Coupling target and radioactive emitter matters: a sub-tissue dosimetric study to identify optimal strategies

Abstract Background Radiopharmaceutical-therapy dosimetry depends on both radionuclide emission physics and the spatial distribution of activity. We evaluated how radionuclide identity and protein-defined subcellular localization influenced cancer-cell nuclear absorbed-dose distributions in one human breast-cancer region of interest (ROI). Methods Multiplex immunofluorescence images from one breast-cancer ROI were segmented into 41,454 cells. Protein-expression maps for nuclear, nuclear-membrane (NM), cytoplasmic, plasma-membrane, and extracellular-matrix (ECM) locations were used as source maps in TOPAS/Geant4 simulations of 161 Tb, 177 Lu, 212 Pb, and 225 Ac. Endpoints were the similarity-dose ( D sim,99 ), simulated surrogate IC 50 and IC 90 threshold-crossing doses over hypothetical per-cell thresholds of 0.2–5.0 Gy, and the percentage of cancer nuclei above the submitted 0.5-Gy threshold for 161 Tb/ 177 Lu or 0.1-Gy threshold for 212 Pb/ 225 Ac. The alpha-emitter IC analysis used the submitted whole-chain factor of 5; coverage used unmultiplied scored dose with the factor-five-equivalent 0.1-Gy threshold described in the Methods. Results For NM localization, D sim,99 was 0.030 Gy for 161 Tb, 0.074 Gy for 177 Lu, 0.412 Gy for 212 Pb, and 0.580 Gy for 225 Ac. Across locations, D sim,99 ranged from 0.030 to 0.074 Gy for 161 Tb, 0.063 to 0.089 Gy for 177 Lu, 0.412 to 1.023 Gy for 212 Pb, and 0.580 to 1.116 Gy for 225 Ac. ECM locations generally produced the largest simulated surrogate IC slopes, although the magnitude and endpoint dependence varied by isotope and some mixed-chain ECM-C IC 90 fits were numerically unstable. At nuclear targeting, the descriptive cancer-nucleus dose fractions were 20.25% for 161 Tb, 15.76% for 177 Lu, 18.32% for 212 Pb, and 18.82% for 225 Ac. Beta/Auger coverage used the submitted 0.5-Gy threshold, whereas the alpha-emitter panels used the submitted 0.1-Gy threshold. Conclusion Radionuclide identity and modeled subcellular localization jointly influenced the submitted nuclear-dose summaries, and no universal optimal compartment was established across endpoints. These single-ROI computational results are hypothesis-generating and require particle-resolved transport, pharmacokinetic, radiobiological, normal-tissue, and multi-patient validation before clinical application.