Population-Based Model Selection Algorithm developed at the example of Lu-177-PSMA Therapy

Ziel/Aim The calculation of absorbed doses for radionuclide therapies strongly depends on time-integrated activity coefficients (TIACs), whose values depend on the measured biokinetic data and the chosen fit function. The aim of this study was therefore to develop an algorithm for the selection of an adequate fit function for the case of a low number of measured time points for kidneys using a population-based approach in Lu-177-PSMA therapy.

Methodik/Methods The biokinetic data of Lu-177-PSMA in kidneys obtained from thirteen patients with metastatic prostate cancer were included in the study. In total, ten exponential functions with various parameterizations of mono- and bi-exponential functions were used. The parameters of the functions with different combinations of global and individual parameters were simultaneously fitted to the biokinetic data of all patients. Goodness of the fit criteria were used to test the quality of the fits and the corrected Akaike Information Criterion (AICc) was used to select the fit function most supported by the data.

Ergebnisse/Results As a result, the function A1*b*exp(-(lambda1+lambda_phys)*t)+A1*(1-b)*exp(-lambda_phys*t) was fitted with an acceptable fit based on the goodness of fit criteria and was selected as the best model based on the AICc weight of 97 % (probability to be the best fit function as supported by the data). In this function, the A1 and lambda1 were fitted individually while parameter b was fitted globally with the estimated value of 0.963 ± 0.004. This function can be used to estimate the time-integrated activity coefficient of a patient with three biokinetic data with an estimation of A1 and lambda1 and fixing b to 0.963.

Schlussfolgerungen/Conclusions We developed an algorithm to define an adequate fit function (to be used for future patients) based on a relatively small number of data, i.e. three time points per patient.

Publication History

Article published online:
08 April 2021