Can ¹⁷⁷Lu-DOTATATE Kidney Absorbed Doses be Predicted from Pretherapy SSTR PET? Findings from Multicenter Data

Before performing ¹⁷⁷Lu-DOTATATE therapy for neuroendocrine tumors, somatostatin receptor (SSTR) PET imaging is currently used to confirm sufficient tumor SSTR expression, but it also has potential to be used to personalize treatment by predicting absorbed doses to critical organs. This study aims to validate the predictive capability of SSTR PET in anticipating renal absorbed dose in the first cycle of ¹⁷⁷Lu-DOTATATE using a multicenter dataset to analyze and derive insights from a broader patient population. Methods: Retrospective data from 5 centers were included in this study: 1 in Canada (n = 25), 1 in Norway (n = 75), 1 in Sweden (n = 18), and 2 in the United States (n = 36 and n = 26). At each center, pretherapy SSTR PET/CT imaging and postcycle 1 ¹⁷⁷Lu imaging-based dosimetry were performed according to site-specific protocols. The mixed-effects model treating centers as random effects was developed using baseline SSTR PET renal uptake values to predict renal absorbed dose from ¹⁷⁷Lu-DOTATATE. Additionally, leave-one-center-out cross-validation and leave-one-sample-out cross-validation were implemented for external and internal validation, respectively, measuring mean absolute error and mean relative absolute error. Results: Across all participating centers, the median cycle 1 renal absorbed dose was 0.56 Gy/GBq (range, 0.14-1.27 Gy/GBq), whereas the median pretherapy SSTR PET renal uptake was 110.7 Bq/mL/MBq (range, 28.6-287.7 Bq/mL/MBq). The differences among center means were statistically significant for both absorbed dose and PET uptake (P < 0.0001 from 1-way ANOVA). A significant (P < 0.05) correlation was observed between kidney SSTR PET uptake and ¹⁷⁷Lu-DOTATATE absorbed dose for each center (center-specific coefficient of determination ranged from 0.14 to 0.53). When data across all centers were aggregated, the mixed-effects model achieved a coefficient of determination of 0.25 (P < 0.01), resulting in an mean absolute error of 0.15 Gy/GBq (SD, 0.11 Gy/GBq) and an mean relative absolute error of 28% (SD, 24%) for external validation and 0.12 Gy/GBq (SD, 0.10 Gy/GBq) and 22% (SD, 20%) for internal validation. Conclusion: The correlations observed between SSTR PET renal uptake and ¹⁷⁷Lu-DOTATATE absorbed dose to kidneys across a multicenter population are statistically significant yet modest. The prediction model achieved a mean relative absolute error 28% or less for both external and internal validation of PET-predicted absorbed doses. The intercenter differences suggest the need for standardized imaging protocols and dosimetry workflows.