ISSUE 014 · Oct 16, 2026 · 9 min read

What if every treatment briefly opened the wall blocking brain-tumor drugs?

In recurrent glioblastoma, transiently open the blood–brain barrier with low-intensity pulsed ultrasound and microbubbles, then compare against the same chemotherapy alone to test whether higher drug exposure becomes longer progression-free survival.

The issue in one sentence

Diagram 01

The proof chain from barrier opening to patient benefit

01Recurrent tumor→

02Microbubble infusion→

03Pulsed ultrasound→

04Transient opening→

05Drug exposure→

06Tumor response→

07Function · survival

ANM proposed model. MRI-visible opening and higher tissue drug concentration are intermediate endpoints. A same-drug control must connect them to tumor control, neurotoxicity, and survival.

01Clinical problem · Evidence

A potent drug is not a treatment if it cannot reach infiltrating tumor cells.

Recurrent glioblastoma infiltrates beyond the resection margin and has few effective options. Although the barrier may be disrupted in the tumor core, the blood–brain barrier around infiltrating disease limits exposure to many systemic agents. Drug failure and delivery failure become confounded in trials.

02Technology stack · Proposal

Bind ultrasound, microbubbles, MRI, and infusion into one delivery window.

At repeat surgery, place a multi-emitter implant in the cranial window. Each chemotherapy cycle pairs intravenous microbubbles with low-intensity pulsed ultrasound. MRI confirms opening, while passive acoustic and physiologic monitoring guard against unsafe cavitation. Infuse chemotherapy in a prespecified sequence and interval.

03Research question · Hypothesis

Does higher drug exposure actually improve tumor control?

In first-recurrence glioblastoma after standard options are exhausted, does adding ultrasound-mediated barrier opening to the same albumin-bound paclitaxel–carboplatin regimen improve six-month progression-free survival without unacceptable severe neurotoxicity or hemorrhage?

04First study · Proposal

Fourteen centers, 180 people, a same-drug phase 2b trial.

Randomize adults with first recurrence, repeat-surgery eligibility, and KPS ≥70 in a 1:1 ratio. Both arms receive identical doses and cycles of albumin-bound paclitaxel plus carboplatin; only the intervention arm receives per-cycle barrier opening through an intraoperatively placed multi-emitter implant. The primary endpoint is six-month progression-free survival by blinded central RANO 2.0 review. Also measure overall survival, cognition, KPS, steroid dose, MRI opening volume, surgical and device complications, and grade ≥3 neurotoxicity.

05Defensible data · Hypothesis

The moat is not the device—it is where opening produced exposure and outcome.

Link emitter-level acoustics, microbubble dose, MRI permeability maps, spatial drug concentrations in repeat-surgery tissue, tumor and immune microenvironment, RANO trajectories, cognition, and toxicity. The same-drug control is what separates delivery effect from drug sensitivity.

06Venture thesis · Assumption

The first customers are neuro-oncology centers and CNS drug developers.

For hospitals, bundle implantation and treatment training, per-cycle barrier opening, MRI and acoustic quality control, and outcomes audit per treatment episode. For pharma, co-develop delivery–pharmacokinetic trials for CNS candidates limited by brain exposure. Per-patient revenue and drug-rescue partnerships remain hypotheses until survival benefit and cross-center reproducibility are shown.

Evidence map

The gap between delivery success and treatment success

3.7×

Mean increase in albumin-bound paclitaxel concentration in sonicated brain tissue in a 17-person phase 1 trial

5.9×

Mean increase in carboplatin concentration in sonicated brain tissue in the same pharmacokinetic analysis

33 people · 90 sessions

Repeated sonication exposure in a 2024 single-arm phase 1/2 trial

No control arm

Reported 14.0-month median overall survival cannot establish treatment effect

Evidence

The barrier opened in humans; longer survival is not yet established

Established: in a 17-person phase 1 study published in 2023, opening after repeated low-intensity pulsed ultrasound with microbubbles diminished over about one hour. Mean drug concentration in sonicated versus non-sonicated brain was 3.7 times higher for albumin-bound paclitaxel and 5.9 times higher for carboplatin. Transient grade 1–2 headache was common; one grade 3 encephalopathy occurred in cycle one at 260 mg/m². In a 2024 single-arm phase 1/2 study, 33 people underwent 90 monthly sessions, MRI opening was seen at 90% of emitters, and no dose-limiting toxicity occurred; grade 3 presyncope, fatigue, wound infection, and connector pain were reported. Median overall survival of 14.0 months in the 12 people given carboplatin immediately before sonication is a nonrandomized exploratory result. Clinical benefit, cost-effectiveness, and drug generalizability therefore remain venture assumptions pending registered follow-up results and a randomized same-drug comparison.

17-person pharmacokinetic phase 1 33-person repeated-opening phase 1/2 Follow-up phase 2 registry SonoCloud-9 registry FDA focused-ultrasound PMA boundary

Failure conditions and safety, ethics, and regulatory risks

Clinical failure condition

Do not scale unless the prespecified six-month progression-free-survival target is met with a hazard ratio of 0.70 or lower and overall-survival and function trends are directionally consistent. MRI opening and tissue drug concentration cannot substitute for patient benefit.

Hemorrhage, edema, and neurotoxicity

An independent DSMB pauses the study if symptomatic intracranial hemorrhage exceeds 3% or opening-attributable grade ≥3 neurotoxicity exceeds 10%. Use real-time acoustic stop thresholds, pre/post MRI, and 24-hour neurologic observation.

Surgical and device harm

Independently adjudicate infection, skin breakdown, connector pain, reoperation, and explantation. Do not perform craniotomy solely for device placement in patients without a clinical indication for repeat surgery.

Confusing delivery with drug activity

Opening is futile when the tumor resists the drug. Use a same-drug control, spatial pharmacokinetics, and prespecified sensitivity analyses to distinguish a well-delivered ineffective drug.

Access and consent ethics

Dependence on tertiary surgery, MRI, and repeated visits can widen geographic and income disparities. Measure travel and caregiver costs, and test understanding separately to limit therapeutic misconception.

Regulatory and business boundary

FDA P150038 authorizes thermal thalamotomy for essential tremor; it does not authorize blood–brain barrier opening or glioblastoma drug delivery. The implant–microbubble–drug combination may require a separate IDE and combination-product strategy. Stop commercialization if center-level reproducibility is poor or total per-treatment cost is not proportional to clinical value.

Bibliography

References

Established facts are separated from research and venture hypotheses that still require validation.

01
Sonabend AM, Gould A, Amidei C, et al. Repeated blood–brain barrier opening with an implantable ultrasound device for delivery of albumin-bound paclitaxel in patients with recurrent glioblastoma: a phase 1 trial. Lancet Oncol. 2023;24(5):509–522.
A phase 1 study of 17 people and 68 sessions showing barrier opening and higher brain drug concentration—not randomized clinical efficacy.PMID 37142373 · NCT04528680
02
Carpentier A, Stupp R, Sonabend AM, et al. Repeated blood–brain barrier opening with a nine-emitter implantable ultrasound device in combination with carboplatin in recurrent glioblastoma: a phase I/II clinical trial. Nat Commun. 2024;15:1650.
A single-arm study of 33 people and 90 sessions supporting repeated-opening feasibility and safety signals, without a survival comparator.PMID 38396134 · NCT03744026
03
ClinicalTrials.gov. Ultrasound-based blood-brain barrier opening and albumin-bound paclitaxel plus carboplatin in recurrent glioblastoma.
Registry for the follow-up drug-combination study; registration and conduct do not establish benefit.NCT04528680
04
ClinicalTrials.gov. Safety and efficacy of transient opening of the blood-brain barrier with the SonoCloud-9.
Primary registry record for the design and endpoints of the single-arm phase 1/2 study reported in 2024.NCT03744026
05
U.S. Food and Drug Administration. ExAblate Model 4000 Type 1.0 System (ExAblate Neuro). Premarket Approval P150038.
Confirms that the approved indication for this MR-guided focused-ultrasound system is unilateral thalamotomy for medication-refractory essential tremor—not barrier opening or tumor therapy.P150038 · Decision Jul 11, 2016

The next issue publishes every day at 9 AM KST.All issues