Bleeding tumors, bleeding genes: new study maps the molecular fingerprint of hemorrhagic glioblastoma
In a finding that could reshape how clinicians interpret one of the most dramatic brain tumor presentations, researchers at The University of Texas Health Science Center at Houston have shown that when glioblastoma bleeds from within, the tumor’s underlying genetic architecture may look meaningfully different from its non-bleeding counterpart, even if survival outcomes remain largely unchanged. The study, published in the Journal of Neuro-Oncology, offers one of the most detailed genomic portraits to date of hemorrhagic glioblastoma, a radiographic subtype that has long puzzled neurosurgeons and neuro-oncologists alike.
Glioblastoma, IDH-wildtype, is the most common and most aggressive primary malignant tumor of the central nervous system, striking roughly three to four people per 100,000 each year. Even with maximal surgical resection followed by radiotherapy and the chemotherapy drug temozolomide, median survival hovers around eighteen to twenty months. Over the past decade, the World Health Organization has reclassified these tumors based on molecular markers rather than appearance under the microscope alone, yet the relationship between a tumor’s genetic alterations and its visible, phenotypic behavior on imaging remains incompletely understood. One such visible behavior is spontaneous bleeding into the tumor itself, a phenomenon known as intratumoral hemorrhage.
The new study set out to answer two questions: whether hemorrhagic glioblastoma carries a distinct genomic signature, and whether the presence of blood within the tumor affects how long patients live. To do so, the team conducted a retrospective analysis of 176 patients with molecularly confirmed IDH-wildtype glioblastoma treated between 2009 and 2020. All patients had undergone both preoperative and postoperative magnetic resonance imaging, and all had comprehensive next-generation sequencing data covering 205 cancer-related genes. The researchers applied the Visually Accessible Rembrandt Images, or VASARI, criteria to define intratumoral hemorrhage, a standardized framework that identifies bleeding within the tumor matrix as intrinsic foci of hypointensity on T2-weighted imaging or hyperintensity on T1-weighted images. Two blinded readers, including a neuroradiologist with more than two decades of experience, independently reviewed the scans, and volumetric measurements of each tumor’s enhancing, necrotic, and fluid-attenuated inversion recovery signal components were calculated on a separate workstation.
The results were striking in their clinical granularity. Of the 176 patients, 105, or 59.6 percent, exhibited intratumoral hemorrhage on preoperative imaging, a rate considerably higher than most previous estimates, which ranged from roughly 2 to 29 percent depending on the study population and imaging criteria. Patients with hemorrhagic tumors were also younger, with a median age of 59 compared to 63 in the non-hemorrhagic group. They were more likely to be Hispanic, comprising 20 percent of the hemorrhagic cohort versus 7 percent of the non-hemorrhagic group, and more likely to present with motor deficits, at 38 percent versus 21 percent. Hemorrhagic tumors were also larger, with a median total volume of 147.6 cubic centimeters compared to 122.2 cubic centimeters, and harbored greater volumes of contrast enhancement and necrosis, findings consistent with a more vascular and biologically aggressive phenotype at presentation.
At the molecular level, the researchers identified several genes whose alteration frequency differed between the two groups, though the findings warrant careful interpretation. Tumors without hemorrhage were more likely to harbor amplifications or mutations in four genes clustered on chromosome 4q12: PDGFRA, present in 26.8 percent of non-hemorrhagic tumors versus 13.3 percent of hemorrhagic ones; KIT, at 19.7 percent versus 8.5 percent; KDR, at 18.3 percent versus 6.6 percent; and PIK3R1, at 11.3 percent versus 2.8 percent. Conversely, the histone methyltransferase gene SETD2 was more frequently altered in hemorrhagic tumors, appearing in 8.5 percent of those cases compared to just 1.4 percent of non-hemorrhagic ones. However, when the investigators applied the Benjamini-Hochberg correction for multiple comparisons, none of these individual gene differences retained statistical significance, meaning the results should be viewed as hypothesis-generating rather than definitive conclusions.
The biological logic behind these gene associations is nonetheless compelling. PDGFRA, KIT, and KDR all encode receptor tyrosine kinases, proteins that sit on the cell surface and relay growth and vascular signals into the cell interior. When amplified together, as often happens in glioblastoma, these genes drive aggressive tumor proliferation and remodel the blood vessels that feed the tumor. KDR in particular is a vascular endothelial growth factor receptor, central to the formation of new, often fragile blood vessels through angiogenesis. Tumors rich in these pathways may build denser and more stabilized vascular networks, which paradoxically could make them less prone to catastrophic bleeding. By contrast, the SETD2 gene encodes an enzyme that adds a chemical mark, trimethylated histone H3 lysine 36, to chromatin, a modification tied to proper gene transcription, RNA splicing, and genome stability. Preclinical work has shown that SETD2 knockout mice develop abnormal forebrain vasculature and spontaneous hemorrhage, hinting at a mechanism by which epigenetic disruption of vascular integrity could predispose certain glioblastomas to bleed. This line of evidence, while preliminary, lends plausible biological weight to the observed association.
Despite these molecular differences, survival outcomes told a quieter story. The researchers found no statistically significant difference in progression-free survival between the two groups, with a median of 8.2 months for hemorrhagic tumors and 8.3 months for non-hemorrhagic ones. Overall survival was similarly indistinguishable, at 18.4 months versus 19.7 months. Even when the team drilled down into what happens after a tumor recurs, patients who received the anti-vascular drug bevacizumab at first progression did no better or worse than those who did not, regardless of whether the original tumor had bled. Multivariable Cox regression analysis, which accounted for age, performance status, extent of resection, chemoradiotherapy, and contrast-enhancing tumor volume, confirmed that hemorrhage status itself was not an independent predictor of death. What did matter, consistent with prior glioblastoma research, were older age, unmethylated MGMT promoter status, and tumor location, with parietal tumors associated with a lower risk of death and methylated MGMT carrying a protective hazard ratio of 0.50.
These findings challenge a lingering assumption that a bleeding brain tumor automatically signals a worse prognosis. Earlier reports, including a 2024 study of 167 glioma patients, suggested that certain mutations such as CDKN2B, KMT5B, and PIK3CA were enriched in hemorrhagic tumors and that hemorrhage might worsen outcomes. The new study, with its more homogeneous cohort of purely IDH-wildtype glioblastomas and its use of standardized VASARI imaging criteria, did not replicate those specific genetic associations, likely because of differences in how tumors were classified and how hemorrhage was defined. The discrepancy underscores a recurring theme in neuro-oncology: heterogeneous patient populations and inconsistent imaging definitions can yield wildly different conclusions, and only carefully stratified studies like the present one can begin to disentangle the true biology.
The clinical implications are nonetheless real. Glioblastoma that presents with intratumoral hemorrhage can mimic a hemorrhagic stroke, and prior work suggests that roughly 3 percent of presumed strokes ultimately turn out to be underlying tumors. Recognizing this is important, since a misdiagnosis could delay definitive treatment. The higher rate of motor deficits and larger tumor volumes seen in hemorrhagic glioblastoma also raises the question of whether hemorrhage reflects a more acute, mass-effect-driven disease course, even if long-term survival is unaffected. On the research side, the tentative links between receptor tyrosine kinase amplification, epigenetic dysregulation, and vascular stability open new avenues for investigating how specific genetic programs govern tumor vessel fragility.
The study is not without limitations. Its retrospective design introduces the possibility of selection bias, since not all patients in the cohort underwent next-generation sequencing or volumetric postoperative imaging. Hemorrhage was identified using conventional MRI sequences rather than more sensitive techniques such as susceptibility-weighted imaging or T2-star imaging, which can detect microscopic blood products that the standard sequences miss. The researchers acknowledge that future studies incorporating these advanced imaging modalities, along with intraoperative and pathological confirmation, will be needed to better characterize the full spectrum of hemorrhagic patterns and their molecular underpinnings. Additionally, MGMT promoter methylation status was not available for all patients, and the relatively modest cohort size limits statistical power, particularly when correcting for dozens of gene comparisons simultaneously.
Still, the work represents an important step forward in linking the visible phenotype of a brain tumor to its hidden genomic landscape. By focusing on a single, molecularly defined tumor type and applying a reproducible imaging framework, the researchers have laid the groundwork for larger, multi-institutional studies that could confirm whether the connections between SETD2, receptor tyrosine kinase signaling, and vascular fragility hold up under scrutiny. For now, patients and clinicians can take comfort in one clear message: while a bleeding glioblastoma may look more aggressive on the scan and announce itself with more dramatic symptoms, it does not, on its own, spell a shorter life.
Subject of Research: People with IDH-wildtype glioblastoma, comparing those whose tumors showed intratumoral hemorrhage on MRI with those whose tumors did not, using next-generation sequencing and survival analysis
Subject of Research: Cancer
Article Title: Molecular features of intratumoral hemorrhage in glioblastoma isocitrate dehydrogenase-wildtype
Article References: Bueno, A., Dono, A., Alcantar, O., Amezquita-Contreras, C., Ocasio, L., Hsu, S., Amsbaugh, M., Zhu, J.-J., Blanco, A. I., Riascos, R. F., Ballester, L. Y., Tandon, N., & Esquenazi, Y. (2026). Molecular features of intratumoral hemorrhage in glioblastoma isocitrate dehydrogenase-wildtype. Journal of Neuro-Oncology, 179(2), Article 51. https://doi.org/10.1007/s11060-026-05749-3
Image Credits: AI Generated
DOI: 10.1007/s11060-026-05749-3
Keywords: glioblastoma, intratumoral hemorrhage, IDH-wildtype, PDGFRA, KIT, KDR, PIK3R1, SETD2, VASARI, next-generation sequencing, MRI, survival outcomes
Cite Scienmag News
APA MLA Chicago
Nathaniel Bowman. (September 6, 2026). New molecular insights into intratumoral hemorrhage in IDH-wildtype glioblastoma. Scienmag. https://scienmag.com/new-molecular-insights-into-intratumoral-hemorrhage-in-idh-wildtype-glioblastoma/
Nathaniel Bowman. “New molecular insights into intratumoral hemorrhage in IDH-wildtype glioblastoma.” Scienmag, 6 September 2026, https://scienmag.com/new-molecular-insights-into-intratumoral-hemorrhage-in-idh-wildtype-glioblastoma/. Accessed 6 September 2026.
Nathaniel Bowman. “New molecular insights into intratumoral hemorrhage in IDH-wildtype glioblastoma.” Scienmag. September 6, 2026. https://scienmag.com/new-molecular-insights-into-intratumoral-hemorrhage-in-idh-wildtype-glioblastoma/
Copy citation Download RIS
Tags: brain tumor hemorrhage genetic insightsclinical significance of hemorrhagic glioblastgenetic differences in bleeding versus non-bleeding glioblastomagenetic markers in hemorrhagic glioblastomagenomic fingerprint of hemorrhagic glioblastomagenomic landscape of hemorrhagic brain tumorsglioblastoma hemorrhageglioblastoma radiographic subtypesIDH-wildtype glioblastoma genetic profilingIDH-wildtype glioblastoma tumor genomicsimpact of geneticimpact of tumor bleeding on glioblastoma prognosisimplications of tumor hemorrhage for prognosis and treatmentintratumoral bleeding in glioblastomaintratumoral hemorrhage molecular mechanismsmolecular differences in bleeding versus non-bleeding glioblastomasmolecular markers associated with tumor hemorrhagemolecular pathways involved in glioblastoma hemorrhagemolecular profiling of hemorrhagic glioblastomaradiographic subtypes of glioblastomatumor microenvironment and bleeding in glioblastoma
