`Drug Development
`Perspectives in
`Clinical Pharmacology
`
`Editecf by
`
`Neal R. Cutler
`Caiifawzia Cfirzical Tz1}:z!s, 8%-s2:‘Iy Hilis, Ctmfomicz, U521
`
`John J. Sramek
`Caf£fa:'m’a Ciirzicczf Triafs, Beverb: Efiffs, Calsfomio, USA
`
`Prem K. Narang
`Pizcmnacia Acfricz, Cfiaical P§sar222acoZ0g}-jPfzamzaco:%i22e£2'es, Cofiunzfms, Ohio, USA
`
`IQHN WILEY 8: SONS
`Chichester - New York - Brisbane ° Toronto - Singapore
`
`|nnoPharma Exhibit 10260001
`
`
`
`Copyright
`
`1994 by Iohri Wiley 3: Sons Ltd,’‘
`Baffins Lane, Chiczhester,
`West Sussex PO19 IUD, England
`Telephone:
`Chichester {(3243} ?‘?9’???
`National
`International +44 243 ?'?9?"??
`
`"except Chapter 12 which is in the public domain.
`
`All rights rese1'ved.
`
`No part of this book may be reproduced by any means?
`or transmirtecl, or translated into a machine language
`without the written permission of the publisher.
`
`Otlzer ll’/Way Editorial Offices
`
`Iolm Wiley fir Sons, loo, 605 Thirtl iwenue,
`New York NY 101580012, USA
`
`Jacaranda Wiley Ltd, 33 Park Road, Milton,
`Queensland 4064, Australia
`
`john Wiley 3: Sons (Canada) Ltd, 22 Worcester Road,
`Rexclale, Ontario M9\X’ 1L1, Canada
`
`John Wiley 8: Sons {SBA} Pte Ltd, 3'?’ Jalan Pemimpin ;%0S-04
`Block B, Union Industrial Building, Singapore 2035?
`
`Library of Czmgress Catczirygirzg-irz-Paciviicatirm Data
`
`Pharmaeodynamiee and drug development : perspectives in clinical
`pharmacology} edited by Neal R. Cutler, john I. Sramek and Pram K.
`Narang.
`cm.
`p.
`Includes bibliographical references and index.
`ISBN 0 4'31 95052 1
`
`1. Drugsmlfiiysiological effect.
`Prem K.
`III. Sramek, John I.
`EDNLM: l. Pharrnarsology, Clinical. QV 38 P3318 1994}
`RM3G{).P48
`1994
`6lS'.?——clc20
`
`I. Cutler, Neal R.
`
`II. Narang,
`
`DNLMHDLC
`for Library of Congress
`
`946103
`{ZIP
`
`British Liizrary Gatafeguirrg in Pztbfieatioza Sam
`
`3‘; catalogue record for this book is available from the British‘ Library
`
`ISBN 0 4?} 950 521
`
`Typeset in 10;‘ 12 pt Plantirt by
`Mathematical Corttposition Setters Ltd, Salisbury, \5i"iltshi1‘e
`Printecl and bouncl in Great Britain by
`Bookeraft (Bath) Ltd, Midserner Norton, Avon
`
`|nnoPharma Exhibit 1026.0002
`
`
`
`
`Contents I
`
`
`
`List of Contriirgutors
`
`0
`
`0
`
`_
`
`Fareword t
`Lesfie Z. Benet
`
`’
`
`I OVERVIEW OF PHARMACODYNAMICS
`
`1 Basic Pharmacodynamic Concepts and Modem
`Raisers} ii’/E535
`
`2 Simultaneous i’harmacakinetic,{1’harmacociynamic Modeling
`Wayrze A. Coiinmz am? Mickaef A. Efdan
`
`3 Factors Influencing Variability in Kinetics and Dynamics
`Pram K. Ncmmg am!‘ Ronaia’ C. Li
`
`4 Populatiomfiased Approaches to the Assessment of
`Phatmacokinetics and Pharmacodynamics
`‘
`Eosepiz C. FZe2'332a}aez* and Ekiward 5'. zinicré
`
`S Generai Perspectives on the Role of Metabolites in
`Pharmacokinetics and Pharmacodynamics
`Rcmdaff D. Seéferz
`
`6 Enantioselectivity in Drug Action and Drug Metabolism:
`Influeace on Dynaanics
`Hey: K.
`I<'2‘0eme:*, Amzetze S. Grass med Micfzei }_3Iic?zeZ£:amrz
`
`’? Regulatory Perspective: The Role of Pharmacokinetics and
`Pharmacodynamics
`Lawrence Lesfea (ma? Roger L. Wéffiams
`
`1! APPLICATION OF PHARMACODYNAMICS IN SELECTED
`THERAPEUTIC DOMAINS
`
`8 Theoreticai Models for Developing Anxiolytics
`P. V. Nickefi and Wzamczs W’. Ufzcfe
`
`9 Pharmacodynamics of Antidepressants
`Karon Dawkém, Husseini K. Marzji and if/éifiam 2. Poster
`
`2:
`
`xiii
`
`3
`
`19
`
`45
`
`‘F3
`
`39
`
`103
`
`1 15
`
`133
`
`15?
`
`|nnoPharma Exhibit 10260003
`
`
`
`vi
`
`CONTENTS
`
`Pharmaccedynamics of Antipsychotic Drugs in
`Schizophrenia
`3033?: 3’. Smmeie and George M. Simpson
`
`11
`
`Pharmacodynamic Modeis Useful in the Evaluation of
`Drugs for Cognitive Impairment
`Micizaefi F. Murphy, Kiazsdizzs R. Sieg,f}*s'ec2’, F. facoia Hujffnnd
`Neai R. Cutie?
`
`181
`
`201
`
`Alzheimer’s Disease: Assessment of Chnlinomimetic Agents
`Scat: A. Reines
`
`225
`
`Antihypertensive Drugs
`L. Micfiznes? P:~is¢:m and Aiber-3 A. Cam‘
`
`Pharmacodynamics of Calcium Antagonist Drugs
`Dawefi R. Abemezizy czmi Nabii S. Andmrwis
`
`Agents in Congestive Heart Failure
`Edmund V. Capparefli
`
`Antiarrhythmic Drugs
`Eanice B. Sclzzuezrzz
`
`Antibiotic Pharmacodynamics
`30322: C. Rotscizafer, 1'<'cm’::z 3’. Waifzer, Kilt‘? 3’. K3555» cmci
`C}m'szop?2er ff. Szefiitscziz
`
`Pharmacodynamics of Antineoplastic Agents
`G233}: L. Rosszer and Mczzrfe 3'. Retain
`
`Controlling the Systemic Exposure of Anticancer Drugs:
`The Dose Regimen Design Problem
`Dania! Z. D’Args2rzio and fofzzvz H. Redraw:
`
`20
`
`Virology and Antiviral Drug Development
`Mficfzaef A. z"i??3€I?IE€<.Z,
`_§*::mzes R. Minor anti Stepizen E.’ Sirens
`
`Ill
`
`FRONTIERS IN PHARMACODYNAMICS: INSIGHT FROM
`MOLECULAR APPROACHES
`
`21
`
`ozvfidrencceptors and their Subtypes: Plaarmacological
`Aspects
`P. 2%. mm Zwieterz
`
`241
`
`253
`
`26?‘
`
`291
`
`315
`
`345
`
`363
`
`3??
`
`409
`
`|nnoPharma Exhibit 1026.0004
`
`
`
`CONTENTS
`
`22 Muscarinic Receptors: Pharmacologicai Subtypes,
`Stmcmrea Function and Regulation
`Lire Mei, Wi§§£a:;z R. Roesiee and‘ Hemy I. Ycmza;~mm:z
`
`23 Serotmiin Receptor Subtypes
`Doknz B. Pritcizest
`A
`
`Index
`
`Vii
`
`433
`
`4-57"
`
`475
`
`|nnoPharma Exhibit 10260005
`
`
`
`This material may be protected by Copyright law (Title 17 U.S. Code)
`
`1 Basic Pharmacodynarnic Concepts
`and Models
`
`ROBERT J. WILLS
`
`Drug therapy is intended to produce and maintain an efficacious pharrnaco~
`logical response(s). The dosing regimen for most drugs has evolved from a
`combination of pharmacokinetics and dose-—response. However, there is strong
`evidence that pharmacological response correlates better with plasma concen»
`trations. If concentrations can be related to pharmacological response, then the
`optimization of therapy stands to be improved.
`flzarmacodynarnics is a measure of the time course of pharmacological
`response to the presence of a given drug. Understanding the pharmacodynarnics
`of a drug is a key step towards understanding the reiationship between concen-
`tration and effect. The value of this understanding is evidenced by the
`resurgence of scientific and regulatory emphasis being placed on the use of
`pharniacokineticsfpharmacodynamics in optimizing therapeutics, in particular
`during drug development. The recent literature is replete with neanmethod
`development for pharmacological endpoints, delineation of the pharmaco-
`dynamics of many drugs and application of kinetics and dynamics to therapeutic
`utility. It is clear that the principles of pharmacodynarnics combined with
`pharmacolrinetics has utility to clinicians and to drug developers alike.
`This chapter will
`review the basic pharmacodjgnamic models} provide
`guidance to model selection, and highlight some considerations for evaluating
`pharmacological response and the relationship to drug concentrations.
`
`PEIARMACOLOGICAL RESPONSE
`
`‘ The pharmacological response is any physiological action attributable to the
`presence of drug. The response can be desired in the case of a pharmacological
`action which is a measure of or is a surrogate to therapeutic effectiveness. The
`response can also be undesired in the case of a toxicological response. Altering
`the dose of a drag to effect a change in the clinical response was the historical
`first step towards optimising drug therapy. However, this empirical approach
`did not always produce the expected response. The limitations associated with
`
`Piaamzacodyrzareaics and Brag Devefoprzmzi: Perspectit-es is Cfiniml Plmmzamingy
`Edited by N. R. Cutler, I. I. Sramek and P. K. Narang
`1994 John Wiley 8: Sons Ltd
`
`|nnoPharma Exhibit 10260006
`
`
`
`4-
`
`OVERVIEW OF PHARMACODYNAMICS
`
`defining} qtiantifying and interpreting the dose-«response gave rise to the
`cori<:entration—-effect
`relationship. The basic principle assumes that
`tirng
`concentrations circulating in plasma are more reflective of drug concentrations
`at the site of action {receptor} than dose, because the dose—response relationship
`is confounded by pi’1at‘1T13C0l-ilfltiticl variability associated with the absorption}
`distribution and elimination of a given drug.
`There are two key components of the pharmacological response that are
`critical to the successful nnderstanding and utility of the pharmacodynamics of
`a drug. The response should have clinical meaning or the establishment of the
`concentration—effect relationship may simply be an exercise of little value. The
`response should be reliably measurable. Developing validated methodologies
`for quantifying pharmacological response continues to be a maior impediment
`limiting the utility of pharmacodynamies.
`in the simplest case, monitoring
`changes in blood pressure is both clinically meaningful and reliable as a
`phaimacodynamic measure of the therapeutic effectiveness fo1*antih3rpertensive
`drugs. However, can this measure be related to drug concentration?
`This is the right question to ask but often difficult to answer. If the response
`changes immediately with a change in concentration, establishing a con-
`centration-effect relationship is highly probable. However,
`the response is
`often delayed in relation to appearance of drug since the site of pharmacological
`action is generally ‘()‘l1tSi{l€ of the vasculatnre. The concentration-effect profile
`takes on the shape of an anticlockwise hysteresis loop (Figure 1). Here,
`it
`becomes somewhat difficult to establish a concentration—effect relationship.
`One approach is to measure drug at the site of action, but this is rarely possible
`for a nnmber of scientific, technical and ethical reasons. Another approach is
`to measure the response at steady state. The approach which will be diseusseci
`
`80
`
`60
`Response4:C}
`
`
`20
`
`0
`
`l .0
`0.5
`Concentration
`
`1.5
`
`Figure 1. Response as a function of concentration shows a time clepenclence relative to
`the rise and decline in concentration. The arrows indicate the direction in time
`
`|nnoPharma Exhibit 1026.000?
`
`
`
`l’l~IrXRNlACOI2>YNAMIC i\rl0i7}El...S
`
`'5
`
`throughout the remainder of this chapter is a modeling approach that takes
`into account
`the time course of drug concentrations and pharmacological
`response.
`V
`
`Prior to discussing pharrnacodynamic models, it is worthwliile mentioning the
`many factors which influence the pharmacological response and by doing so
`confound the delineation of the pharniacodgnianiics even though the response
`may have clinical meaning and may be reliably measurable.
`lnter—subiect
`variability to pharmacological response is the most influential factor that affects
`the pharmacodynamics of any drug. The disease state and the pharmacolcinetic
`state {organ function) of the patient,
`the pharmacolcinetic behavior of the
`drug (metabolism, protein binding, etc.) and the hiochemicalfphysiological
`mechanisms involved contribute to inter-subject variability of pharmacological
`response.
`
`Further examples of influential factors that confound pharmacodynamics
`include indistinct pharmacological measures such as sedation or pain, which are
`highly variable among patients and therefore difficult to quantitate reliably.
`Responses that occur after the drug has been eliminated or that last well beyond
`the presence of a drug are difficult
`to relate to concentration since these
`situations often represent a cascade of biochemical or physiological events.
`Tolerance to the continued presence of a drug alters the concentration-effect
`rclationship as a function of time and adds complexity to the pharmacodynamic
`understanding of a drug. Diurnal variation. compensatory responses and
`responses (desired or undesired) sensitive to the rate of appearance of plasma
`concentrations require design and data analysis modifications to gain an under»
`standing of the concentrationweffect relationship. There is a considerable body
`of literature that discusses the pharmacological response in relation to concen-
`trations of drug (1-5). The point here is that pharmacological response is a
`complex measure that needs to he understood and reliably measurable to be
`meaningful and useful.
`
`PHARMACODYNAMIC MODELS
`
`The use of mathematical models to aid in testing hypotheses that would not
`otherwise he experimentally feasible has proven to be a useful developmental
`tool. Pharmacodynamic models are no exception. In experimental medicine,
`modeling can be a valuable surrogate where experiments are difficult or
`impossible to conduct because of practical
`limitations or ethical concerns.
`Pharrnacodynamic models were born out of existing chemical models. The Emx
`model is an adaptation of the Michaelis~Menten {6} equation, which describes
`the kinetics of enzymatic reactions. The sigmoid Ema model is a derivation
`of the Hill C?) equation, describing the mass action of chemical dissociation,
`and the Langmuir (8,9) equation. describing the phenomena of physical
`adsorption. It is believed that Langmuir’s work seeded the development of the
`
`|nnoPharma Exhibit 10260008
`
`
`
`6
`
`OVERVIEW OF ?HARMACODYNAMICS
`
`dr11g—rcceptor theories advanced by Clark (l0) and serve as the basis for
`concentration~effect relationships.
`
`THE LINEAR MODEL
`
`The simplest of all pharmacodynaniic models is the linear model. Here concen-
`trations of drug are proportionally related to a pharmacological response. The
`model takes the form of the equation for a straight fine where E is the effect,
`which is usually expressed as a fraction of maximal response, C is the drug
`concentration, 5 is the slope or the rate of change in response with a given
`change in concentration, and EL. is the effect when no drug is present. Even
`though this model can account for a baseline effect,
`
`E=S><C+En
`
`(I)
`
`this model cannot predict a maximal response, since this model predicts a
`continuous increase in effect with increasing concentrations. Unfortunately, the
`absence of a maximal response is inconsistent with most physiological stimuli.
`Another deficiency of this model is the limited usefulness since for most drugs
`the relationship between effect and concentration is not linear.
`an attempt to adjust for the non—linearity gave rise to the log~linear model,
`which, through log transformation of concentrations, provides a linear approxi-
`mation of a nonlinear relationship. In the equation, the baseline effect is
`replaced by I, an empirical constant of no physiological meaning. This model
`cannot predict the effect
`
`E"—~S><lOgC-E-f
`
`(2)
`
`in the absence of drug, nor a maximal effect. The model can only predict effects
`which fall between 20% and 30% of the maximal response (1 1). This model can
`be useful if the 20-80% range of maximal response can be ascertained, and if
`this occurs over a wide range of concentrations. An example of the application
`of the log-linear model is depicted in Figure 2.
`
`THE Emx MODEL
`
`This model most often describes the concentration~effect relationship. The
`equation for the model describes a liyperbolic relationship where E and C are
`as defined above, Em“ is the maximum effect attributable to the drug, and the
`EC” is the concentration that elicits 50% of the maximal effect. Unlike the
`linear model, this model predicts a maztimal response:
`
`E
`
`: Elnitx X C
`Ecgo +
`
`(3)
`
`An alternative to transforming the data when the response is inhibitors? is to use
`the inhibitory Sum model. In this model E3 is the effect when no drug is
`
`|nnoPharma Exhibit 10260009
`
`
`
`IWIARMACODYNAMIC MODELS
`
`7
`
`C
`
`’
`3,
`I 0
`It
`
`/G
`
`too
`
`80
`3 ‘so
`.22C!)
`2:
`a0:5
`0-
`
`40-«
`
`
`
`2.5
`
`3.0
`
`1.5
`
`2.0
`
`Drug in central compartment (mg I mg}
`
`Figure 2. Log-linear fit of neuromuscular blocking effect (% paralysis) of tubocurarine
`and the amount of tubocurarine in the body. The solid circles are the actual data, which
`Show a traditional Sohape. The open circles and the line represent the log-linear fit over
`the 20-80% effect range (Reprinted with permission from reference 11}
`
`present, and Em“ is the maximum reduction in response. If a drug is capable
`of completely abolishing an effect, than Enm becomes Ba, and Equation 3
`reduces to a fractional Ema model:
`
`Eflhll ”@“+—c>
`
`C
`
`“*9
`
`An example is illustrated in Figure 3, where the relationship between
`trirnoprostil, a PGE2 analogue, plasma concentrations and inhibition of meal-
`inducecl gastric acid secretion is best described by an inhibitory Emmi model
`(12).
`
`THE SIGMOID Eniax MODEL
`
`Dftentimes the concentration-effect curve takes on a more pronounced S-shape,
`which is not adequately described by the Em; model. Wagner Cl) and later
`Holford and Sheiner (3) adapted the Hill (2?) equation to the Enm model as a
`means of improving the fit. The difference involves the use of an exponent, 32,,
`which determines the slope of the curve and has no physiological meaning
`
`|nnoPharma Exhibit 10260010
`
`
`
`8
`
`OVERVIEW OF PHAREWXCODYNAMICS
`
`I00
`
`90
`
`80
`
`Q
`
`T‘’{}
`
`60
`
`C
`.9
`E5
`'59w
`E in
`Q .2
`36
`5 :3
`3 ‘E 50
`“RE
`‘Egg
`;§ § 40
`
`30
`
`20
`
`IO
`
`G g
`
`cg
`$
`E
`
`---—---Moder:
`E: 0.95?°G
`1.‘! + 0
`
`--~~-b;<::daI2 {3'0xC}0_87
`- (3.9 X f;}0.87_,_ (~;25)Q.8?
`o Actuaidata
`1'73‘-‘-‘
`
`
`
`_!...........................l
`3
`4
`Mean plasma coru::entration (ng! ml)
`
`Figure 3. An inhibimry 5",“ mode} fit of meal-izzduced gastric acid secretion and
`trimeprostii plasma Concentrations. The solid circles are the anzmai data and the lines
`represent the best fit. Model 1 shows the parameters for the fit; IC5¢.= 1.1 ngfml and
`E,1m"—~95.?‘%/o inhibition. Mbdel 2 refers to 3 sigmoid Enm model linked to a kinetic
`made} where 12:13.8? §R:3printed with ps:s~missior1 from refbrrence 12)
`
`attributed to it This modei collapses to an Em“ model when the exponent has
`a value of 1 {see Figure 3}:
`
`E
`
`__ E-Essa; X C".
`ECS0!£ + C}!
`
`(S)
`
`3
`Simiiar to the Enm an inhiE:»i‘t0r
`The influence of the expon<~:r1t,
`profile is shown in Figure 4.
`
`res anse can be incur aerated into this model.
`:2, on the shape of aha c<mcem1*ation—effect
`
`|nnoPharma Exhibit 1026.0011
`
`
`
`l’H3lRl\*irXCOl3YI‘~i£&l»ilC MODIELS
`
`9
`
`$00
`
`80
`
`response
`Percentageofmaximal
`
`60
`
`40
`
`20
`
`Concentration
`
`Figure 4. Response as a function of concentration conforming to the sigmoid Emx
`model with the same ECso, the same Emx and varying exponent values
`
`THE PROBABILITY MODEL
`
`If the pharmacological response is subjective and difficult to grade, or when the
`importance of a pharmacological response is whether it occurs or not and not
`the clegree of the response} it may be useful to determine the probability of
`achieving such an effect as a function of concentration. There are several
`statistical approaches that can he used, such as the Kaplan—Meier analysis (13)
`and the Cox regression analysis (14). As an example, Antal at :12. (15) applied
`this model to the reduction in the number of panic attacks during the final week
`of therapy of alprazolam compared with the number of panic attacks at baseline
`using the Cox regression analysis. They determined that a 73% chance of being
`a major responder (75% reduction in panic attacks compared to baseline) is
`associated with an aIp1'a:»:olam steady—state plasma concentration of 481-lg/ml.
`Likewise, Aural or :12. (15) determined that there was a 50% probability of
`sedation emergence associated with an alprazolam plasma concentration of
`40 ngfml. A comparison of the probability profiles for these two different
`outcomes appears in Figure 5. The authors conclnderl that at an alprazolam
`concentration necessary to elicit a 335% probability of a major response there was
`< 50% chance of the emergence of sedation. In general, this model is well suited
`to therapeutics where response is often sufoiective, a large placebo effect
`generally exists and the inter-subject variability is usually large.
`
`MODEL r3DPlPTATiONS
`
`There are many literature examples of where the standard models are adapted
`to meet the specific needs of a particular drug. For exarnple, Smith er al. (16)
`
`|nnoPharma Exhibit 10260012
`
`
`
`:0
`
`OVERVIEW OF PI~IARMACODYNAMICS
`
`Probability
`
`o 2
`
`0 £l£
`
`0
`
`25
`
`50
`
`125
`100
`3'5
`Alprazolam concentration (no I m!)
`
`150
`
`175
`
`20!}
`
`Figure 5. Probability of being 2 maior responder (solid line} or for the emergence of
`sedation {dashed line} as a function of alprazolatn plasma concentrations in patents
`suffering from panic attacks (Reprinted with permission from reference 15}
`
`to account for
`incorporated an exponential function into the Emax model
`pharmacological tolerance. The development of tolerance could be represented
`by an exponential decay in the maximum effect as shown in Equation 6.
`
`"_ Em X e ""“ x C"
`E "
`E050» + Cu
`
`The tolerance function can easily be incorporated into the E3250 term, resulting
`in an exponential increase. Another adaptation involves the dynamic modeling
`of ioop diuretics where C and E030 are replaced by ER and '.E§R5o——the urinary
`excretion rate, and that which is associated with an increase in the sodium
`excretion rate which is 30% of maximal, respectively (i?',18).
`
`THE LINK MODEL
`A more elaborate approach to understanding and predicting oharmacoiogical
`response as a function of concentration when the response is delayed in relation
`to the appearance of drug (hysteresis) is to use the link model.
`
`E
`
`:
`
`{?2a;Z/V1)‘'‘
`(}e€gZf V1)” + (Css5g}”
`
`{7}
`
`This model, first proposed by Sheiner at as’. (19), links a pharmacokinetic model
`to a pharmacociynamic model; and was applied to the sitnnitaneous fitting of
`
`|nnoPharma Exhibit 10260013
`
`
`
`”*IiARMACODYNAlviIC MODELS
`
`I l
`
`he phartnacoltinetics and pharmacodynsniics of d—-tuhocurarine. The phar-
`tacodynamic model has the look of the sigmoid Ema model, where C3350 is the
`St€3C1}’*8l2il'.€
`concentration corresponding to SCI‘?/o maximal effect, ken is a first-
`order rate constant for rernoval of drug from the site of action, is is the exponent
`hat determines the shape of the curve, and Z/ V1 represents the sum of
`exponentials
`for
`the appropriate pharrnacoi-zinetic model. This modeling
`echnique collapses the hysteresis, allowing for a true eoncentration~—effect
`characterization. it is the most comnionly used of the link models.
`Two other link approaches have been described and applied. A non-
`aararnetric approach was proposed by Fuseau and Sheiner (20) and later
`odiiied (23). This approach makes fewer assumptions about the specificity of
`the kinetic/ dynamic models and the underlying physiological mechanisms. A
`systems approach proposed by Veng—Pedersen (22) also makes fewer assump-
`ions about the specificity of the kineticjclynamic models where linear and
`on—linear systems are described using operators.
`These link-model approaches are an effective Way to adjust for hysteresis, and
`hus describe and understand the concentration—effect relationship. The theory,
`advantages, limitations and applications are discussed in more detail in the next
`chapter and elsewhere (23).
`
`MODEL SELECTION
`
`SCIENTIFIC BASIS
`
`Model selection begins with the understanding of the pharmacological response.
`Is the response the result of inhibition or stimulation? is the response a direct
`action of the drug or a cascade of events? is there dinrnal variation in response?
`Does the response have a baseline or a maximum? What is the biochemistry?
`Is the duration of the response important to clinical outcome? The answers to
`these kinds of questions, as well as others, lead to the understanding of the
`pharmacological response, yet are often not answerable. The challenge facing
`experts working in this area is to gain as much of an understanding as possible
`to best select a model.
`
`DESIGN CONSIDERATIONS
`
`Study design is a key component toward obtaining meaningful data. When
`conducting pharmacodynamic studies, it is imperative to examine the placebo
`response, since this will define the baseline pharmacological response. It is also
`strongly recommended that
`these studies he conducted in a double-blind
`fashion to protect against bias and to lend validity to the dynarnic n’ieasure(s}.
`It has to be decided whether to conduct a single-dose or a I’I’111ltipl€-(i088 study.
`This choice is dependent on what is known about the study drug. The choice
`
`|nnoPharma Exhibit 10260014
`
`
`
`l2
`
`OVERVIEW OF PHARMACODYNAMICS
`
`of study population, healthy subjects or patients, has to be made. The
`meaningful choice is to use patients, but the decision needs to be based on
`riskjhenefit, on probability of compliance for patients, since these studies are
`often rigorous, and on the relevance of healthy subjects to the patient population
`(24). It is a good idea to test the study population for poor responders. A
`pre~study rtm~in should also be considered to properly acclimate the study
`population. Consider excluding as many external factors as possible that can
`introduce variability across the study population, such as concornitant medi-
`cation, smoking and other controllable factors. Lastly, ample and appropriate
`sample times are critical to ensure that the model will be meaningful, hut care
`has to be taken so as not to affect the outcome of the dynamic response(s).
`The goal here is to reduce variability, enhance validity and optimize clinical
`utility.
`
`THE DnTA
`
`Once data have been collected, they should he visually inspected. This step is
`often taken too lightly as predetermined expectations sway the experimentalist.
`The data themselves are generally the best indicator of the appropriate dynamic
`model. Exaniine the changes in plasma concentrations of drug and changes in
`response to doses of drug and placebo over time. hire the implicit assumptions
`of dose response in play, are there anomalies, etc.? Ask yourself, what is
`maximally or minimally expected in the presence and absence of drug? What
`makes sense phartnacologically? For example, one would not expect a calcium
`channel blocker to drop blood pressure to zero.
`Based on this data overview, decisions as to the handling of the data need to
`be made.
`Ideally, evaluating each individual separately is the most robust
`approach, since individual treatment provides intra- as well as inter~sul>iect
`variability. Pooling of data becomes necessary when the data density from
`individuals is not sufficient for analysis or the data are highly variable. Pooling
`has the risk of maslting individual anomalies that may be indicative of a small
`subset of responders. An example might be a small subset of hypersensitive
`responders whose shift to the left on the concentration—response curve could be
`lost when data are pooled.
`0
`The next step is to decide whether to use the raw data or to transform the
`data. This is critical because transforming the data carries assumptions about
`the data themselves and the pharmacological mechanism.
`In particular,
`transforined data alter the distribution of the data and can conceal baseline
`
`variability. Transformation also determines which equations may he used with
`a given dynamic model. In general, data should not be transformed Without a
`theoretical basis, but in practice it is hest to incorporate the choice of raw versus
`transformed data as part of the selection of a model.
`
`|nnoPharma Exhibit 10260015
`
`
`
`PHARMACODYNAMIC MOI3l3l.S
`
`l3
`
`MODEL SELECTEON
`
`Choosing a pliarrnacodynamic model comes from the data analysis and
`ltnowledge of the pharmacological mechanism. At this point there is some trial
`and error necessary prior to the final selection. Initial parameter estimates such
`as EC5g or Enm for any model can usually be obtained from inspection of the
`concentration—effect plot of raw or transformed data. These estimates should be
`used in a non-linear regression analysis of the concentration—effect data,
`employing whatever rnodel/(s) have been initially selected. Overlaying the
`predicted data on top of the actual data will provide guidance as to the next
`steps. Qftentirnes, the results of these regressions will eliminate poor models,
`because a predicted parameter violates what is already known to be. This point
`is best made by using the example of Kroboth st of. (25).
`The example involves
`the inhibition of gastric acid secretion by the
`H3*l3l0CkCf3 nizatidine. The percentage inhibition (transformed) of gastric
`acid secretion as a function of nizatidine plasma concentrations (includes
`placebo response) were fit using the Em}. and the sigmoid Emax models. The
`predicted regressions are shown in Figure 6(a) and (b), respectively. Here the
`choice is obvious in favor of the sigmoid Ema}: model when the data were
`transformed.
`
`The authors also modeled the raw data using the inhibitory sigmoid E...“
`model. The net
`inhibition of gastric acid secretion (placebo minus drug
`response) was fit as a function of concentration. The regression is shown in
`Figure 6(c). Here the model predicted a maximal inhibition of 36. ml3q relative
`to baseline. The actual data showed the potential for greater suppression. In this
`case, transforming the data was more appropriate. The reader is referred to
`Kroboth er at. (25) for the details surrounding this example.
`
`MODEL JUSTIFICATION
`
`Good science dictates that statistical justification be used to discern between
`models even in the obvious example given above. The better model is based on
`a smaller residual sum of squares, smaller 95% confidence intervals for
`parameter estimation and smaller standard deviations around the parameter
`estimates. The distribution of the residuals can be used to discern between the
`
`‘fits that rnodeis predict at the extremes. The application of the F-test (26) and
`the Akaike Information Criterion (2?) can also be used to distinguish the better
`model when the difference between the models is an additional parameter.
`Testing for the robustness of the model is equally encouraged, since an
`obvious goal of modeling is to predict events or outcomes for a larger popula-
`tion. The more perturbations that can be explained by the model, the greater
`the predictability to a real system. As a simple example, the robustness of a
`kinetic/dynamic model developed for oral administration may be tested by
`
`|nnoPharma Exhibit 10260016
`
`
`
`14
`
`QVERVIEW OF PHARMACODYNEKMICS
`
`100
`
`
`
`
`
` 03O Acidsuppression(%)
`
`{3}
`
` Q 1
`
`o
`zoo 40:2: 600 goo 1o0n12eo1~:eo1ez201a0o2ooo
`Prasma concentration mg I mi)
`
`‘K30
`
`0
`
`20o 490 scao soc 10-£>012<}€}14001£3001€~(}02<X)0
`
`Plasma concentraiican (mg I mi}
`
`C)
`
`/'~.U“m.»
`
`(O/O)
`
`Acidsupprasséon
`P§acebo—treatment{mEq) 0
`
`(C)
`-
`
`0
`
`500 1ooo1soo2o<)o25o03ooo3aoa4oc>o45a-05000
`Coneantration (ng Irni)
`
`Figtzre 6. Ca) Gastric acid suppression as a functi<>:t{ of nazatidine plasma concsemrations
`{solid circles) along with a regression fit (solid line) using the Em“ model. (b) Same data
`along with a reg1‘essi(3n fit {su-lid line) using the sigmoid Enm model. {:3} Inhibition of
`gastric acid secraztion re1ati\.*c:t(3 placebo as :1 function of nazazidine plasma czoncentrations
`(solid sziycles) along with the regression fit {solid line) using the inhibitory Emax modei
`(Reprinted with permission from reference 25}
`
`|nnoPharma Exhibit 1026.001?
`
`
`
`Pi-Ir5iRivir§C{)D"Yi\lrX3\=ilC MODELS
`
`13
`
`comparing the sirnttlatecl data for intravenous administration with actual data
`collected after intrattenons adniinistration.
`
`SUMMARY
`
`If a goal of health care professionals is to provide nieaningful therapeutics, then
`an action to this goal would he the development of a rational process for
`optimizing drug therapy.
`Integration of pharmacokinetics and pharmaco-
`dynamics is one such process that has been applied successfully to a number of
`therapeutics. In the case Where there is a wide enough separation between
`concentrations that produce the desired response and concentrations that
`produce the undesired response (applicable to most drugs}, the utilization of
`pharmacokinetics and pharmacodynamics may help establish an optimal dose
`and regimen during drug development, but will not be needed to individnalize
`patient therapy since it is more practical to prescribe doses of drug that ensure
`therapeutic effectiveness in most patients where the occurrence of toxicity is of
`little or no consequence. For drugs where the separation is narrower, pharrnaco~
`kinetics and pharrnacodynarnics can be utilized to individnalize therapy to
`ensure a separation of desired from undesired responses. Applicable to either
`situation, understanding the concentratiomeffect
`relationship can aid in
`optimizing drug therapy.
`The intent of this chapter was to review pharmacoclynarnic concepts and
`models as they apply to the concentration—effect relationship. Progress has been
`made towards utilizing kinetics and dynamics in understanding the variability
`of pharmacological response. The challenge that continues to face researchers
`working in this area is to develop more sophisticated technology for measuring
`pharmacological response and to increase the knowledge of the underlying
`biochemical and physiological mechanisms associated with a given disease. The
`regulatory, economic and social expectations being placed on health care world-
`wide provide the necessary impetus.
`
`REFERENCES
`
`t...‘
`
`‘s§i?agt1er]G. Kinetics of pharmacologic response. 3'. Tizeor Biof. 1968; 20: l?’3w201.
`.
`2. Holford NHG, Sheiner LB. Kinetics of pharmacologic response. Pfznrmacol Ylzer.
`1982; 16: 143-166.
`3. Holford NHG, Sheiner LB. Understanding the dose—effect relationship: clinical
`ap