Purpose
Accurate and precise renal dosimetry during 177Lu‐labeled prostate‐specific membrane antigen (PSMA) radioligand therapy is crucial for therapy decisions. Sampling schedules for estimating the necessary time‐integrated activity coefficients (TIACs) are not optimized and standardized for clinical practice. Therefore, a simulation study to determine optimal sampling schedules (OSSs) was performed on 13 virtual 177Lu‐PSMA I&T therapy patients.
Method
A total of 880 clinically feasible sampling schedules for planar imaging (three time points) were investigated. To simulate the hybrid planar/SPECT method, an additional quantitative SPECT/CT measurement following one planar image was considered. For each sampling schedule and patient, the activity values were generated separately. Measurement noise was modeled by drawing random numbers of log‐normal distributions. The used fractional standard deviations (FSD) differed depending on the imaging modality. For activity values assigned to planar imaging, systematic noise between 25% and 75% of the total noise was simulated. After fitting with a mono‐exponential function, the root‐mean‐squared errors of the deviations of the simulated TIACs from the ground truth for 1000 replications were used to determine the OSS. The uncertainties of the TIACs and renal dose coefficients were estimated.
Results
For the hybrid planar/SPECT method, OSSs were determined to be (3–4, 72–76, 124–144) h post injection (p.i.) with the quantitative SPECT/CT scan shortly after the second measurement. The accuracy and precision of the determined TIACs were in the range of (−3.0 ± 6.2)% and (−1.0 ± 6.5)%. This precision was improved by a factor 2–3 compared to dosimetry based on planar images only. Similar results were obtained for the renal dose coefficients. The virtual patients' renal dose coefficients were (0.68 ± 0.24) Gy/GBq indicating that a population‐based method yields an uncertainty of 35%.
Conclusions
Dosimetry based on the hybrid planar/SPECT method with OSS outperforms dosimetry based on planar images. The high variability in dose coefficients between the virtual patients demonstrates the need for individualized dosimetry.