Slide 1: Towards a flexible model-informed precision dosing software solution using MonolixSuite
Towards a flexible model-informed precision dosing software solution using MonolixSuiteTM Dr. Manuel Ibarra Dr. Laura González Full Professor Associate Professor Department of Pharmaceutical Sciences Department of Computer Science Faculty of Chemistry – Universidad de la República Faculty of Enginee
Slide 2: About us
o ’ ~ we . ots Las Piedra Pando oe apf Pes aaros tance ‘Santiago 2 ae é Guudad HOSPITAL DE CLINICAS a8 ca UNIVERSIDAD Dr. Manuel Quintela mares DE LA REPUBLICA URUGUAY Montevideo Wy
Slide 3: Precision Dosing: tailoring treatments to patients’ needs
Precision Dosing: tailoring treatments to patients’ needs Individualization of drug treatment regimens based on patient characteristics known to alter drug absorption, disposition and/or response, with the aim of maximizing the probability of attaining both goals: efficacy and safety. Opposed to the
Slide 4: Precision Dosing: tailoring treatments to patients’ needs
Precision Dosing: tailoring treatments to patients’ needs Individualization of drug treatment regimens based on patient characteristics known to alter drug absorption, disposition and/or response, with the aim of maximizing the probability of attaining both goals: efficacy and safety. Delivering the
Slide 5: Quantitative pharmacology: Pharmacometrics
Quantitative pharmacology: Pharmacometrics I Trocóniz – UNAV, Spain.
Slide 6: MIPD – key to bridge knowledge and improve the accuracy of dosing recommendations
MIPD – key to bridge knowledge and improve the accuracy of dosing recommendations Gonzalez D et al (2017) Clin Transl Sci
Slide 7: Model-Informed Precision Dosing (MIPD)
Precision dosing + Quantitative pharmacology = Model-Informed Precision Dosing (MIPD) Use of modeling & simulation approaches (e.g. pharmacometrics: population PK/PD models, PBPK, regression models, decision trees, etc.) in combination with individually measured patient and disease characteristics t
Slide 8: Model-Informed Precision Dosing (MIPD): Nonlinear mixed effects models
Model-Informed Precision Dosing (MIPD) • Nonlinear mixed effects models Base model Covariate model Structural model Fixed effects Explain variability Stochastic model Random effects Characterize the unexplained variability
Slide 9: Model-Informed Precision Dosing (MIPD)
Model-Informed Precision Dosing (MIPD) C vs t PO 𝜂𝜂𝑘𝑘𝑘𝑘 80 𝜔𝜔2𝑘𝑘𝑘𝑘 70 𝑘𝑘𝑘𝑘𝑖𝑖 = 𝑘𝑘𝑘𝑘𝑝𝑝𝑝𝑝𝑝𝑝 + 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 60 Conc. (mg/L) 2 𝜂𝜂𝑘𝑘𝑘𝑘 ~𝑁𝑁 0, 𝜔𝜔𝑘𝑘𝑘𝑘 50 40 30 𝜂𝜂𝑖𝑖𝑖𝑖𝑖𝑖 20 10 𝜂𝜂𝑉𝑉𝐷𝐷 0 0 5 10 15 20 25 30 𝜔𝜔2 𝑉𝑉𝐷𝐷 𝑉𝑉𝑉𝑉𝑖𝑖 =…
Slide 10: Individualizing pharmacometrics models
Individualizing pharmacometrics models Empirical Maximum A-posteriori Estimation Balance between prior knowledge (parameter distribution in a given population) and individual data 𝑝𝑝 𝑛𝑛 � (𝜃𝜃𝑗𝑗 − 𝜃𝜃𝑗𝑗 ) 2 (𝑦𝑦𝑖𝑖 − 𝑓𝑓 𝜃𝜃𝑗𝑗 , 𝑥𝑥 )2 𝑀𝑀𝑀𝑀𝑀𝑀 𝑂𝑂𝑂𝑂𝑂𝑂 ≅ � 2 +� 𝜔𝜔𝑗𝑗…
Slide 11: Empirical Maximum A-posteriori Estimation
A balance between prior knowledge and collected data Empirical Maximum A-posteriori Estimation 𝒑𝒑 𝒏𝒏 � 𝒋𝒋 )𝟐𝟐 (𝜽𝜽𝒋𝒋 − 𝜽𝜽 (𝒚𝒚𝒊𝒊 − 𝒇𝒇 𝜽𝜽𝒋𝒋 , 𝒙𝒙 )𝟐𝟐 𝑀𝑀𝑀𝑀𝑀𝑀 𝑂𝑂𝑂𝑂𝑂𝑂 ≅ � +� 𝝎𝝎𝟐𝟐𝒋𝒋 𝝈𝝈𝟐𝟐 𝒋𝒋=𝟏𝟏 𝒊𝒊=𝟏𝟏 M Lavielle – sia.webpopix.com
Slide 12: Empirical Maximum A-posteriori Estimation
A balance between prior knowledge and collected data Empirical Maximum A-posteriori Estimation 𝒑𝒑 𝒏𝒏 � 𝒋𝒋 )𝟐𝟐 (𝜽𝜽𝒋𝒋 − 𝜽𝜽 (𝒚𝒚𝒊𝒊 − 𝒇𝒇 𝜽𝜽𝒋𝒋 , 𝒙𝒙 )𝟐𝟐 𝑀𝑀𝑀𝑀𝑀𝑀 𝑂𝑂𝑂𝑂𝑂𝑂 ≅ � +� 𝝎𝝎𝟐𝟐𝒋𝒋 𝝈𝝈𝟐𝟐 𝒋𝒋=𝟏𝟏 𝒊𝒊=𝟏𝟏 M Lavielle – http://shiny.webpopix.org/mcmc/bayes1/
Slide 13: Bayesian updating
Bayesian updating Upton D. et al. 2014. AAPSJ Dashboard systems: implementing pharmacometrics from bench to bedside.
Slide 14: Bayesian forecasting
Bayesian forecasting A. Fuchs. Implementation of Bayesian Therapeutic Drug Monitoring in Modern Patient Care
Slide 15: A complicated and time-consuming task
A complicated and time-consuming task Create dataset Estimate individual parameters Individual variables and Simulate observations Report 15
Slide 16: MIPD works!
MIPD works! Bayesian AUC-guided dosing in years 2 and 3 was associated with fewer additional blood samples per subject (3.6, 2.0, and 2.4; P < 0.003), shorter therapy durations (8.2, 5.4, and 4.7 days; P < 0.03), and reduced nephrotoxicity (8%, 0%, and 2%; P < 0.01). The median inpatient stay was 20
Slide 17: MIPD works!
MIPD works! Roggeveen, L.F., Guo, T., Fleuren, L.M. et al. Crit Care 26, 265 (2022). https://doi.org/10.1186/s13054-022-04098-7
Slide 18: MIPD works!
MIPD works! • Clinical benefits • Financial impact – Reduce the incidence of preventable ADRs – Improve adherence – Reduce hospitalization days
Slide 19: MIPD has not been widely integrated into clinical practice
MIPD has not been widely integrated into clinical practice Kluwe, F .et al. (2021), Perspectives on Model-Informed Precision Dosing in the Digital Health Era: Challenges, Opportunities, and Recommendations. Clin. Pharmacol. Ther., 109: 29-36. doi: 10.1002/cpt.2049
Slide 20: MIPD Software is quickly evolving
MIPD Software is quickly evolving Kantasiripitak W. et al. Front. Pharmacol., 07 May 2020 Volume 11 - 2020 | https://doi.org/10.3389/fphar.2020.00620
Slide 21: Our own difficulties and needs
Our own difficulties and needs • Software accessibility ($) • Engage medical staff and hospital authorities • Ability to apply the models that best fit our needs and perspectives • Start off on the right foot in terms of model performance
Slide 22: Vancomycin: Inter-model variability
Inter-model variability Vancomycin Broeker A. et al. 2019. Clinical Microbiology and Infection 25 (2019) 1286.e1e1286.e7
Slide 23: Tacrolimus: Inter-model variability
Inter-model variability Tacrolimus
Slide 24: Vancomycin: Model validation
Model validation Vancomycin Broeker A. et al. 2019. Clinical Microbiology and Infection 25 (2019) 1286.e1e1286.e7
Slide 25: Tacrolimus (n=40): Model validation
Model validation Tacrolimus (n=40)
Slide 26: Isoniazid/Rifampicin (n=38): Model validation
Model validation Isoniazid/Rifampicin (n=38) Isoniazid Rifampicin
Slide 27: MIPD in Uruguay: Vancomycin in critically ill patients
MIPD in Uruguay Vancomycin 31 critically ill patients
Slide 28: Software perspective
Software perspective As MIPD is advancing to the clinical setting, there is a need for flexible software solutions that support its application by pharmacometric groups, contributing to demonstrate the advantages of the approach in clinical practice and clinical research.
Slide 29: Grupo Interdisciplinario en Dosificación de Presición
Grupo Interdisciplinario en Dosificación de Presición Interdisciplinary model-informed precision dosing group – Pharmaceutics – Medical doctors • Nephrology • Neonates • Infectious diseases • Intensivists – Veterinary doctors – Statisticians – Computer engineers
Slide 30: Developing a software solution
Developing a software solution • Key characteristics – Accessibility (cloud-based, server-client architecture) – User friendly with 3 modules: • Physician (able to simulate from the population parameters) • Pharmacist at the TDM unit (able to perform the bayesian forecasting) • Admin (able to change
Slide 31: Finglix
Finglix • A prototype to implement MIPD at the University Hospital (Udelar), Montevideo - Uruguay
Slide 32: Finglix: Arquitecture
Finglix: Arquitecture Frontend : Backend {a = Monolix z = a ae =. Panel de : To. ‘Adminsitracion : TFTA) : om : com oF : Cola de Servidor mensajes Aplicacion web Sistema de archivos datos
Slide 33: Finglix: tools
Finglix: tools Frontend Backend q | Monolix bbRabbit 0 = <r ad =S Ee um Cola de ker Servidor mensajes =_ : Been |]: ges Tae = : y—— ae ® Aplicacion web: => i DB ‘Sistema de archivos
Slide 34: Finglix: tools
Finglix: tools Create dataset Estimate individual parameters New drug measurements Simulate Report
Slide 35: Finglix
Finglix
Slide 36: Finglix: Next steps
Finglix: Next steps • Provide dose recommendation based on exposure/response targets • Covariate visualization • Compartmentalize information by medical service • Methods: – Model averaging – Weights for available observations – IOV handling
Slide 37: Take home message
Take home message • Interdisciplinarity is essential for MIPD success • Several key challenges to be solved on site • Need to prove to the medical staff and health care center authorities the benefits of MIPD • Great software available for MIPD – Pharmacometricians might want to use their own valida
Slide 38: Thank you
w) GIDP GRUPO INTERDISIPLINARIO EN DOSIFICACION DE PRECISION UNIVERSIDAD DE LA REPUBLICA URUGUAY Thank you SB SimulationsPlus MIDD® Model Informed Drug Development + 2023