Glioblastoma targeted therapy still runs into biology, not just drug design
Glioblastoma remains the most lethal primary brain tumour in adults, and the gap between molecular knowledge and therapeutic gain is still wide. Deep genomic profiling has clarified many of the alterations that drive these tumours, yet targeted treatment has produced only limited benefit in routine practice. The problem is not a lack of candidate pathways. It is the collision of intratumoural heterogeneity, pathway redundancy, cellular plasticity and the stubborn barrier posed by the central nervous system itself.
Only a narrow subset of patients has seen clear progress from targeted approaches, and that success has largely been confined to tumours with either BRAF V600E mutations or NTRK fusions. Those alterations are biologically cleaner than the sprawling genomic chaos typical of glioblastoma, making them easier to match with a drug. For the broader glioblastoma population, the tumour often behaves like a moving target: one clone may be susceptible while another resists, and the dominant pathway at diagnosis may no longer matter after treatment pressure reshapes the disease.
The practical obstacle starts with delivery. Even a well-chosen inhibitor can fail if it cannot reach therapeutic concentrations inside the brain. Limited blood-brain penetration remains a central reason that many molecular strategies underperform in glioblastoma, where the anatomy of the disease and the physiology of the brain work against drug exposure. Strategies designed to improve access include focused ultrasonography, convection-enhanced delivery, efflux avoidance and chemical modification of compounds to improve penetration. Each of these approaches aims at the same bottleneck: getting enough active drug to the tumour at the right time.
Heterogeneity turns one tumor into many therapeutic problems
A single glioblastoma can contain multiple biologically distinct subclones. Some carry truncal drivers that are shared across the tumour mass, while others reflect later branching evolution and may be present only in limited regions. When therapy is matched to a subclonal alteration, the response may be partial and short-lived because other compartments of the tumour remain untouched. This is one reason multiregional sampling has gained prominence. A biopsy from one site may not represent the disease in full, and a treatment plan based on that sample can miss the dependencies that matter most.
Cellular plasticity deepens the problem. Glioblastoma cells can shift state under therapeutic pressure, changing phenotype without losing malignant potential. Pathway redundancy provides another escape route: when one signalling route is blocked, parallel circuitry can sustain growth, survival or invasion. This creates a situation in which suppression of one oncogenic axis may slow disease only briefly before the tumour leans on another.
The review places special emphasis on extrachromosomal DNA dynamics, a feature that helps explain oncogene amplification and resistance. Extrachromosomal DNA can amplify oncogenic signals in a flexible, rapidly adaptable form, allowing the tumour to alter its molecular strength in response to therapy. In a disease already defined by genomic instability, that mobility matters. It gives glioblastoma a mechanism for accelerating resistance without requiring a stable chromosomal rewrite.
Target classes are expanding, but the proof remains uneven
Targeted therapy in glioblastoma now spans several major classes, including receptor tyrosine kinases, intracellular signalling proteins, cell-cycle dysregulation and synthetic-lethal vulnerabilities. The therapeutic menu has also expanded to include strategies aimed at genome integrity, telomeres, epigenetic modulation and tumour-neural circuitry. These directions reflect a broader shift in thinking: glioblastoma is not only a proliferative disease, but also one shaped by DNA repair stress, chromatin regulation and its interactions with surrounding neural tissue.
The most credible path forward may not be a single universal drug class. Antibody-drug conjugates and theranostics are emerging as ways to improve selectivity, deliver payloads more efficiently and align imaging with treatment. Their appeal lies in precision at two levels at once: molecular recognition and local drug release. If those platforms can be adapted to the CNS environment, they could bypass some of the weaknesses that have hindered small-molecule strategies.
A one-size-fits-all model has already failed here.
Trial design now matters as much as target selection
More informative clinical trials are part of the therapeutic answer. Novel trial designs can test agents faster, broaden eligibility when biologically justified and reduce the delay between molecular insight and clinical readout. Surgical window-of-opportunity designs are particularly relevant in glioblastoma because they can measure pharmacodynamic effects in real tumour tissue rather than infer activity from indirect endpoints alone. That matters in a disease where brain penetration, intratumoural distribution and target suppression are all highly uncertain until tissue is examined directly.
Liquid biopsy-based monitoring is also gaining importance. Repeated tissue sampling is difficult in the brain, and disease evolution can outpace standard imaging. Circulating biomarkers may help track resistance, clonal shifts and treatment effect over time, although the review stops short of claiming that they are ready to replace tissue-based assessment. Their promise lies in longitudinal surveillance, especially when the tumour changes faster than conventional decision-making can keep up.
Pathway-based glioblastoma classification and master kinase mapping offer another conceptual shift. Rather than defining the disease only by static mutations, these approaches attempt to align drugs with functional tumour states. In a tumour where the same genetic driver may behave differently across regions or under treatment pressure, pathway activity may be more actionable than a single alteration in isolation. This is a more disciplined way to think about precision oncology in the brain: not merely whether a target exists, but whether it is truly driving the disease at that moment.
Low-grade glioma gains have not translated to glioblastoma
The contrast with low-grade glioma is stark. Several targeted drugs have already secured regulatory approval in that setting, showing that molecular precision can work when the biology is more tractable and the therapeutic target is more stable. Glioblastoma has not followed that path. The difference is rooted in disease structure as much as drug class: in low-grade glioma, a smaller set of driver alterations can dominate tumor behavior, whereas glioblastoma is shaped by broader heterogeneity, deeper resistance networks and more severe delivery constraints.
That is why progress in glioblastoma has been incremental rather than impactful. Success will likely require combining better brain delivery, more accurate tumour profiling, stronger prioritization of truncal dependencies and treatment platforms that can adapt to changing tumour states. The review makes clear that no single tactic is enough on its own. The field will need multiple coordinated solutions, and even then the central uncertainty will remain whether a target can be reached, suppressed and sustained inside the human brain without the tumour finding another way around it.