Estimation of kidney doses from [¹⁷⁷Lu]Lu-DOTA-TATE PRRT using single time point post-treatment SPECT/CT

Dosimetry after [¹⁷⁷Lu]Lu-DOTA-TATE therapy can be demanding for both patients and the clinical service due to the need for imaging at several time points. In this work we compare three methods of single time point (STP) kidney dosimetry after [¹⁷⁷Lu]Lu-DOTA-TATE therapy with a multiple time point (MTP) dosimetry method.

Method 1 (MTP): Kidney doses were calculated from 31 patients including 107 therapy cycles. Post-therapy SPECT images were acquired on day 0, 4 and 7 along with a CT scan on day 4. A mono-exponential fit was used to calculate kidney doses using cycle specific data. Method 2 (Consistent effective half-life): The effective half-life \(\left({\text{T}}_{\text{e}\text{f}\text{f}}\right)\) calculated in cycle 1 was assumed consistent for subsequent cycles of therapy and the activity scaled using a single day 3–5 SPECT/CT. Methods 3 and 4 (Hänscheid and Madsen approximations): The Hänscheid approximation and Madsen approximation were both evaluated using a single SPECT/CT acquired on day 0, 4 and 7. All STP methods were compared to the MTP method for accuracy.

Using the MTP method, mean right and left kidney doses were calculated to be 2.9 ± 1.1 Gy and 2.8 ± 0.9 Gy respectively and the population \({\text{T}}_{\text{e}\text{f}\text{f}}\) was 56 ± 13 h. For the consistent \({\text{T}}_{\text{e}\text{f}\text{f}}\) , Hänscheid and Madsen methods, the percentage of results within ± 20% of MTP method were 96% (n = 70), 95% (n = 80) and 94% (n = 80) respectively.

All three single time point methods had > 94% of results within ± 20% of the MTP method, however the consistent \({\text{T}}_{\text{e}\text{f}\text{f}}\) method resulted in the highest alignment with the MTP method and is the only method which allows for calculation of the patient-specific \({\text{T}}_{\text{e}\text{f}\text{f}}\) . If only a single scan can be performed, day 4 is optimal for kidney dosimetry where the Hänscheid or Madsen approximation can be implemented with good accuracy.