Somewhere in an operating room this spring, a neurosurgeon did something that has never been done before: while removing a patient’s brain tumor, the surgical team threaded a thin cannula into the remaining tissue and, guided by real-time MRI, pumped in millions of copies of an engineered virus designed to destroy glioblastoma cells from the inside out.
The therapy is called TGX-007, and it is now being given to the first patients enrolled in a Phase I/II clinical trial known as ADePT. It marks a new chapter in the effort to treat the deadliest common brain cancer with tools borrowed from gene therapy, and it arrives at a moment when a separate viral approach has already shown that injecting engineered viruses directly into glioblastomas can wake up the immune system in ways that chemotherapy and radiation never have.
Why glioblastoma has resisted almost everything
Glioblastoma is the most aggressive primary brain tumor in adults. Roughly 12,500 Americans are diagnosed each year, and the standard combination of surgery, radiation, and the chemotherapy drug temozolomide, known as the Stupp protocol, has not meaningfully changed since 2005. Median survival remains around 15 months. The tumor grows tentacle-like projections that infiltrate healthy brain tissue, making complete surgical removal nearly impossible, and it suppresses the local immune environment so effectively that the body’s own defenses rarely mount a sustained attack.
Drugs delivered through the bloodstream face an additional obstacle: the blood-brain barrier blocks most large molecules from reaching the tumor at therapeutic concentrations. That barrier is the central reason researchers have turned to direct injection during surgery, bypassing the problem entirely by placing the treatment inside the tumor while the skull is already open.
How TGX-007 works
TGX-007 is built on an adeno-associated virus serotype 1 (AAV-1) vector, a type of virus that has been hollowed out so it cannot replicate. Instead of viral genes, it carries two therapeutic payloads.
The first is HSV-tk, a gene encoding the herpes simplex virus thymidine kinase enzyme. Once the gene is expressed inside dividing tumor cells, doctors administer the antiviral drug ganciclovir intravenously. HSV-tk converts ganciclovir into a toxic metabolite that kills the cells expressing it, functioning as a molecular suicide switch. Because the enzyme is only active in cells that take up the vector and are actively dividing, healthy neurons, which rarely divide, are largely spared.
The second payload is interleukin-12 (IL-12), an immune-signaling protein that recruits and activates T cells. The goal is to turn the tumor site into a beacon for the immune system, drawing killer T cells into a microenvironment that glioblastoma normally keeps immunologically cold.
Delivery happens through a technique called convection-enhanced delivery, or CED. Rather than simply injecting the virus and hoping it diffuses outward, CED uses positive pressure through a cannula to push the vector through tumor tissue over a wider area. MRI guidance allows the surgical team to monitor distribution in real time, aiming to cover the infiltrative margins where microscopic disease often lingers after the visible tumor has been removed.
Evidence from a related but different viral approach
TGX-007 is not the first engineered virus to be injected into a glioblastoma during surgery. A separate program, CAN-3110 (also called rQNestin34.5v.2), completed a first-in-human Phase 1 trial led by E. Antonio Chiocca and colleagues at Brigham and Women’s Hospital and Harvard Medical School. That trial treated 41 patients with recurrent high-grade glioma. CAN-3110 is an oncolytic herpes simplex virus, meaning it is engineered to replicate selectively inside tumor cells, physically bursting them open and spreading to neighboring cancer cells in a chain reaction.
The results, published in peer-reviewed form, showed that patients whose tumors exhibited strong immune activation after treatment, including expansion of specific T-cell clones and markers of cytotoxicity, survived longer than those without such responses. A subsequent mechanistic analysis by Chiocca and colleagues, published in Cell in 2025, linked persistent T-cell activity and cytotoxicity to a single intratumoral injection, suggesting the virus triggered lasting immune memory against glioblastoma cells.
The distinction between the two programs matters. CAN-3110 relies on viral replication to destroy tumor cells directly. TGX-007 uses a non-replicating vector as a delivery truck for its two gene payloads. One spreads and lyses; the other expresses foreign genes and waits for a systemic drug to trigger cell death. Whether a non-replicating AAV carrying IL-12 can produce the same depth of immune activation as a replicating oncolytic virus is an open and important question.
Still, the CAN-3110 data provide the strongest clinical evidence to date that direct viral delivery into brain tumors can reshape the local immune landscape. That evidence is part of the scientific rationale underpinning the ADePT trial.
What the ADePT trial has not yet shown
It is worth being precise about what is known and what is not. The ADePT trial is registered, has regulatory clearance to dose patients, and has begun enrollment. Its design, delivery method, and planned endpoints are described in the ClinicalTrials.gov registry and a conference abstract published in Neuro-Oncology.
But no patient outcome data have been published. No primary data on infusion volumes, real-time MRI cannula performance, or early adverse events have been publicly posted. The trial lists long-term vector genome detection and HSV-tk mRNA expression as planned endpoints, but results have not appeared. Without them, it is not possible to confirm whether the AAV-1 vector achieved spatially restricted gene expression or whether IL-12 production reached levels sufficient to recruit T cells in glioblastoma’s immunosuppressive microenvironment.
Safety questions are also unanswered. Convection-enhanced delivery carries its own risks, including catheter-related hemorrhage and off-target distribution of the vector into healthy brain tissue. IL-12, while potent as an immune activator, has caused severe systemic inflammation in other clinical contexts when expression was not tightly controlled. Whether the localized delivery method keeps IL-12 levels confined to the tumor bed is something only clinical data can resolve.
The CAN-3110 immune correlates, including TCR sequencing and cytotoxicity assays, have not been reported for TGX-007. Extrapolating one program’s immune findings to the other would be premature without matched data. The latest publicly available clinical update on CAN-3110 was published in late 2023, and no new efficacy or safety results from that trial have appeared since.
What intratumoral viral therapy still needs to prove
For the roughly 12,500 families who hear the word “glioblastoma” each year in the United States, the start of a trial like ADePT is both a reason for cautious hope and a reminder of how early this science remains.
The completed CAN-3110 Phase 1 trial, with its peer-reviewed survival data and immune profiling in 41 patients, offers the clearest picture so far of what intratumoral viral therapy can and cannot do. It demonstrated that a single injection can provoke sustained T-cell responses in some patients. It did not establish that every glioblastoma will respond similarly, or that the approach will outperform the current standard of care. The immune-survival link is encouraging, but it comes from a small, heterogeneous group of heavily pretreated patients. Larger, randomized studies are needed to determine whether oncolytic viral therapy changes long-term outcomes.
TGX-007 sits at an even earlier stage. Its dual-payload design raises specific questions that only clinical experience can answer: Will IL-12 expression stay localized enough to avoid systemic toxicity? Can the HSV-tk/ganciclovir system selectively eliminate transduced tumor cells without damaging surrounding neurons? Will convection-enhanced delivery reliably cover the infiltrative margins where microscopic disease persists?
Patients considering enrollment should understand that the primary goals of ADePT at this stage are to establish safety, define dosing, and map biological responses, not to guarantee therapeutic benefit. These are the necessary first steps before any therapy can advance to the larger trials that would determine whether it truly extends survival.
What the field has gained, though, is a proof of concept. The durability of immune activation seen with CAN-3110 suggests that even a single, well-designed intervention inside the tumor can change how the immune system perceives glioblastoma. TGX-007 is an attempt to build on that foundation with a more controlled platform, combining direct tumor killing with immune stimulation in a single surgical procedure. Whether it succeeds is a question that only the data from ADePT, once they arrive, can begin to answer.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
