ability to regulate podoplanin and TF might explain the increased risk of VTE, especially in combination with oncogenic expression of EGFR.However, a direct correlation between PTEN activity and VTE in glioblastoma patients has not been described yet.

CDKN2A/B
In addition to genetic alterations in EGFR and PTEN, homozygous deletion of CDKN2A and CDKN2B is also frequently observed in glioblastoma patients.Both genes are involved in cell-cycle control by regulating pRB and p53 signaling. 118DKN2A encodes both p16 INK4a and p14 ARF by alternative splicing, which were found to be inactivated in 52 and 49% of glioblastoma tumors, respectively. 119In classical glioblastoma, 94% of samples showed homozygous deletion of CDKN2A in co-occurrence with amplification of EGFR. 100 CDKN2A deletion may exert a stimulating effect on TF expression, as p14 ARF suppresses TF-induced procoagulant activity in glioblastoma cells by regulating tissue factor pathway inhibitor-2 (TFPI-2). 120Furthermore, p14 ARF normally inhibits Mdm2, the negative regulator of p53.Disruption of p14 ARF activity therefore results in inactivation of p53, which may lead to TF upregulation in combination with KRAS mutations, as described in colorectal cancer, or PTEN mutations, as described in non-small cell lung cancer. 116,121Thus, CDKN2A deletion may increase hypercoagulability in glioblastoma patients by downregulation of TP53.Co-occurrence of EGFR amplification and CDKN2A deletion in classical glioblastoma may therefore result in an increased risk of developing VTE.
Homozygous deletion of CDKN2B, encoding p15 INK4b , has been described in 47% of all glioblastoma patients. 119Interestingly, CDKN2B mutations were associated with a significantly increased risk of cancer-associated thrombosis in patients with solid tumors, independent of cancer type. 122In the same study, CDKN2A was also part of the top 10 of somatic mutations that associate with an increased VTE risk in cancer patients, although significance was lost after false discovery rate adjustment.Moreover, our group recently reported a link between CDKN2A deletion and VTE in a cohort of 324 glioblastoma patients. 123A targeted DNA-sequencing approach demonstrated a 12-month adjusted cumulative incidence of VTE of 12.5% in glioblastoma patients with a CDKN2A deletion, compared to 5.4% in glioblastoma patients with wildtype CDKN2A expression.This resulted in an HR of 2.53 (95% CI: 1.12-5.73,p ¼ 0.026).Thus, frequently observed homozygous deletion of CDKN2A/B in glioblastoma patients may significantly increase the risk of VTE.
In the same study, we used the cBioPortal for Cancer Genomics to study a potential link between CDKN2A deletion and mRNA expression levels of podoplanin and TF.Based on the Glioblastoma Multiforme dataset of the PanCancer Atlas (TCGA), we found an inverse correlation between CDKN2A expression and podoplanin mRNA levels (p ¼ 0.009). 123This is the only report on the relation between CDKN2A/B alterations and expression levels of podoplanin so far.A similar effect was observed for TF mRNA expression, although not statistically significant (p ¼ 0.058).
Altogether, there are several genetic aberrations that may affect the procoagulant genetic signature in glioblastoma patients, as summarized in ►Table 1 and ►Fig. 1.Moreover, the aforementioned signaling pathways are all interconnected, thereby amplifying hypercoagulability.That is, EGFR amplification and PTEN deletion both induce RAS signaling, which is frequently overactivated in cancer and by itself induces a plethora of procoagulant effects, such as upregulation of pro-angiogenic VEGF or TF itself. 83,124Furthermore, TP53 mutations are often observed in proneural glioblastoma, and are known to promote TF expression in combination with inactivating alterations in either PTEN or CDKN2A.Intra-tumoral heterogeneity may impact the local tumor microenvironment even further, resulting in hypercoagulable niches within the tumor that affect the procoagulant systemic state.Thus, the combination of procoagulant mutations within the glioblastoma tumor may result in a patient-specific genetic risk profile for VTE, which needs to be fully addressed in order to identify glioblastoma patients with the highest risk of VTE.In this regard, sequencing approaches may be of great value to develop personalized treatment strategies for glioblastoma-related VTE.

Future Directions
Due to the poor overall survival of glioblastoma patients and the high burden of glioblastoma-related VTE, novel biomarkers and treatment strategies are highly warranted.Furthermore, a glioblastoma-specific risk assessment model for VTE may significantly improve decision making regarding the use of thromboprophylaxis.In the final part of this review, an overview is given of the potential future directions to further develop a personalized approach for preventing VTE in patients with glioblastoma.

Current Biomarkers for Glioblastoma-Related VTE
Several studies have attempted to identify biomarkers to optimize risk calculation for VTE in glioblastoma.Suggested biomarkers mainly include clinical parameters for VTE risk prediction in the general population.In a study with highgrade glioma patients (n ¼ 141), of which 68.1% with glioblastoma, three potential biomarkers were identified: low platelet count, elevated D-dimer, and high soluble P (sP)selectin. 125In addition, FVIII activity and leukocyte count both showed borderline significance.Exploratory risk assessment models including either low platelet count and elevated sP-selectin or low platelet count, high leukocyte count, and elevated D-dimer resulted in a VTE probability of 23.0 and 37.7%, respectively.Furthermore, high D-dimer plasma levels, elevated von Willebrand factor levels, and decreased clotting time were associated with increased hypercoagulability in glioblastoma patients as compared to patients with meningioma. 126Another study described a 2.1-fold increased VTE risk in high-grade glioma patients, of which 85% with glioblastoma, with elevated FVIII activity. 127A benefit of these coagulation markers is their current diagnostic use in the clinic, resulting in detection methods being widely available.However, these parameters are not glioblastoma-specific, and the implementation of TF and/or podoplanin may further increase the prediction value of potential risk stratification tools for VTE in glioblastoma patients.
Intriguingly, in the general cancer population, high instead of low platelet count has been identified as a risk predictor for VTE. 50,128This may be due to the fact that glioblastoma patients exhibit increased expression levels of podoplanin, which induces platelet aggregation and, consequently, platelet consumption. 69This finding underlines the tumor-specific biology in brain cancer patients and further warrants a VTE risk assessment model for this specific population.
Podoplanin as a Biomarker and Therapeutic Target for Glioblastoma-Related VTE Tumoral podoplanin expression may be a relevant biomarker for glioblastoma-related VTE, as the risk of VTE is increased in glioblastoma patients with high podoplanin levels as compared to patients with low levels. 69This can be examined by immunohistochemical staining of podoplanin-positive tumor tissue following surgical resection, although specific antibodies for tumor-expressed podoplanin are required as podoplanin is also expressed in healthy tissue. 129nterestingly, the podoplanin-specific antibody NZ-1 could be used to block podoplanin-mediated platelet aggregation in glioblastoma cells, which may be useful for the reduction of VTE. 130Furthermore, combination of this antibody with chimeric antigen receptor-transduced T-cells resulted in Tcell recognition of podoplanin-positive glioblastoma cells, which inhibited growth of glioma xenografts in vivo. 131Thus, podoplanin may be a promising therapeutic target for glioblastoma, but the implications in a clinical setting remain to be investigated.
Circulating podoplanin is assumed to contribute to glioblastoma-related VTE as well, 68,69,132 but this has not been demonstrated so far.Nevertheless, since levels of circulating podoplanin have been measured in glioblastoma patients, 73 and podoplanin-induced platelet activation likely plays a role in thrombogenesis, the use of circulating podoplanin as a prognostic biomarker for glioblastoma-related VTE may be promising.
TF as a Biomarker and Therapeutic Target for Glioblastoma-Related VTE TF-EVs may be used as biomarker for VTE in glioblastoma, as both TF-EV levels and activity have been associated with increased VTE incidence. 73,91A randomized controlled trial with advanced cancer patients reported a VTE risk reduction of 80% in patients with high TF-EV levels using thromboprophylaxis compared to high TF-EV patients without prophylactic treatment, 133 thus demonstrating the value of TF-EVs as prognostic biomarker for VTE.However, no glioblastoma patients were included in this study.Nevertheless, some promising TF-targeting treatment strategies for glioblastoma have been described over the last years.Induced expression of TFPI-2, which inhibits TF-mediated coagulation, resulted in impaired tumor growth and vessel formation in human glioblastoma cells in vitro and in vivo. 134In line with this, the TF-targeting antibody 10H10 was shown to reduce tumor cell invasion and vascular activation in a human xenograft glioblastoma model. 135Finally, the tick-derived TF inhibitor Ixolaris was found to block TF-induced procoagulant activity by attenuating tenase complex assembly in glioblastoma cells in vitro. 136Furthermore, in vivo glioblastoma tumor growth in Ixolaris-treated mice was inhibited as a result of VEGF downregulation and decreased tumor vascularization.Thus, TF-targeting treatment strategies in glioblastoma may affect both tumor progression and prothrombotic activity.To date, TF-directed therapies are increasingly studied in clinical trials aimed at a broad Thrombosis and Haemostasis © 2024.The Author(s).
VTE in Patients with Glioblastoma Kapteijn et al.   showing the highest overexpression. 103Taken together, every glioblastoma patient exhibits its own procoagulant profile, which warrants personalized treatment to determine the benefit-risk ratio of thromboprophylaxis.A glioblastoma-specific VTE risk assessment model including tumor genomics is required to identify glioblastoma patients in which the risk of bleeding due to extended anticoagulation is outweighed by the decreased risk of VTE.

Conclusion
Glioblastoma patients are among the cancer patients with the highest risk of developing VTE.Increased local hypercoagulability is caused by a combination of vascular pathology and hypoxia, which is fueled into a systemic procoagulant state due to a variety of genetic aberrations within the tumor that affect the expression of procoagulant proteins.Particularly, the role of TF and podoplanin has been increasingly linked to hypercoagulability and development of VTE over the last years.In terms of a glioblastoma-specific VTE risk stratification model, expression levels of procoagulant EVs as well as readily available genomic markers may add great value to decision making about the use of thromboprophylaxis for glioblastoma patients.Potentially in combination with RNA-sequencing methods, this will lead to personalized VTE risk prediction which is required to improve future treatment strategies for glioblastoma-related VTE.

Fig. 1
Fig.1Proposed molecular and genetic mechanisms that underlie VTE in patients with glioblastoma.Upregulation of EGFR in glioblastoma cells results in increased signaling through PI3K-AKT and RAS-RAF-ERK, which both induce TF expression.Loss of PTEN prevents inhibition of AKT, which together with EGFR leads to increased expression of TF.Deletion of CDKN2A precludes expression of p14 ARF , which normally inhibits Mdm2, the negative regulator of p53.Consequently, Mdm2 represses p53 activity, resulting in TF upregulation.Altogether, this leads to increased expression of TF on the cell surface as well as on extracellular vesicles (EVs) in the circulation.Here, TF induces the coagulation cascade by activation of FVII into FVIIa, ultimately resulting in fibrin aggregation.On the other hand, increased expression of podoplanin in glioblastoma cells likely causes release of podoplanin-positive EVs, which induce platelet activation and aggregation through the platelet receptor CLEC-2.Collectively, in combination with TF-mediated fibrin aggregation, this results in clot formation, thus increasing the risk of VTE in patients with glioblastoma.EGFR, epidermal growth factor receptor; FVII, factor VII; TF, tissue factor; VTE, venous thromboembolism.
VTE in Patients with Glioblastoma Kapteijn et al.

Table 1
Common genetic aberrations in glioblastoma and their potential mechanistic impact on VTE 123 in Patients with Glioblastoma Kapteijn et al.spectrum of cancer types and stages,137but clinical studies specifically focusing at TF-related treatment for glioblastoma and glioblastoma-related VTE are warranted.Personalized Treatment Using Sequencing ApproachesTumor genomics needs to be considered when assessing the risk of VTE in glioblastoma patients, since tumor heterogeneity and hypercoagulability are largely dictated by the underlying genetic profile of the tumor.DNA sequencing may therefore be a promising approach to discover novel VTE biomarkers.In the largest study to date to associate tumorspecific genetic aberrations with VTE (n ¼ 11,695), Dunbar et al described a link between increased VTE risk in patients with solid tumors and tumor mutations in STK11, CDKN2B, KEAP1, KRAS, CTNNB1, and MET.122This cohort consisted for 4% of high-grade glioma patients.Specifically focusing on a cohort with 324 glioblastoma patients, our research group showed that tumoral CDKN2A deletion is associated with an increased risk of VTE using targeted DNA sequencing.123Furthermore,we used next-generation RNA sequencing to discover novel tumor-expressed genes and signaling pathways that associate with glioblastoma-related VTE.This nested case-control study consisting of 23 glioblastoma patients with VTE and 23 glioblastoma patients without VTE demonstrated a potential role for Sonic Hedgehog signaling, with classical Sonic Hedgehog target gene GLI1 Abbreviations: CI, confidence interval; HR, hazard ratio; TF, tissue factor; VTE, venous thromboembolism.Thrombosis and Haemostasis © 2024.The Author(s).