Individualized tacrolimus dosing in pediatric kidney transplantation: development of PBPK-based predictive models, simulations and applications (ModSimTer-Tacro Study)
Lucía Díaz-García (1), Andrea Guzmán-de Antonio (1), Alejandro Zarauza Santoveña (2), Diego Morante Martínez (2), Sergio Sánchez-Herrero (3), Antonio Carcas-Sansuán (1).
(1) Clinical Pharmacology Department, IdiPaz, La Paz University Hospital, School of Medicine, Autonomous University of Madrid, Madrid. Spain, Pediatric Nephrology Department, La Paz University Hospital, Madrid, Spain, (3) Empresarios Agrupados Internacional, Madrid, Spain
Objectives: Development of individual dosage (ID) methods is a long-standing goal pursued by clinicians. Therapeutic individualization requires processing a significant amount of information such as diagnoses and clinical and biological parameters related to drug efficacy and safety1-2. The reduction in efficacy and safety with the appearance of adverse events are quite common phenomena in relation to the prescription of some drugs, especially in certain populations (children, elderly, multiple pathologies and polypharmacy) and in complex pathologies such as transplantation. Therefore, the development of ID tools is of great value to improve patient care by reducing the inaccurate use of drugs and improving the safety and efficacy of treatments3. This study is a continuation of previous clinical trials in pediatric kidney transplant patients with Tacrolimus as the main immunosuppressant treatment. Despite immunosuppressive therapy is a therapeutic option that significantly increases the immediate and long-term survival rate after transplantation, patients can suffer significant clinical complications that compromise the functionality of the graft. The development and validation of predictive models, allowing for the concept of Model-Informed Precision Dosing tools (MIPD), enables improve an adequate balance of immunosuppression and this has become one of the main objectives of research translational in this area4-5.
Methods: This is an ambispective observational study (retrospective and prospective data) for the development and validation of tacrolimus MIPD strategies. In the prospective cohort, clinical data is being collected from the clinical electronic history and routine clinical practice samples are being taken to determine genetic polymorphisms of the genes involved in the metabolism and bioavailability of the drug and proteomic and metabolomics profiles. The retrospective cohort is divided in two cohorts, first cohort comes from previous research projects (PMID: 23907143, PI18/00136-TACRO-Omics and FC/HULP_00/2014, doi: 10.24217/2530-4984.17v1.00005). Second cohort comes from patients in routine clinical care at Pediatric Nephrology Department recruited in this study that will provide to the previous predictive model real life information and an internal validation to test this model
Results: At present, 76 patients have been recruited from the hospital's total cohort of 120 kidney transplant patients. During the study will collect different kind of variables, clinical data including defined relevant clinical events such as infections, lymphoproliferative syndrome, graft rejection etc, laboratory variables, PK variables, treatments, genetic variables and omics variables. The statistical analysis of the significant variables will be used to build the PK model. PK model will be constructed with two cohort (60:40), first cohort will be a training cohort to improve on the previous predictive model and second cohort will be used as an internal validation cohort to test this model. The machine learning algorithms used will be Artificial Neural Networks, Support Vector Machine, Random Forest, Naïve Bayes and KNN. For each individual algorithm, the optimal model will be evaluated through the Bayesian Information Criteria (BIC). Once the optimal model for each algorithm has been chosen, they will be evaluated jointly using the measures of accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and Kappa index.
Conclusions: The validation of the final model has the objective to develop an application that will be available in hospital computers or mobile devices with the purpose to the clinicians to adjust the dosage of immunosuppressive therapy in paediatric kidney transplant patients. Modelling of physiological systems are not common use in real clinical practice conditions so this ID tool combining clinical data and biomarkers will improve transplant patients management.
References:
[1] van der Meer AF, Marcus MA, Touw DJ, Proost JH, Neef C. Optimal sampling strategy development methodology using maximum a posteriori Bayesian estimation. Ther Drug Monit. 2011 Apr;33(2):133-46.
[2] Marouani H1, Zografidis A, Iliadis A. Kinetic nomograms assist individualization of drug regimens. Clin Pharmacokinet. 2011 Dec 1;50(12):773-9. doi: 10.2165/11594000-000000000-00000
[3] Prado-Velasco M, Borobia A, Carcas-Sansuan A. Predictive engines based on pharmacokinetics modelling for Tacrolimus personalized dosage in paediatric renal transplant patients. Sci Rep. 2020 May 5;10(1):7542.
[4] Bassingthwaighte, J. B., Butterworth, E., Jardine, B. & Raymond, G. M. Compartmental modeling in the analysis of biological systems. Methods Mol. Biol. 929, 391 -438 (2012).
[5] Barrett, J.S., Della Casa Alberighi, O., Laer, S. & Meibohm, B. Physiologically based pharmacokinetic (pbpk) modeling in children. Clin Pharmacol Ther 92, 40 -9 (2012).