Justine Henriot 1,2, Sunay Rodriguez Pérez 1, Michelle Andersson 1,3, Bjorn Baselet 1, Tomas Opsomer 1, Maarten Ooms 1, Lara Struelens 1, Michel Koole 2, Clarita Saldarriaga Vargas 1
1 Belgian Nuclear Research Centre (SCK CEN), Nuclear Medical Applications Institute (Mol, Belgium), 2 KU Leuven, Department of Imaging and Pathology, Nuclear Medicine and Molecular Imaging, (Leuven, Belgium), 3 Université Libre de Bruxelles, Medical Physics Department, Institut Jules Bordet (Brussels, Belgium )
Objective. Peptide receptor radionuclide therapy (PRRT) with [¹⁷⁷Lu]Lu-DOTA-TATE is an established treatment for somatostatin receptor type 2 (SSTR2)-expressing neuroendocrine tumours, yet it is currently administered using fixed-activity regimens that do not account for inter-patient variability. This radiopharmaceutical combines the β⁻-emitting radionuclide ¹⁷⁷Lu with DOTA-TATE, a somatostatin analogue. The DOTA-TATE vector offers intrinsic modularity, as it can be labelled with multiple radionuclides for diagnostic and therapeutic purposes, including ¹⁶¹Tb, a promising therapeutic alternative to ¹⁷⁷Lu due to its comparable physicochemical properties and additional emission of short-range Auger and internal conversion electrons. In this context, there is a clear need for predictive tools that enable individualized treatment planning while remaining adaptable to different radionuclide-vector combinations. The objective of this study is to evaluate whether a generalized preclinical physiologically based pharmacokinetic (PBPK) model can provide accurate, organ- and lesion-level pharmacokinetic predictions across multiple DOTA-TATE–based radiopharmaceuticals, thereby supporting both treatment personalization and modular radiopharmaceutical development.
Methods. A murine whole-body PBPK model was implemented using MATLAB Simbiology based on a published model (1). Model generalizability was first assessed by comparing two cross-radionuclide modelling strategies. In strategy 1, PBPK models were developed and verified for [¹⁶¹Tb]Tb-DOTA-TATE in healthy and tumour-bearing mice and subsequently applied to [¹⁷⁷Lu]Lu-DOTA-TATE; strategy 2 followed the reverse approach. For each strategy, the model was initially developed in healthy mice, incorporating novel in vitro binding data (2), plasma protein binding, and competition between labelled and unlabelled DOTA-TATE for receptor targeting. Kinetic parameters and tissue-specific SSTR2 densities were estimated by fitting the model to preclinical biodistribution data in kidneys, liver, pancreas, spleen, lungs, blood, and tumour. The model was then verified against independent datasets (3,4) and further developed and verified using biokinetic data of mice bearing a CA20948 rat pancreatic tumour (5). The best model was subsequently evaluated for broader generalizability using preclinical data from 54 published studies of DOTA-TATE labelled with various therapeutic β–emitters (177Lu, 161Tb, 153Sm) and imaging tracers (64Cu, 68Ga, 125I, 111In). The impact of model structure simplification on predictive performance was also investigated. Comparison of modelling strategies and evaluation of the generalizability were performed considering the two-fold limits [0.50;2.00] and using geometric mean fold errors (GMFEs) of organ-specific areas under the curves (AUC) or individual concentrations.
Results. Both strategies produced fitted curves in good agreement with the experimental data, with most predicted-to-observed ratios within the two-fold range and residuals between -3 and 5. Although strategy 1 yielded slightly lower GMFEs than strategy 2 (healthy mice: 1.50 vs 1.67; tumour-bearing mice: 1.34 vs 1.36), predictive performance did not differ significantly between strategies (Wilcoxon signed-rank test, p = 0.5506). As [¹⁷⁷Lu]Lu-DOTA-TATE is clinically approved, strategy 2 (model development based on [¹⁷⁷Lu]Lu-DOTA-TATE data) was selected for the broader generalizability assessment. Application of the [¹⁷⁷Lu]Lu-DOTA-TATE–based models to additional radionuclides resulted in good or acceptable predictive performance in 68% of the 54 datasets. Consistent accuracy was achieved for DOTA-TATE radiolabelled with ¹⁷⁷Lu and ¹⁶¹Tb in C57BL/6 or BALB/c mice, as well as with 68Ga and 111In across BALB/c, Swiss, and other nude strains. Structural model simplification by lumping non-SSTR2-expressing organs and the adrenal glands into a single compartment had minimal impact on predictions, with relative changes in organ AUC ranging from −2.14% to +0.08%.
Conclusion. PBPK models were successfully developed and validated for [¹⁷⁷Lu]Lu-DOTA-TATE and [¹61Tb]Tb-DOTA-TATE. The [¹⁷⁷Lu]Lu-DOTA-TATE models (healthy and tumour-bearing mice) reliably extended to three other radionuclides, including 161Tb, supporting its translation for PRRT and 68Ga, their theragnostic pair. Overall, the generalizability assessment demonstrates model robustness across radionuclides and experimental settings, while structural simplification preserved predictive accuracy, highlighting suitability for therapeutic translation. Collectively, this framework provides a practical basis for dosimetry-guided personalization of DOTA-TATE PRRT and supports the development of next-generation SSTR-targeted radiopharmaceuticals.
References:
1. Zaid NRR, Kletting P, Winter G, Prasad V, Beer AJ, Glatting G. A Physiologically Based Pharmacokinetic Model for In Vivo Alpha Particle Generators Targeting Neuroendocrine Tumors in Mice. Pharmaceutics. 2021;13:2132.
2. Spoormans K, Struelens L, Vermeulen K, Saint-Hubert MD, Koole M, Crabbé M. The Emission of Internal Conversion Electrons Rather Than Auger Electrons Increased the Nucleus-Absorbed Dose for 161Tb Compared with 177Lu with a Higher Dose Response for [161Tb]Tb-DOTA-LM3 Than for [161Tb]Tb-DOTATATE. J Nucl Med. 2024;66.
3. Busslinger SD, Mapanao AK, Kegler K, et al. Comparison of the tolerability of 161Tb- and 177Lu-labeled somatostatin analogues in the preclinical setting. Eur J Nucl Med Mol Imaging. 2024;51:4049-4061.
4. Schüler E, Österlund A, Forssell-Aronsson E. The amount of injected 177 Lu-octreotate strongly influences biodistribution and dosimetry in C57BL/6N mice. Acta Oncologica. 2016;55:68-76.
5. Bernard BF, Krenning E, Breeman W a. P, et al. Use of the rat pancreatic CA20948 cell line for the comparison of radiolabelled peptides for receptor-targeted scintigraphy and radionuclide therapy. Nucl Med Commun. 2000;21:1079.
Reference: PAGE 34 (2026) Abstr 12031 [www.page-meeting.org/?abstract=12031]
Poster: Drug/Disease Modelling - Absorption & PBPK