lec 23 - applications of PK/PD modeling in drug discovery and development (wang)

model informed drug development learning and confirming

• M&S = modeling and simulation → performed before each decision point to quantitatively assess risk in moving forward

• preclinical

  • efficacy

  • toxicology

  • PK-PD

• phase I

  • toleration

  • human PK-PD

• phase IIa/IIb

  • efficacy and safety

  • dose/exposure-response

  • dose adjustments

• phase III

  • therapeutic index

  • covariate effects

• registration labeling/phase IV

  • results relative to competitors

  • regional differences

  • therapeutic index

PK and PD

• PK = quantitative analysis of the time-course of a drug in the body

  • ADME

  • what body does to drug

• PD = quantitative analysis of the time course of a drug effect

  • onset

  • duration

  • intensity

  • what a drug does to the body

  • possible PD endpoints

    • target engagement

    • target-mediated signaling pathway (gene/protein expression)

    • cell proliferation/apoptosis

    • tumor growth

    • disease score

    • symptom

PK/PD pt 2

• left graph

  • regimen A = therapeutic window

  • regimen B = excessive adverse effects

• right graph

  • plasma warfarin concentration vs perfect decrease in normal prothrombin activity

  • as warfarin plasma concentration decreases there is a greater decrease in normal prothrombin activity

  • delayed effect (?)

PK/PD pt 3

• PK

  • disposition kinetics

    • dosing regimen → Cp = plasma concentration of drug

  • biophase distribution

    • Ce = effect site concentration → linked to k_eo (rate constant describing how quickly Cp equilibrates with Ce)

    • exposure

• PD

  • biosensor process

    • drug binds to receptor or target leads to → biosensor activation

  • biosignal flux

    • generated by k_in (rate of biosignal production) and k_out (rate of signal loss/degradation)

    • how quickly the system reacts to drug presence

    • duration of action and signal delay

  • transduction

    • biosignal is translated into a response

  • response

success rates from first-in-human to registration

• CV drugs had highest success rate

• CNS/oncology = low success rates

• overall industry average success = 11%

• 1991-2000 success rate

  • most success in phase I

  • downward trend

reasons for attrition (loss)

• in 1991 → biggest reason = PK/bioavailability

  • → smallest reason = formulation

• in 2000 → biggest reason = efficacy

  • → smallest reason = formulation

drug development and the FDA

• preclinical testing

• investigational new drug application (IND)

  • approval for clinical trials

• phase 1 studies

• phase 2 studies

• phase 3 studies

• new drug application (NDA)

preclinical testing

• pharmacology

  • in vitro and in vivo pharmacology studies

• drug metabolism and PK

• toxicity

  • 2-4 week studies in 2 animal species after multiple dosing

• multi-scale modeling informed discovery and development

  • PK/PD time profile in animals

    • predicts human efficacious dose prior to phase I

    • integrates in vitro evidence with in vivo preclinical data

  • TK (toxicokinetics) modeling to assist dose selection int eh toxicity study in animals

phase 1

• dose levels predicted based on preclinical studies

• first-in-human study

• healthy volunteers (or patients with target disease)

• 20-80 subjects

• short duration

• maximal tolerated dose, adverse events

• determine PK parameters in humans — dose proportionality

• unblinded, uncontrolled

• usually NOT evaluating for efficacy; efficacy assessed for oncology drugs

• refine the human PK/PD model using the observed human data

• model the additional biomarker data in phase 1 trial

• model-based dose selection for phase 2 trial

phase 2

• evaluating efficacy

• dose-response

• patient with target disease

• several hundred subjects

• several months

• double blind, placebo controlled

  • double blind = neither participants nor the researchers know who gets treatment vs placebo

• adverse events, PK

• formulation and dose-dependent food effect prediction

  • predicts how different formulations and food intake affect drug absorption

• drug-drug interaction prediction

• special population predictions (peds, organ impaired population, etc)

• modeling and simulation to analyze all exposure-response relationship and explore dose choices for phase 3 study

• identify co-variates effects E-R (exposure-response)

  • co-variates = pt factors that influence drug response such as

    • age, weight, sex

    • genetics

    • kidney/liverfunction

phase 3

• evaluating effectiveness, risk-benefit

• patients with target disease, more diverse population

• 1000s of subjects

• several years

• double blind, placebo controlled, active control, randomized

  • active control = trial drug vs existing standard treatment

• adverse events, PK, dosing intervals, drug-drug interactions, drug-disease interaction

• most expensive stage of drug development

• population PK and PD

  • collection of relevant PK/PD info in patients who are representative of the target population to be treated with the drug

  • ID and measurement of variability during development and evaluation

  • explanation of variability by identifying factors of demographic, pathophysiological, environmental or concomitant drugs

phase 4

• after NDA approval

• monitoring ongoing safety in large populations

• pts with target disease

• 1000s subjects

• several years

• uncontrolled, observational

• epidemiologic data, pharmacoeconomics

• special population predictions (preggo, asian)

• predict PKPD with new formulations

• compare results in competitors

types of PK studies

• single/multiple dose

• mass balance (ADME)

• bioavailability/bioequivalence

• food effect (for oral formulations)

• special populatons

  • male, female

  • pediatric, elderly

  • renal or hepatic impairment

• drug-drug interactions

• QTc prolongation

mass balance studies

• study that tracks the entire fate of drug using radiolabeled compounds

• commonly use radiolabeled compounds (_¹⁴C)

• usually single dose

• usually healthy volunteers

• provides information on metabolites and excretion routes

• blood samples, urine, feces are collected

  • in animal studies, can also collect bile and tissues

bioequivalence

• approval of generic drugs

• when changes are made for the approved marketed formulation

• to demonstrate similarity to previous formulation

• rate and extent of absorption

• absence of a significant difference in concentration-time profiles in blood for 2 drug products

  • Cmax, Tmax, AUC

translational research

• the transfer of new understanding of disease mechanisms gained in the lab into the development of new methods for diagnosis, therapy and prevention and their first testing in humans

human dose prediction: 3 primary attributes determine human dose

bioavailability, CL, potency

• preclinical estimates of the 3 attributes is translated into human expectations with uncertainty

• integrate all the human expectations into expected human dose requirement

  • integrate known 1st principle relationships:

    • dose = CL/F*potency

human CL prediction: useful interspecies scaling methods

allometry → scaling using the relationship between body size and biological processes across species to predict drug CL

• rule of exponent → universal method, SA requires correction factors based on exponent b

  • if b < 0.55 → simple allometry is ok

  • if if 0.55<b<0.71 → use simple allometry

  • if 0.71<=b<1 → use CL x MLP (max. lifespan potential)

  • if 1<b <1.3 → use CL x BrW (brain weight)

  • if b>=1.3 → use FCIM method

• FCIM, fu corrected intercept method → universal method based on the intercept obtained from the SA log-log plot and ratio of unbound fraction in plasma between rats and humans

• metabolism normalized allometric scaling if metabolism is major pathway

  • CL_{NAS =} a(BW)^b

  • CL_NAS = CL_{animal-in-vivo} * (human_{CLint-hep-in-vitro}/animal_{CLin-hep-in-vitro})

• in vitro-in vivo extrapolation (IVIVE) based on microsome or hepatocytes

  • bro just look at the equations in the photo

  • PBSF = physiological-based scaling factor

  • Q blood= hepatic blood flow

human Vss prediction: useful interspecies scaling methods

• oie-tozer equation → universal method, uses fup and Vd from animals

  • look at photo for equation b/c i aint typing all that

  • fup = fraction unbound in plasma

  • accounts for binding in plasma and tissues UNLIKE simple allometry

• simple allometry using total or free Vss (2 or more species)

  • look at photo for equation

  • power law method using body weight scaling

  • can also normalize to unbound drug

  • used when multiple species data are available

• PBPK based approach

  • in silico method (computer based/mathematical models) that uses logP (or logD), fup, pKa, and tissue composition data to predict drug’s tissue distribution

  • predicted tissue:blood partition coeff (Kp) values can be used to build a full PBPK distribution model

  • tissue distribution or quantitative whole=body autoradiography (QWBA) method

• rat-dog-human proportionality equation

  • assume Vss(human) = Vss(monkey) or Vss(human) = Vss(rat)

applications of physiologically-based pharmacokinetics (PBPK) model

• predicting the impact of intrinsic physiological changes on PK

  • hepatic impairment

  • renal impairment

  • pediatric

  • pregnancy/fetus

  • polymorphism

• predicting the impact of extrinsic factors on PK

  • age, health, demographic formulation PK

  • drug-drug interaction

    • PPI/ARA-DDI

      • ARA = acid-reducing agents

  • food effect

PBPK model: the interplay of system-specific factors and drug-specific factors

• drug specific parameters

  • molecular weight

  • pkA

  • logD/logP

  • pH-solubility profile

  • dissolution

  • particle size

  • dosage form

  • dosing regimen

  • permeability

  • Km, Vmax

  • fup, B:P, fu_inc

• system-specific paramteres

  • demographic and genetic factors

    • age

    • weight

    • height

    • sex

    • genetics

    • race

    • disease

  • physiological factors

    • GI transit time

    • gastric pH

    • bile salt concentration

    • organ size and the associated tissue types

    • blood flow

    • drug metabolizing enzymes

    • drug transporters

    • plasma protein

    • hematocrit

some tools for PBPK modeling

• simcyp

• gastroplus

• pk-sim

• PSE

• stella (GI-Sim)

population PK and PD

• evaluate concentration-response relationships

• estimate mean PK/PD parameters

• evaluate variability

• determine influence of co-variates on PK/PD

  • individual parameters

  • concomitant medication

• address regulatory concerns

  • special populations

  • product labeling

population PK/PD: sources of variability in population

• body size

• age

• protein binding

• difference in CL

  • hepatic CL

  • renal CL

• differences in endogenous substances

• differences in disease stage

• drug-drug interactions

co-variates ID by population PK analysis

• adalimumab → immunogenicity and age

• akakinra → weight

• alosteron → dose and sex

• aripiprazole → age, race smoking

• busulfan → smoking, pediatric

• daptomycin → renal function

• delaviridine → ketoconazle co-admin

• fexofenadine → geriatric, renal, and hepatic impairment

• galantamine → fluoxetine coadmin

• pramipexole → cimetidine coadmin

• rifapentin → sex

• rosuvastatin → race

• tiagabine → carbamazepine coadmin

• valgancyclovir → kidney, heart, and liver transplant pts

population PK/PD data analysis

• intensive sampling vs sparse sampling

• hypothesis testing vs exploratory analysis

• non linear mixed effect modeling

  • combines fixed effects (avg clearance in population) and random effects (variability between individuals)

  • non-linear = accounts for complex, real world PK relationships

  • complicated and time consuming

  • extensive statistical analysis

• identify and quantify sources of variability

  • between subject variability → differences across individuals

  • within subject variability (between occasion variability) → change sin individual across time or treatment cycles

  • inter-study variability → differences across separate studies or populations

  • residual variability → unexplained noise in data

population PK data

• left panel = population spaghetti plot

  • each line represents one individual

• observed vs predicted plots

  • tight clustering = good fit

software

• General PK/PD and compartmental modeling tools

  • Phoenix WinNonlin → for non-compartmental and compartmental analysis

  • ADAPT 5 → for nonlinear mixed-effects modeling

  • Berkeley Madonna → fast differential equation solver for PK/PD modeling

  • Matlab → general purpose math software

• Population PK modeling tools

  • Nonmem → gold standard for population PK modeling

  • Monolix → for nonlinear mixed effects modeling

  • S-ADAPT → simulation and estimation engine

• PBPK Modeling pathways

  • Gastroplus → used to simulate ADME

  • Simcyp → simulates drug interactions and special populations