Feasibility of single time point dosimetry during 177Lu-PSMA-617 treatments of prostate cancer

Sammanfattning

Introduction: Standard dosimetry methods for radionuclide therapies re- quire imaging at multiple time points post injection. An alternative method has been presented in the literature that uses previously acquired pharma- cokinetic data to create so-called dose factors. Multiplying the dose factors with an activity concentration (Bq/ml) taken from a single time point mea- surement results in an estimated absorbed dose (Gy). The dose factors were presented for 177Lu-PSMA-617 therapies for patients with metastatic cas- trate resistant prostate cancers. Before the new method (1-point method) is applied for clinical use it needs careful validation by comparing it to the standard dosimetry method (2-point method). In this study dosimetry was performed on kidneys, salivary glands and tumors. The resulting absorbed doses were compared between methods to evaluate the applicability of the 1-point method. Method: Dosimetry was performed on 5 patients, using SPECT/CT images acquired 24 and 96 h post injection. Images were reconstructed with OSEM, using attenuation correction, resolution recovery and model based scatter correction (ESSE). Partial volume correction was performed by convolving the VOI with a Gaussian point spread function and determining the ratio of VOI counts with and without effects of limited spatial resolution. Two dosimetry methods were used, the 2-point method that incorporated images from both imaging time points and the 1-point method. Absorbed doses using the 1-point method were calculated for the two different imaging time points separately. Salivary glands and kidneys were delineated manually, tumor delineation was made using an automated method based on differ- ence of Gaussians. Tumors were categorised into soft tissue and bone lesions based on their mean density and corresponding coefficient of variation, using support vector machines for classification. Results: Mean absorbed dose (standard deviation) per injected activity to kidneys was 0.56 +- 0.14 (2-point method), 0.59 +- 0.11 (1-point method 24h-image) and 0.44+-0.16 (1-point method 96h-image) Gy/GBq, to sub- mandibular glands it was 0.36 +- 0.15 (2-point method), 0.34 +- 0.14 (1-point method 24h-image), 0.23 +- 0.12 (1-point method 96h-image) Gy/GBq, and for parotid glands 0.31 +- 0.15 (2-point method), 0.32 +- 0.13 (1-point method 24h-image) and 0.19 +-0.14 (1-point method 96h-image) Gy/GBq. Categori- sation of tumor type using support vector machines on test data correctly categorised 50 out of 51 tumors. Conclusion: The 1-point method for performing dosimetry was deemed sufficiently accurate based on the low mean difference in absorbed dose com- pared to the 2-point method. This applies to images taken 24 h post in- jection, imaging at 96 hours resulted in poor agreement with the 2-point method.