Dosimetry Reconstruction in Radiopharmaceutical Therapy Using a Sparse Network of External γ-Sensors

Abstract

Radiopharmaceutical therapy (RPT) has demonstrated promise in the treatment of neuroendocrine and prostate cancer. Due to the highly varied biodistribution and non-homogeneity of total integrated dose across cancer patients, a system for real-time dosimetry based on continuous measurement is desirable to deliver sufficient dose for tumor ablation while preventing toxicity from off-target uptake by organs at risk (OAR). Single time point imaging (mostly SPECT)-based dosimetry offers a snapshot of the body-wide dose distribution at a given time point, but even single SPECT imaging is generally limited in availability, often leading to significant inaccuracies in estimating total integrated dose. Therefore, accurate personalized dosimetry in RPT is an unmet need and requires continuous dosimetry measurements of tumors and OARs across multiple half-lives of the therapeutic radiopharmaceutical. Using a priori knowledge of tumor and OAR locations from pretherapy imaging, we have developed a novel algorithm that utilizes a network of custom uncollimated, optical fiber-based γ-counting probes to isolate the real-time in vivo tumor and OAR uptake in 177Lu-PSMA-617 and 225Ac-MACROPA-YS5 therapy. The proposed system was successfully validated in athymic mice models bearing varying numbers of tumors from two human prostate cancer cell lines (PC3-pip, PC3-flu). Uncollimated γ counts using the developed probes were acquired outside of the mice for 10 minutes, starting at 0 hr, 6 hrs, 12 hrs, 24 hrs, and 48 hrs after the injection of 177Lu-PSMA-617. The percent injected activity per mL of tissue (%IA/mL) of each tumor and OAR was reconstructed at every time point and compared to the %IA/mL extracted from SPECT/CT immediately after the recording. The developed system's %IA/mL reconstruction in PC3-pip tumors, PC3-flu tumors, kidneys, and bladders is highly correlative with the %IA/mL extracted from state-of-art in vivo dosimetry techniques, with %IA/mL ranging from 0.1% to 160% assuming a 29.6 MBq 177Lu-PSMA-617 administration. The least squares linear regression fit between the reconstructed activity and the activity measured from SPECT/CT is given by Estimated %IA/mL = 0.91 x SPECT %IA/mL, with an R2 of 0.991, and Pearson's r of 0.9975. There is a nearly 1:1 mapping between the proposed model and SPECT/CT. A novel dose reconstruction algorithm for personalized dosimetry in RPT that utilizes a sparse set of external γ-counters and a priori knowledge of tumor and OAR locations was developed and validated in in vivo human prostate cancer murine models. The proposed system enables continuous dosimetry measurements of multiple tumors and OAR noninvasively, with high accessibility, high temporal resolution, and across multiple classes of ɑ and β-based RPT. Similar experiments using 225Ac-MACROPA-YS5 are ongoing and additional results will be reported.