← Back to Home

Map the Brain Tumor’s Blood Supply Before Deciding Where the Drug Should Enter

An MD Anderson team used AI to help identify multiple arteries supplying malignant brain tumors, in an attempt to enable intra-arterial therapy to cover more of the tumor. Preliminary results from three patients improved the drug-delivery range but cannot yet answer questions about efficacy and safety.

By SURL BioNews

Treating malignant brain tumors is challenging not only because effective drugs must be found, but also because those drugs must be able to reach the entire lesion. Tumors are often supplied by multiple small arteries. If infusion is performed through only one of them, large areas may remain uncovered even when the catheter has been positioned close to the tumor.

A team at The University of Texas MD Anderson Cancer Center therefore developed an AI-assisted method to identify multiple arterial branches supplying the tumor before superselective intra-arterial infusion. The study, titled “From Single Pedicle to Whole-Tumor Coverage,” was presented at the 23rd annual meeting of the Society of NeuroInterventional Surgery.

In this workflow, AI creates an individualized map of the patient’s tumor blood supply, and physicians then use advanced imaging to confirm the identification results. The catheter is subsequently advanced into the different tumor-feeding arteries, with the infusion volume adjusted according to each vessel’s contribution to tumor blood flow. The goal is to expand coverage of the lesion while reducing the flow of drugs into unintended brain tissue.

The available data currently come from three patients with malignant brain tumors. The team identified multiple tumor-feeding arteries in each patient. According to results presented at the meeting, multi-pedicle infusion covered more than 85% of each tumor, compared with less than 65% coverage from proximal infusion through a single pedicle. The researchers also said the procedure could be performed again two weeks later.

These figures primarily demonstrate the range of drug delivery, rather than treatment efficacy. The publicly available information does not specify the patients’ tumor types, the therapies infused, or the follow-up duration, nor does it provide clinical outcomes such as tumor shrinkage, survival, or neurological function. Experience in only three patients is also insufficient to characterize rare complications or establish the safety of repeated procedures.

Validation of the AI itself also remains incomplete. The current report does not describe the model’s training data, its accuracy in identifying blood vessels, or whether it was compared with neurointerventional physicians in a blinded assessment. It is also unclear whether the results can be reproduced with imaging equipment and patient populations at different hospitals. Before proceeding to more formal clinical trials, the researchers must not only demonstrate that increased coverage genuinely improves patient outcomes, but also establish standardized procedures for image interpretation, human review, and infusion dosing. At this stage, this is a more detailed roadmap for drug delivery, not yet evidence that brain tumor treatment outcomes have improved.

References

  1. Society of NeuroInterventional Surgery via EurekAlert
  2. News-Medical.Net