Patient-specific dosimetric workflow with a reduced number of time-points for 177Lu treatments

International audienceAim/Introduction: Patients with unresectable neuroendocrine tumors can be treated by internal radiation therapy administrating 7.4 GBq of Lu177-labeled somatostatin analog. Image-based absorbed dose estimation to organs requires several time-points acquisitions but their number is limited by clinical and logistical constraints. The goal of this work was to propose an adaptive dosimetric workflow for estimating absorbed doses from a reduced number of acquisitionacquisitionMaterials and Methods: Thirteen patients with neuroendocrine tumors received four injections of 7.4 GBq of 177Lu-DOTATATE except one patient because of haematological toxicity. Three SPECT/CT acquisitions were performed after the first cycle at 1h, 24h and 96h or 144h and a single acquisition after three latter cycles at 24h for estimating absorbed doses by kidneys, liver and intrahepatic tumors, spleen and bone marrow. Monte Carlo simulations were used to estimate dose rates for each acquisition to take into account self- and cross- doses. These dose rates were then integrated at the organ level (ODR). Time dose rate curves (TDC) were fitted with a tri-exponential function to respect the patients' physiology. For cycles with less than three acquisitions, three methods were proposed and evaluated. If the acquisition at 24h of cycle 1 was missing, the ODR was approximated by the next first ODR at 24h scaled according to the injected activities at each cycle. When only one acquisition was available, two cases were distinguished. Absorbed doses were estimated from the TDC of a previous cycle for the samepatient scaled to the ODR available. Otherwise, TDC of other patients were scaled to the ODR and their integrals were averaged to obtain absorbed doses as proposed by Jackson and al. (1) with time activity curvesResults: Estimated errors when the ODR at 24h was replaced was inferior to 15.9% except for one patient. With only one acquisition, the lowest error was obtained for acquisitions at 96h or 144h with two methods: less than 11% for liver and spleen and less than 20% for kidneys. Thesemethods are being tested for bone marrow dose estimation. Absorbed doses estimated with this workflow were: 2⼟.50.9 Gy (left kidney), 2.7⼟1.2 Gy (right kidney), 2.8⼟1.7 Gy (liver) and 3.4⼟1.6 Gy (spleen) all cycles taken together. Conclusion: The proposed workflow allows the estimation of organ doses from a reduced number of acquisitions for patients treated with 177Lu. References: (1) Jackson and al. JNM 202