PHARMACOKINETIC

SERIES

Dosage Regimen Design: Pharmacodynamic Considerations Roger

L. Williams,

E ach

dose of a drug in a patient is an experiment to ameliorate a pathologic state. Over time, the data base for the experiment is developed by a diverse group of individuals, including not only pharmaceutical and regulatory scientists and health care professionals, but also the patient. Specific objectives of the experiment are to maximize a desired set of drug responses and minimize a set of undesired responses. The challenge to pharmaceutical and regulatory scientists in creating and labelling pharmaceutical products is to bring concise and useful information to the experimental setting. The information required can be extensive, from an understanding of drug action at the molecular level, to biopharmaceutic and pharmacokinetic information to guide dosimetry, to a series of instructions to the patient and clinician that define positive and negative drug effects for a given dosing regimen. The greater degree to which this information is available for a specific drug, the more rationally therapy can proceed. An important part of the body of information required for rational therapeutics relates to the pharmacokinetics of a drug or active metabolite. Pharmacokinetics is a science that describes the time course of drug and/or metabolite(s) in the body relative to dose. Pharmacokinetic information can guide dosimetry so that a drug or its active metabolite achieves and maintains a useful concentration at the site of action. As reviewed by Dr. Thompson in Part I (Pharmacokinetic Principles) of this two-part series, the application of pharmacokinetic methodology to therapeutics has become an integral part of drug development. Pharmacokinetic methodology has been especially valuable in drug development to establish dosing criteria and to guide the development of formulations that deliver a drug reliably and consistently. The pharmacokinetic description of the time course of a drug or metabolite in the body is only part

From the Office of Generic Drugs, Center for Drug Evaluation and Research, Rockville, Maryland. The views expressed are solely those of the author and do not necessarily represent the views or the policies of the Food and Drug Administration. Address for reprints: Roger L. Williams, MD, Office of Generic Drugs, Center for Drug Evaluation and Research, HFD 600, 5600 Fishers Lane, Rockville, MD 20857.

J COn Pharmacol

1992;32:597-602

MD

of the ing

information

regimens.

required Further

to develop

information

rational

is needed

dos-

to relate

drug dose to drug effect. With this information in hand, the health care professional can decide how much drug to give to achieve a specific plasma or blood concentration that will in turn achieve a desired effect. The decline in effect with loss of drug from the body will determine when the next dose of drug is to be given. Ideally, dosimetry should be based on frequent measurements of pharmacologic effect over time, with the time of the next dose determined by the time at which effect falls below a desired level. Certain characteristics of a drug sometime allow this possibility. For example, prothrombin time and loss of analgesia are effect parameters that are used clinically to determine when the next dose of anticoagulant or analgesic is administered. Frequently, however, clear-cut pharmacodynamic effects

are

not

always

available,

and

the

clinician

must rely on prior knowledge of drug effect relative to concentration and on a pharmacokinetic description of the drug to determine dosing regimens. Optimally, it would be desirable to use concentration of drug at the actual site of drug action to guide dosimetry, but because of sampling and assay limitations, pharmacokinetic parameters usually describe drug concentration in a readily accessible fluid such as blood and urine. If drug effects are directly and consistently related to a concentration in an accessible fluid, the health care professional and patient can be provided with a reasonably standard set of dosing instructions and can monitor therapeutic goals not only with clinical observation but also with the services of a drug level laboratory.

DRUG Just course

EFFECTS as

pharmacokinetics of

a drug

or

can one

or

more

measure of

its

the

time

metabolites,

drug effect can be measured in terms of several different positive and negative actions over time. Selection of a drug effect, development of standardized methodology to measure the effect, and application of the methodology to a specific clinical setting are challenges to the clinical and regulatory scientist. In a sense, the science of pharmacokinetics was made

597

WILLIAMS

TABLE Levels

I

of Pharmacologic

Response Clinical

Drug

ReceptorImmediate

Propranolol Lovastatin

Beta adrenergic

Cimetidine

Histamine antagonist 4, Reduction vitamin formation t Insulin secretion 4.Cyclooxygenase

HMG-COA

Warfarin Tolbutamide NSAID

blockade

reductase inhibition H2 receptor K epoxide

possible by extraordinary advances in drug and metabolite assay methodology. Analyses of drug effect are at an earlier stage of development, and commonly accepted, standardized methods for assessing drug effect in a patient population are needed for many drugs. Drug effects can be defined both in terms of a primary interaction at the level of the receptor as well as in terms of immediate or long-term clinical effect (Table I). Pharmaceutical scientists have become increasingly adept at determining drug effect at a molecular level, and it is not an unreasonable goal for this information to be available for any new chemotherapeutic agent. However, measurement of primary drug action at the level of the receptor is usually impractical in the clinic, and measurement of long-term clinical benefit is frequently difficult, time-consuming, and expensive. For these reasons, the focus of both the pharmaceutical and regulatory scientist, as well as the clinician and the patient, is frequently an immediate and readily measured clinical effect, such as fall in blood pressure, reduction in gastric acid secretion, reduction in plasma glucose or alleviation of a specific set of symptoms. These effects are sometimes only markers of actual clinical benefit. In the clinic, a dose of drug is administered to maximize a desired set of responses and minimize an undesired set of responses. The relationship between dose, amount of drug in the body, and drug effect can be defined through the application of pharmacokinetic and drug effect models. Development of pharmacokinetic/effect models requires information about the primary active agents in the body (drug or metabolite) produced by administration of a drug, the ability to measure these active species in an accessible biologic fluid, knowledge that a relationship exists between concentration of the active agents

598

5

Effect

J Clin

Pharmacol

1992;32:597-602

Long.Term

Hypotension 4. Serum cholesterol 4. Gastric acid secretion

4. Cardiovascular 4. Cardiovascular

4. Coagulation

4. Thrombosis/rethrombosis

Improved glucose control 4, Prostaglandin formation

Improved diabetic 4. Inflammation

Ulcer

risk risk

healing

control

and effect, and finally an ability to measure one or more effects deemed clinically important. Choice of an effect of interest thus is an important objective for both the clinical scientist and clinician. Although drug effects are generally thought drugs produce different positive and A negative drug effect can sometimes

of singly, most negative effects. be a manifes-

tation of the primary action of the drug that occurs at higher doses (e.g., hypoglycemia that occurs with excessive doses of an sulfonylurea) or that occurs elsewhere in the body from the desired site of action (e.g., hypotension with theophylline as opposed to bronchodilatation). Sometimes a negative drug effect bears no relation to its primary action (e.g., hypersensitivity to penicillin). A valuable relationship between drug concentration and effect sometimes may not be available. Some drugs produce their effects by destroying a diseased cell or interacting irreversibly with a receptor. Anticancer drugs are frequently designed to destroy rapidly dividing tissues or are targeted to attach to and destroy a cell with a particular class

of receptors

(e.g.,

immunotoxins).

Similarly,

the

aipha-adrenergic blocking drug phenoxybenzamine binds irreversibly to the alpha receptor. For these agents, no clear relationship may exist between drug dose or concentration and drug effect. Similarly, effects of some drugs last much longer or bear no clear relationship to drug concentration. The effect of many beta-adrenergic blocking drugs appears to last substantially longer than would be predicted by the time course of the drug in the body. Where immediate markers of pharmacologic effect are not available, long-term benefit may be the only pharmacologic effect of interest, but correlation of long-term

benefit to a blood or plasma concentration is usually not possible. Although drug concentration/effect correlations are difficult to develop for many drugs, for many

DOSAGE

REGIMEN

others some definable and useful relationship exists between a pharmacologic effect of interest and the concentration of the drug or metabolite in an accessible biologic fluid. For these drugs, pharmacokinetic and effect models may be especially useful.

imal

DESIGN

to

maximal

concentration dicity, which tion of a large Temporal

PHARMACODYNAMIC

CONSIDERATIONS

Concentration/Effect

Relationships

Pharmacodynamics can be defined as the steadystate relationship between drug concentration and effect at the site of action.’ When a relationship exists between drug concentration in blood or plasma and the concentration of drug at the site of action, drug effects can also be related to blood or plasma drug concentration. Models of these relationships can provide clinically useful parameters regarding baseline and maximal effects (sensitivity aspects) and the change in the observed relationship with time (temporal aspects). The various relationships that can be defined between drug concentration and effect have been the subject of several reviews.2’3 Although several models can be defined (fixed effects, linear effects, loglinear effects), the most generally applicable is a specific form of an equation developed originally to describe the disassociation of oxygen and hemoglobin.4 E

Emax

EC,0

+

C C”

In the standard Emax model, the exponential term in equation 1 is unity. A more general form of the equation is referred to as the Hill equation or sigmoid Emax model, in which the exponential term is other than unity. In the Emax model, drug effect in initially linearly related to concentration but approaches a maximum (Emax) beyond which increments in concentration cause no further increment in effect. This particular concentration/effect relationship is thus intuitively attractive because it postulates no effect when drug concentration is zero and a maximum effect beyond which no increment is possible. The concentration at which effect is half-maximal is termed the EC50. With certain assumptions, this model reflects a reversible interaction between one drug molecule and one receptor (when the exponential term in Equation 1 equals unity) or between more than one molecule (when the exponential term in equation 1 is other than unity).’ Several variants of the Emax concentration/effect relationship have been developed to account for different drug-receptor relationships and interactions of more than one drug at the site of action.’ Drugs that move from mm-

PHARMACOKINETIC

SERIES

effect

over

a comparatively

short

range exhibit a high degree of sigmoiis modeled in equation I by the selecvalue for the exponential term.

Aspects

of the

Drug/Effect

Relationship

Because pharmacodynamics defines the steady-state relationship between effect and concentration, when all transfer and interaction processes are at equilibrium, clinical experiments to define the relationship should be conducted at steady state. These experiments are sometimes cumbersome and may require constant infusion of a drug over an extended period before measurement of pharmacologic effect. Pseudo-steady-state studies can be performed when effect is measured during the terminal log-linear decline of a drug from the blood or plasma, but studies employing this technique also generally require extended periods of observation and include the possibility that a stable relationship between concentration at the effect site and drug concentration in an accessible fluid was not attained. The pharmacodynamics of a drug can be studied in the absence of steady-state conditions through the application of a model that links a pharmacokinetic model describing the time course of drug concentration in an accessible fluid and a pharmacodynamic model (e.g., Emax) that relates the concentration of the drug or metabolite in the effect compartment (Ce) to effect. The link between the two primary models is defined by a hypothetical effect compartment containing drug of negligible mass in which Ce is predicted by the exit rate constant (kej of drug from the effect compartment. K becomes a parameter that adjusts for the temporal disassociation between effect and blood or plasma concentration that can arise as a result of delays in drug reaching its site of action or time required to produce a pharmacologic effect. First proposed by Segre5 and developed extensively by Sheiner and others,6’7 this model has been described and applied in several clinical settings over the last decade. Initially, the overall model consisted of three parametric submodels, in which concentration versus time was predicted by a pharmacokinetic model, effect concentration versus effect was predicted

by

a pharmacodynamic

model

and

the

two

models were joined by a link model. Nonparametric pharmacokinetic8 and pharmacodynamic9 submodels have also been proposed. Pharmacokinetic/pharmacodynamic link models work well in the presence of linear kinetics and when the relationship between Ce and effect is constant. When kinetics are nonlinear or when the observed relationship between Ce and effect is not

599

WILLIAMS

stable (e.g., metabolite tration-effect

when drug tolerance occurs or an active accumulates), interpretation of concenrelationships can be difficult.

important. Examples include determination of antibiotic sensitivity in bacterial cultures (in vitro), determination of coagulation parameters in the clinic (ex vivo), genesis

APPLICATIONS

and animal (preclinical

Despite Whether

steady-state

or non-steady-state

studies

are

used to define a relationship between a measurable concentration and effect, the assessment of pharmacologic effect is crucial to rational drug development and therapeutics. Before reaching the clinic, a new drug can be tested in several in vitro, ex vivo, and in vivo

screens.

The

value

of these

has been amply documented on them to exclude drugs the

from enters vivo

risk

that

a valuable

preclinical

screens

drug

will

be falsely

one

or more

cal

studies

assessing

600

5

will benefit precise and

II). Many

methods

11

conduction

12

time

13 14 15 16

time

QT interval

Theophylline Theophylline d-Tubocurarine Verapamil

FEy1

17 18

infusion

19 20 21 22

FEy1

Warfarin

Nerve stimulation/muscle PR interval QT interval Prothrombin time

Warfarin

Prothrombin

Tocainide

Arrythmia

complex

frequency

paralysis

synthesis

rate

Dedrug

from continued destandardized ways of

Examples

time

of

to

those redrugs, have

10

QT interval Action potential (in vitro) Activated partial thrombopiastin Ventricular tachycardia induction

to as be

clini-

ways

Reference

Quinidine

1992;32:597-602

In vivo different

effect.

Measurement

Nizatidine Propranolol Propranolol

J Clin Pharmacol

many

sophisticated and complex. because of these advances,

Selected

Refractory period Analgesia Bicycle exercise time Exercise ST response Blood pressure, heart rate Suppression of acid secretion Exercise heart rate Heart rate with isoproterenol

Morphine Nitroglycerin

measure-

parameters. on

effect, particularly of cardiovascular

pharmacologic

Intracardiac conduction Refractory period Blood pressure Treadmill exercise Heart rate Blood pressure Left ventricular ejection

Intracardiac QT interval

carcino-

II

Effect

encainide

nonclinical

(Table

pharmacologic to the evaluation

Studies:

Drug

Flecainide Heparin 3-Methoxy-0-desmethyl

effect relied

pharmacologic

assessing

Pharmacokinetic/Pharmacodynamlc

Digoxin Disopyramide

clinical have

testing in the clinic velopment of more

TABLE

Atenolol

of these

and

observation of interest both and regulatory scientist professional and patient will

become increasingly spite or perhaps

excluded

further clinical evaluation. Even after a drug the clinic, continued application of in vivo, ex and preclinical in vivo effect measurements are

N-acetylprocainamide

utility

ments, the primary the drug development well as to the health

assess lated

over time, but reliance from further testing entails

the

tests for mutagenesis in vivo).

23 24 6, 7, 25 26 27 28 29

PPIMN

DQAC

An outstanding example fronting drug development and clinicians in assessing fully is the current effort treatment

of the

diseases

of the challenges conand regulatory scientists pharmacologic effect useto develop drugs for the caused

by

the

human

im-

munodeficiency virus (HIV). Several in vitro, ex vivo, and in vivo preclinical models have been developed to identify potentially effective anti-HPv drugs, but the value of these preclinical models for all possible drugs, many of which act at different stages in the life cycle of the virus, remains uncertain. Once an anti-HIV drug enters clinical evaluation, surrogate markers of true clinical benefit have been defined, but again the clear relation of these markers either to drug dose and/or concentration or to long-term clinical benefit has been difficult to prove. Documentation

drugs

of

the

has

long-term

become

the protracted ical considerations

a placebo, and rate the effects

clinical

increasingly

course that

benefit

difficult

of the disease, ethical restrict dosimetry

of

anti-HIV

because

of

and politand use of

the availability of drugs that of the virus but do not cure.

amelio-

REFERENCES

1. Sheiner

LB: Pharmacodynamic

cokineticsfor San

Pharmaceutical

Fnancisco,

2. Holford

February NHG,

Clin

tionship.

3. Holford

modeling Scientists,

4-9,

methodology

to

define

SERIES

13). Pharmaof California.

1990.

Sheinen

LB: Understanding

Pharmacokinet

1981:6:429-453.

HGH, Sheiner LB: Kinetics Then 1982;16:143-166.

Pharmacol

(Lecture University

the dose

of pharmacologic

effect

rela-

response.

4. Hill AV: The possible effects of the aggregation of the molecules of haemoglobin on its dissociation curves. J Physiol (London) 1910;40:iv-vii.

5. Segre systems.

C: Kinetics II Farmaco

of interaction 1968:23:906-18.

between

drugs

7. Stanski

the

time

course of drug in an accessible biologic fluid is now well established as are models that relate concentration to effect. When steady-state conditions are not readily available to define a dose- or concentrationeffect relationship, non-steady-state pharmacokinetic/pharmacodynamic models can be applied. Despite these methodologic advances, many aspects of clinical drug development and therapy do not readily lend themselves to pharmacokinetic/pharmacodynamic scrutiny. Barriers to this scrutiny include: drug assay problems or lack of an accessible biologic fluid that can be sampled to measure drug concentration; lack of a immediate pharmacologic effect in the clinic that can be related to concentration; and lack of a relationship between an immediate measurable pharmacologic effect and long-term clinical benefit. In the evaluation and application of any drug, the following questions are useful: I) what is the pharmacologic effect of interest? 2) is it a positive (benefit) or negative (risk) effect? 3) are there more than one positive or negative effects of interest? 4) how is the effect measured and can the measurement be applied reasonably in a clinical setting? 5) is the effect immediate (for example, blood pressure reduction) or long-term (for example, prevention of stroke with an antihypertensive)? 6) can the effect be usefully related to a concentration to guide dosimetry? and 6) if a concentration/effect relationship is likely, can it

PHARMACOKINETIC

be defined at steady state or must non-steady-state pharmacokinetic/pharmacodynamic models be constructed? Focus on these issues is crucial to the success of any drug development program and more importantly to the primary therapeutic experiment between clinician and patient.

and

biological

6. Sheiner LB. Stanski DR. Vozeh S. Miller RD. Ham J: Simultaneous modeling of pharmacokinetics and pharmacodynamics: application to d-tubocuranine. Clin Pharmacol Then 1979:25:358-71.

SUMMARY Pharmacokinetic

DESIGN

DR. Ham j, Miller RD. Sheiner and pharmacodynamics of d-tubocuranine narcotic and halothane anesthesia in

LB: Pharmacokinetics during nitrous oxide man. Anesthesiology

1979:51:235-41. 8. Unadkat JD, Bartha F, Sheiner LB: Simultaneous modeling pharmacokinetics and pharmacodynamics with nonparametric netic and dynamic models. Clin Pharmacol Then 1986:40:86-93. 9. Fuseau E, Sheiner LB: Simultaneous modeling of pharmacokinetics

and

pharmacodynamics

with

a nonparametnic

pharmacody-

namic model. Clin Pharmacoi Then 1984:35:733-41. 10. Jaillon P. Rubenson D, Peters F, Mason JW, Winkle trophysiologic

Am J Cardiol

effects

of n-acetylprocainamide

in

12. ual

Kelman subjects.

AW,

13.

Whiting

Whiting

B, Holford

RE, Winchester beta

Kroemer

15.

Whitfield

and

anticoagulant

beings.

B: Modellingof

J Pharmacokin

Sheiner

LB:

HK.

Turgeon

disposition in vitro.

LR,

Clin

J. Parker

Quantitative

RA,

Roden

subjects Then

Levy

C: Relationship

effect

of hepanin

pec-

in individanalysis

relationship.

in human Pharmacol

DC:

for angina

drug response 1980:8:115-130.

Biopharm

NHG,

M, Harrison

blockade

the disopyramide concentration-effect Pharmacol 1980:9:67-75. 14.

RA: Elec-

human

1981:47:1124-1140.

11. Jackson C, Schwartz J. Kates Atenolol: once-daily cardioselective tons. Circulation 1980:61:555-560.

enantiomers: logic actions

of

ki-

DM:

of

J Clin

Br

Flecainide

and electrophysio1989:46:584-590.

between

in plasma

concentration

of normal

subjects:

magnitude Pharmacol 16. Roden

and predictability of intenindividual differences. Clin Ther 1980:28:509-516. DM. Lee JT, Woosley RL. Echt DS: Antiarrhythmic efficlinical electrophysiology, and pharmacokinetics of 3-meth-

cacy, oxy-o-desmethyl

encainide

(MODE)

in

patients

with

inducible

ventricular tachycardia or fibrillation. Circulation 1989:80:12471258. 17. Dahlstrom BE. Paalzow LK. Segre C. Agren AJ: Relation between morphine pharmacokinetics and analgesia. J Pharmacokinet Biopharm 1978:6:441-49.

601

WILLIAMS

18. Zimnin D, Reichek N. Bogin KT, Aurigemma C, Douglas P. Berko B. Fung HL: Antianginal effects of intravenous nitroglycerin over 24 hours. Circulation 1988:77:1376-1384. 19. Callaghan JT, Bergstrom WW: Intravenous nizatidine Pharmacol Then 1985:37:162-65.

RF. Obermeyen BD, King EP. Offen kinetics and acid suppression. Clin

20. Lalonde RL. Pieper JA, Straka RJ. Bottonff MB and Mirvis DM: Propranolol pharmacokinetics and pharmacodynamics after single doses and at steady-state. Eun / Clin Pharmacol 1987:32:31518. 21. McDevitt DC, Shand DC: Plasma concentrations time-course of beta blockade due to propranolol. Clin Ther 1975:18:708-713. 22. Holford NH, Coates PE, Cuentent TW, Riegelman LB: The effect of quinidine and its metabolites on the diogram and systolic time intervals: concentration-effect ships. Br / Clin Pharmacol 1981;11:187-95. 23.

Falliers

602

S

CJ: Pharmacodynamic

J Clln Pharmacol

and

1992;32:597-602

spirometric

and

the

Phanmacol

sustained-release theophylline pharm 1975:17:125-30.

24. Mitenko PA, Ogilvie RI: Rational ylline. N Engi / Med 1973;289:600-03. 25.

responses

to a

Ci, Van Beem

HBH,

model

McLeod for

Clin

intravenous K, Sibbald

pancuronium.

26. Echizen H. Brecht T, Niedergesass baum M: The effect of dextro-, levo-, atnioventnicular conduction in humans. 217.

Pharmacol

doses

Bio-

of theoph-

A, Watson Br

MJ: A

J Anaesth

S. Vogelgesang B, Eicheland racemic verapamil on Am Heart J 1985:109:210-

27. O’Reilly RA: Studies on the optical enantiomorphs farm in man. Clin Pharmacol Then 1974;16:348-54. 28.

5, Sheiner electrocarrelation-

Hull

pharmacodynamic 1978; 50: 1113-1123.

mt /

capsule.

Nagashima

effecfs

cal Then

in man;

R, O’Reilly the

RA. Levy

anticoagulant

C: Kinetics

action

of war-

of pharmacologic

of warfarin.

Clin

Pharma-

1969:10:22-35.

29. Meffin PJ, Winkle Response optimization with tocainide. Clin

RA,

Blaschke

TF,

Fitgerald

J,

of drug dosage: antiarrhythmic Pharmacol Then 1977;22:42-57.

Harrison

DC:

studies