[Article] Impact of Different Models based on Blood Samples and Images for Bone Marrow Dosimetry after 177Lu-labeled Somatostatin-Receptor Therapy

Impact of different models based on blood samples and images for bone marrow
dosimetry after 177Lu-labeled somatostatin-receptor therapy

Delphine Vallot1 , Séverine Brillouet1, Séléna Pondard1, Lavinia Vija1, Jean‑Sébastien Texier1, Lawrence Dierickxand Frédéric Courbon1
1 Oncopole Claudius Regaud, Toulouse, France


Background: Peptide receptor radionuclide therapy with 177Lu‑DOTATATE is a recog‑ nized option for treating  neuroendocrine tumors and has few toxicities, except for the kidneys and bone marrow. The bone marrow dose is generally derived from a SPECT/CT image‑based method with four timepoints or from a blood‑based method with up to 9 timepoints, but there is still no reference method. This retrospective single‑center study on the same cohort of patients compared the calculated bone marrow dose administered with both methods using mono, bi‑ or tri‑exponential models. For the image‑based method, the dose was estimated using Planetdose© software. Pearson correlation coefficients were calculated. We also studied the impact of late timepoints for both methods.

Results: The bone marrow dose was calculated for 131 treatments with the blood‑ based method and for 17 with the image‑based method. In the former, the median absorbed dose was 15.3, 20.5 and 28.3 mGy/GBq with the mono‑, bi‑ and tri‑exponen‑ tial model, respectively. With the image‑based method, the median absorbed dose was 63.9, 41.9 and 60.8 with the mono‑, bi‑ and tri‑exponential model, respectively. Blood samples after 24h post‑injection did not evidence any change in the absorbed bone marrow dose with the bi‑exponential model. On the contrary, the 6‑day post‑ injection timepoint was more informative with the image‑based model.

Conclusion: This study confirms that the estimated bone marrow dose is signifi‑ cantly lower with the blood‑based method than with the image‑based method. The blood‑based method with a bi‑exponential model proved particularly useful, with‑ out the need for blood samples after 24h post‑injection. Nevertheless, this blood‑based method is based on an assumption that needs to be more validated. The important difference between the two methods does not allow to determine the optimal one to estimate the true absorbed dose and further studies are necessary to compare with biological effects.