Overview

Discovering new drugs through pharmaceutical research and development is a complex and lengthy process. Starting from the seventies, mathematical modelling approaches have been used to analyse biological and pharmacological data, to make predictions, to characterize pharmacological response and to design clinical trials with either confirmatory or exploratory purposes. This approach to pharmaceutical research is called Model Based Drug-Development (MBDD).

An important objective during drug development is the quantitative assessment of clinical drug efficacy and safety. In this context it is fundamental to properly characterize drug pharmacokinetics (PK) and pharmacodynamics (PD) and to determine, for example, which is the minimum effective dose or which dose is associated to outbreak of undesired adverse events. Within this framework, PK models aim to quantitatively describe the Absorption, Distribution, Metabolism and Elimination (ADME) processes of a drug. PK-PD models are focused on the relationship between drug concentration and the consequent pharmacological effect on a specific endpoint and/or a relevant descriptive causal biomarker.

Since PK models describe only observations of drug concentration in plasma with minimum model complexity, they have typically limited capacity for extrapolations and to predict concentration-time profiles in tissues and organs. Whole-Body Physiologically Based Pharmacokinetic models (WB-PBPK) represent a possible solution. These models combine the knowledge of anatomy and physiology of the organism with physio-chemical properties of the specific drug, and allow the prediction of the drug concentration also at the effect site.

A further attempt to better describe not only the pharmacometrics, but also the mechanism of action of a drug at the microscopic level is represented by a new emerging discipline in the field of mathematical modelling for biomedical research and pharmaceutical companies: the Systems Pharmacology (SP). SP has been defined as “the interface between Pharmacometrics and Systems Biology” [P. H. van der Graaf, 2012] and its objective is to develop mechanistic and quantitative models for the description of drug interaction with organism, integrating data and models at different scales and evaluating how individual components and the system interact each other.

Clinical trials that enrol a large number of patients are performed to quantitatively assess clinical efficacy (and safety) in a large population. The use of the population approach to PK-PD modelling is instrumental to characterize both the typical and individual responses, as well as the magnitude of the variability in the population. However, extensive sampling is not always feasible and identifying a complex mathematical model based on sparse data can be very difficult. The design of optimal sampling schedules plays an essential role in reducing costs while maximizing clinical trial power.

Some examples of mathematical and statistical tools are represented by the ordinary differential equations, the compartmental models and the nonparametric modelling of temporal profiles. The development and analysis of Non Linear Mixed Effects models (NLMEM) is one of the most interesting topics of our research.

Key words: Non Linear Mixed Effect Models (NLMEM), Maximum Likelihood (ML), Optimal Design, Bayesian approach, MCMC, Covariate analysis, Visual Predictive Checks (VPC), PK-PD, TMDD, DDI, mPBPK, WB-PBPK, mAb, FIH, PTS, DEB, IVIV, FIM, DMD, ADC, SP

Tools: NONMEM, Monolix, WinBUGS, OpenBUGS, Stan, PsN, Xpose, Berkeley Madonna, Matlab, R, Shiny, PFIM, PopED, PKSim, MoBi

Research Areas:

Oncology:

  • Solid tumour: development of a population PK-PD tumour-in-host model,  based on a set of tumour-host interaction rules defined by the Dynamic Energy Budget (DEB) theory and the Simeoni TGI model, to describe both the dynamics of cachexia and the tumour growth inhibition in xenograft rats
  • Solid tumour: development of a PK-PD tumour-in-host model, based on the Dynamic Energy Budget (DEB) theory, to describe the anti-angiogenic modulation of tumour growth in xenograft mice
  • Solid tumour: based on preclinical studies, derive steady-state properties between drug-driven and biomarker-driven tumour growth inhibition models in order to predict new drug antitumor potency
  • Solid tumour: development of a semi-physiological population PK model to describe the food effect on the PK of an antitumor drug, based on clinical data
  • Solid tumour: translational modelling approach to estimate first dose in humans of an antibody-drug conjugate (ADC) from preclinical data of efficacy and toxicity
  • Leukaemia (blood cancer): development of a population PK model and PK-PD models to describe the relationships between drug concentration and biomarker/tumour-size/survival time courses considering also the drug-drug interaction (DDI) in case of co-administered treatments
  • Drug-drug interaction: assessment of the effect of co-administrations of antiangiogenic and cytotoxic drugs on tumour growth inhibition
  • Drug-drug interaction: assessment of the effect of co-administration of an antitumor drug together with inhibitors/inducers of CYP3A4 enzymeon adverse events (eg. neutropenia, liver toxicity)

Degenerative diseases:

  • Duchenne Muscular Dystrophy (DMD): assessment of drug effect based on biopsy data

Pharmacometric modelling of biological drugs:

  • Autoimmune diseases: development of minimal PBPK (mPBPK) models incorporating Target Mediated Drug Disposition (TMDD) to describe PK-PD properties of monoclonal antibodies (mAb)

Mechanism-based PK/PK-PD models:

  • Infectious diseases: development of a WB-PBPK model to describe and predict antibiotic concentration at the site of action, with a special focus on drugs used to treat pulmonary tuberculosis
  • Respiratory diseases: development of a WB-PBPK model to study locally acting inhaled drugs
  • Hypercholesterolemia: integration of a WB-PBPK model with a metabolic network to describe, respectively, the distribution of Atorvastatin in different organs and its metabolism in the liver

Optimal design/Clinical trial design:

  • First In Human (FIH) studies: evaluation of safe and effective first dose in human based on preclinical data, scaling approaches and computation of Probability of Technical Success (PTS)
  • Paediatrics: clinical trial design optimization in paediatric trials, especially in case of rare disease, and extrapolation from adult to paediatric, using prior knowledge combined with modelling and simulations
  • Paediatrics: optimization of serum ferritin assessment to guide chelation therapy and to support individualized dosing strategy
  • Paediatrics: evaluation of adaptive clinical study designs (sequential, Bayesian, enrichment approaches) vs standard parallel design, mainly in terms of required sample size and time to completion
  • Oncology: design optimization of clinical studies in oncology using the Fisher Information matrix (FIM) within the nonlinear mixed-effect model (NLMEM) framework

Drug delivery:

  • In vitro release: development of a model describing the time course of the in vitro release of progesterone vaginal rings  at different doses
  • In vitro-in vivo (IVIV) correlation: development of an in vitro-in vivo population PK model to predict the time course of plasma concentration and AUC for the in silico assessment of in vivo bioequivalence

Software development:

  • Development of a PharmML-to-WinBugs converter, which supports Bayesian model estimation and simulation tasks in the DDMoRe Interoperability Framework platform
  • Development of a WinBUGS a connector integrating WinBUGS in the DDMoRe Interoperability Framework platform

EU collaborations:

Drug and Disease Model Resource (DDMoRe) - IMI EU project 2011-2016

The working group:

Research fellows and PhD students: Elisa Borella, Letizia Carrara, Silvia Grandoni, Silvia Lavezzi, Giulia Lestini, Nicola Melillo, Lorenzo Pasotti, Italo Poggesi, Giovanni Smania, Elena Tosca

Supervisors: Giuseppe De Nicolao, Paolo Magni