Berkeley Lab Clears a Bottleneck in Targeted Cancer Radiation

Targeted alpha therapy has had a measurement problem. The drugs are blunt little atom bombs: a single decay of actinium-225 sends four alpha particles tearing through a cancer cell's DNA, while leaving most nearby healthy tissue alone. But the same physics that makes the drug lethal also makes it hard to see. Standard PET scanners were built to spot the back-to-back photons that a positron emits. Actinium-225 does not emit positrons. The radiation it does produce is a cascade of fast gamma rays from a daughter isotope, thallium-209, and no commercial scanner was reading them.

A team at Lawrence Berkeley National Laboratory has now shown that a clinical PET/CT, in research mode, can be taught to read that cascade. The technique is called time-of-flight cascade-gamma imaging, or TOF-CGI, and the first in-human image appears in a paper published October 1 in the Journal of Nuclear Medicine.

The trick is timing. When actinium-225 decays, its granddaughter thallium-209 fires two gamma photons almost simultaneously, 60 picoseconds apart, along a line that points back to the original decay. A PET scanner does not need the positrons to do useful work; it just needs two coincident photons, and these qualify. The harder problem is that the cascade energy is high and the photon energies are uneven, so a stock scanner's energy window rejects most of them. The Berkeley group retuned a Siemens Biograph Vision 600 PET/CT, widened the energy acceptance, and ran the data through a custom GPU-accelerated reconstruction that uses the picosecond-scale timing to localize each event.

The team tested the system on vials, on clinical-size phantoms, and on one patient with prostate cancer who had been dosed with [225Ac]Ac-PSMA-617, a standard targeted alpha therapy that routes the isotope to prostate-specific membrane antigen. The resulting image showed activity inside a known prostate-bed tumor and a higher signal-to-noise ratio than the SPECT scan the patient had also received, according to the paper. The reported minimum detectable activity in a 20-minute vial scan was 5.5 kilobecquerels; phantom sensitivity came in at 26.0 counts per second per megabecquerel, roughly an order of magnitude better than SPECT, the authors wrote.

The patient result is a single data point, not a clinical proof. The setup is also research hardware: a Siemens Biograph Vision 600 in investigational mode, a custom reconstruction pipeline, and anisotropic resolution of about 20 millimeters transversely and 42 millimeters axially, coarse by clinical PET standards. The numbers above are experiment-specific and do not generalize to every scanner in every hospital.

What does generalize is the concept. A research scanner that can finally localize actinium-225 in tissue is the piece that was missing for an otherwise fast-moving class of drugs. Actinium-225 emits four alpha particles per decay and has a roughly 10-day half-life, long enough to manufacture, ship, and dose a patient, and short enough that the radiation does not linger after treatment. Those properties, plus the four alphas per atom, are why several PSMA-targeted and other alpha therapies are already in clinical trials.

The catch is that, without imaging, clinicians had no way to verify where those alphas actually landed. Actinium's daughter isotopes can drift away from the original targeting molecule, redistribute through the body, and accumulate in the kidneys. A method that can map the cascade means researchers can quantify that redistribution and dose more carefully.

"This work is a step closer to seeing where the drug goes," said Caravaca in the lab's announcement.

The paper also lists collaborators at UCSF and at Siemens Medical Solutions USA, and Berkeley Lab says the technology is available for licensing. None of that answers the two open questions the imaging advance does not solve. The first is supply: actinium-225 has to be made in a cyclotron or extracted from legacy sources, and that pipeline is the gating constraint on the entire class of drug, not just on this measurement. The second is validation: a single prostate-cancer image is a feasibility demonstration, not a clinical performance claim. Wider deployment, the authors note, will require longer scans, lower doses, and trials on whole-body scanners, including long-axial-field-of-view PET systems that are now entering nuclear medicine departments.

The next paired test, on a long-axial-field-of-view system, is what turns a research image into a clinical tool.