Originally published in Medscape
Fifteen years ago, Jinan Behnan, PhD, was a graduate student at Lund University in Lund, Sweden, performing a routine dissection that would stay with her for more than a decade.
While removing a glioblastoma (GBM) tumor from a mouse model, Behnan felt something physically abnormal: The skull was paper-thin. The implication was immediate. She suspected osteoclasts, the body’s “bone-eating” cells, were being activated to attack and dissolve the bone.
Though she brought the idea to her superiors, it went uninvestigated due to limited funding and resources. But the idea stayed with her.
Now an assistant professor of neurological surgery at the Albert Einstein College of Medicine, New York City, and leading her own lab, it took Behnan 11 years to find the right environment and resources to start real work on her initial hunch and an additional 4 years to finish.
That persistence has paid off in a new study published in Nature Neuroscience, led by Behnan, which reveals that GBM may not be just a local brain malignancy but a systemic disease that hijacks the skull marrow to fuel its growth.
The research has also uncovered a therapeutic paradox that may finally explain why immune-modulating therapies in these patients so often fail.
Building Tools to See the Unseen
GBM is the most aggressive malignant primary tumor of the central nervous system. Despite decades of surgical and pharmaceutical advancements, the prognosis remains grim, with median survival rates between 15 and 19 months.
For Behnan, the scientific validation was a slow burn that didn’t ignite until she secured her Marie Curie fellowship in Sweden in 2019, where she approached Göran Andersson, PhD, an expert in molecular genetics, about her theory that GBM’s systemic disease profile involved osteoclasts dissolving skull bone.
Though Andersson initially questioned the link between bone cells and brain cancer, a pilot study involving three animals proved the concept, with collaborators confirming the presence of osteoclast activity.
But just as the research gained traction, the COVID pandemic paralyzed the lab, costing Behnan her critical animal models.
Forced to relocate to save the work, she eventually found a champion in Emad N. Eskandar, MD, chair of the Leo M. Davidoff Department of Neurosurgery at Montefiore Einstein, who offered her the resources to establish her own brain tumor lab.
Still, proving GBM systematically eats the skull required a technological leap. Behnan needed to image the microarchitecture of the skull bone of many mice with induced GBM at high resolution, a logistical and financial nightmare for a junior investigator.
So she found help.
Wade R. Koba, BS, a biomedical engineer at Albert Einstein College of Medicine, designed a custom microCT scanner capable of scanning four mouse skulls simultaneously. He calibrated the scanner and three dimensionally-printed a custom rig that kept the animals’ heads motionless, preventing the blurring that would otherwise ruin the images.
Working with Erika Yamashita, PhD, the team conducted high-resolution bone analysis on the microCT scans, quantifying the extent of skull erosion across multiple animals.
To visualize the cellular mechanism inside the calvarium — the top of the skull — Behnan and Abhishek Dubey, a PhD student at Albert Einstein College of Medicine, learned multiphoton imaging techniques from David R. Fooksman, PhD, also at Einstein. Koba printed a custom frame to adapt Fooksman’s microscope — originally built for femoral bone imaging — to the precise height requirements for brain imaging.
Then came the delicate work: Behnan performed cranial window surgery — removing a section of skull to expose the brain — while Dubey mixed fast-solidifying dental cement. Together they’d place a glass coverslip over the opening and use the cement to seal it to the surrounding skull, creating a stable optical window before transferring the tumor-bearing mouse to the multiphoton stage.
While Behnan stabilized the head and monitored the animal’s breathing throughout, Dubey checked imaging channels. Together they had mapped the imaging window, tag-teaming to keep mice stable through 6- to 8-hour sessions.
When the 1.5-terabyte files from the microscope overwhelmed the lab’s existing software, Nathaniel J. Killian, PhD, a specialist in Python programming and image analysis, built a custom pipeline to translate the massive data set into visible images.
“The light sheet made a very beautiful image,” said Behnan.
‘Moth-Eaten’ Lesions
The combination of Koba’s microCT, Yamashita’s bone analysis, Killian’s computational pipeline, and Behnan and Dubey’s multiphoton imaging provided the proof Behnan had waited more than a decade to see.
When the first high-resolution CT images came back, the contrast between the healthy control group and the GBM-bearing mice was undeniable.
For Koba, who looked at the raw data from an engineering perspective, the pattern of destruction was distinct. “I do remember [seeing the skull erosion firsthand],” Koba said of that first validation. “And that was such a nice finding to see.”
The images confirmed that the tumor was actively remodeling the skull, creating porous, “moth-eaten” lesions.
To ensure these findings weren’t limited to mice, the team examined medical records from human patients. They compared CT scans from 26 GBM patients against 22 people of the same age and gender who did not have cancer. The patients with brain tumors had significantly thinner skulls than the healthy group.
A Factory Production Line, Hijacked
Under normal conditions, the skull’s bone marrow acts as a balanced production line, generating a 50/50 split of myeloid cells and lymphoid cells.
But when GBM develops, it hijacks this production line. Using single-cell RNA sequencing and high-dimensional flow cytometry — the latter enabling deep molecular profiling and functional characterization of the immune cells — the team documented a dramatic shift: the tumor forces myeloid production to 70%-90% of total cellular output, while lymphoid lineages deteriorate.
The tumor becomes “a desert for T cells,” Behnan said, starved of the lymphocytes that might attack it. Even more striking, B-cell populations were reduced by 60%-94%, a near eradication.
And while the skull marrow showed signs of inflammatory activation, femoral bone marrow exhibited what researchers called a “cold” signature, with marked suppression of Type II interferon (interferon-gamma) response pathways. This suggests the tumor simultaneously suppresses systemic immunity in distant marrow sites while recruiting inflammatory cells from the skull — a dual-action immune evasion strategy.
As the calvarium dissolves and newly formed skull channels form, they seem to serve a sinister purpose. “It’s like a highway connecting the brain tumor to the marrow,” Behnan said. These channels increased in both number and size across all tumor stages, from early engraftment (day 8) to late-stage morbidity, facilitating bidirectional trafficking of cells and molecular signals.
The tumor can insidiously remodel a patient’s skull for its own purpose: survival and proliferation.
When the Cure Is the Poison
GBMs often produce colony-stimulating factor (CSF)-1, a growth factor that typically attracts macrophages to help the tumor spread.
E. Richard Stanley, PhD, a senior author of the study and expert in CSF-1 biology, views the interaction between the tumor and the skull as a subversion of standard developmental biology.
Having identified the supply line, the skull channels, and the mechanism of access — bone erosion — the team wondered whether blocking RANKL, a protein that promotes activation of osteoclasts, could stop progression of the tumor. After all, blocking the protein that sets off the erosion of the skull should work, right?
Well, not necessarily. They treated the tumor-bearing mice with osteoporosis drugs — zoledronic acid (a bisphosphonate) and an anti-RANKL antibody. The drugs worked. MicroCT analysis confirmed that bone thickness and density were restored to healthy levels.
But then came the shock. “We blocked the phenomenon in the skull, but the tumor inside it progressed, which was puzzling for us,” said Behnan.
Not only did the tumor grow faster in the mesenchymal subtype models, but the osteoporosis drugs also had a devastating effect on immunotherapy. When combined with anti-PD-L1, which should unleash the immune system, osteoporosis drugs eliminated the survival benefit.
“That was the thing that really surprised me the most,” said Stanley. The cure for bone loss became a poison for the immune response.
The anti-PD-L1 immunotherapy worked by eliminating neutrophils being pumped by the hijacked skull marrow, but osteoporosis drugs interfered with this process.
It appeared the osteoclasts, while physically destructive to the skull, were playing a complex, dual role. By chemically inhibiting them to save the bone, the researchers inadvertently protected inflammatory neutrophils from being pumped out of the skull marrow due to GBM.
This finding provides a potential molecular explanation for the failure of other targeted therapies in this disease subtype. As Stanley noted, previous attempts to block the CSF-1 receptor, which regulates both macrophages and osteoclasts, also failed in GBM in general.
“Maybe there’s something about the microenvironment of the tumor that’s changing the response here,” Stanley said, pointing to a “unified theory” where blocking this pathway — whether via RANKL or CSF-1 — consistently unleashes the tumor’s myeloid helpers.
Stanley suggested the next step is to find a way to specifically activate the antitumor variety without disrupting the delicate balance of bone remodeling.
For Eskandar, a neurosurgeon who sees the human impact of GBM firsthand, the clinical implications create what seems like an impossible dilemma: a patient on immunotherapy who develops severe osteoporosis.
The timing couldn’t be more relevant. Patients with GBM are routinely prescribed high-dose corticosteroids like dexamethasone to manage cerebral edema, and a ubiquitous side effect is iatrogenic osteoporosis. The standard of care for steroid-induced osteoporosis is administration of bisphosphonates or RANKL inhibitors— exactly the drugs this study suggests may be harmful.
This hypothesis finds support in retrospective data. A 2013 clinical trial tested zoledronic acid to treat osteoporosis in patients with primary malignant glioma but showed poor survival outcomes and was largely considered a failure. Overall, 27 of 59 (45.8%) patients survived to 6 months, and only 19 of 59 (32.2%) survived to 12 months. By comparison, contemporaneous standard-of-care GBM trials reported that 255 of 309 (82.5%) patients survived to 6 months and 192 of 309 (62.1%) survived to 12 months. The trial was terminated early due to concerns about rapid tumor progression.
At the time, the mechanism for these poor outcomes was unknown. But this study provides a possible retrospective mechanistic explanation for that failure.
So, does this make treating GBM trickier than before? Do these results raise doubt, demoralizing clinicians rather than empowering them? Eskandar doesn’t think so.
“I think knowledge is power,” he said. “Knowing that [osteoporosis drugs] might be counterproductive is a net gain.”
Looking ahead, Eskandar sees only one viable path forward for GBM. “It’s not going to respond to any one agent or therapy. It’s just not going to happen,” he said. “Relying on the immune system to help us is the only way forward.”
Keeping the Band Together
After 15 years, Behnan has finally found the right environment, a place where engineers, computational experts, and neurosurgeons collaborate to validate findings.
“To cure glioblastoma, we need to put the efforts of those different fields together,” Behnan said. “We need the immunologist, the vaccine biologist. We need them to work together with the neurosurgeons to cure this disease.”
Eskandar agrees. “I think the biggest misconception is that [glioblastoma is] kind of hopeless,” he said. “It’s very complex, and it’s very challenging, but there are definitely opportunities.”
Part two of their study is just beginning. For Behnan, that thinning skull she felt 15 years ago has proven to be not just the first clue in unraveling one of oncology’s most vexing puzzles, but a reminder that transformative discoveries often begin with a scientist who refuses to let go of what her hands tell her is true.
The study was supported by the National Institutes of Health and the National Institute of Neurological Disorders and Stroke grants and the Marie Skłodowska-Curie Actions Global Fellowship. The funders had no role in study design, data collection and analysis, the decision to publish, or the preparation of the manuscript. The study authors declared no competing financial interests.