Clinical Pharmacokinetics
`
`
`
`,
`iliird eolilion
`
`
`
`MALCOLM ROWLAND, PhD.
`‘
`
`
`
`
`
`
`
`
`Contep’rs and Applications
`
`Department ol Pharmacy
`
`Universily ol Manchester
`
`Manchester, England
`
`THOMAS N. TOZER, PhD.
`
`School of Pharmacy
`
`University ol Calilornia
`
`San Francisco, California
`
`
`
`A, Lea & Febiger Book
`
`
`
`
`
`é‘g LiPPINCOTTWILLlAMS a; WILKINS
`I, ' A Wolters Kluwér Company
`, Philadelphia ' Baltimore - New York ' London
`Buenos Aires - Hong Kong - Sydney - Tokyo
`
`BDSI V. RB PHARMACEUTICALS LTD
`IPR2014-00325
`
`Page 1
`
`Page 1
`
`RB Ex. 2026
`BDSI v. RB PHARMACEUTICALS LTD
`IPR2014-00325
`
`

`

`AVAVv
`V
`
`Executive Editor: Donna Balaa’o
`Developmental Editors: Frances Klass, Lisa Stead
`Production Manager.- laurz’e Forsytb
`Project Editor.- Roberl D. Magee
`
`Copyright © 1995
`Lippincott Williams 5c Wilkins
`530 Walnut Street
`Philadelphia, Pennsylvania l9l06-3621 USA
`
`
`
`All rights reserved. This book is protected by copyright, No part of this book may be reproduced in any
`form or by any means, including photocopying, or utilized by any information storage and retrieval system
`without written permission from the copyright owner.
`
`Accurate indications, adverse reactions, and dosage schedules for drugs are provided in this book, but it is
`possible they may change. The reader is urged to review the package information data of the manufacturers
`of the medications mentioned.
`
`Printed in the United States of America
`
`First Edition 1980
`
`Library of Congress Cataloging—in-Publication Data
`
`1213141516171819 20
`
`Rowland, Malcolm.
`Clinical Phannacokinetics : concepts and applications / Malcolm
`Rowland, Thomas N. Tozer. — 3rd ed.
`p.
`cm.
`“A Lea & Febiger Book.”
`Includes bibliographical references and index.
`ISBN 978—0-683-07404-8
`ISBN 0—683—07404—0
`
`1. Pharmacokinetics.
`11. Title.
`
`2. Chemotherapy.
`
`I. Tozer, Thomas N.
`
`2. Drug Therapy.
`
`QV 58 R883c 1994]
`
`[DNLM: 1. Pharmacokinetics.
`RM301.S.R68
`1994
`615.7—dc20
`DNLM/DLC
`
`for Library of Congress
`
`94—26305CIP
`
`7776 Publishers have made every eflort to trace the copyright boldersjor borrowed material. If they have in.—
`acluerterttly overlooked any, they will be pleased to ma/ee the necessary arrangements at tbefirst opportunity.
`
`Page 2
`
`Page 2
`
`

`

`
`
`VARIABILITY
`
`
`
`I
`
`itcrtVEs
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Th1e reader will beable to:I
`
`List six motor sources 0fvariabilityIn drUQ response
` 2 Evaluate whether variabilityIn drug response iscaused by a variabilityIn pharmacokinet'Ics
`
`pharmacodynamiIcs, or both, given response and pharmacokineticdata.
`
` 3. State why variability around the mean and shape of thefrequency distribution histogram of
`
`a parameter are as important as the mean itsell.
`
`
` 4, Explain how variability'In hepatic enzyme activity manifests itselfIn variabilityIn both phar-
`
`macokinetic parameters and plateau plasma drug concentrations for drUgs of highand low
`
`
`hepatic extraction ratios
`
`
` 5. Suggestan approach for initiating a dosage regimen foran individual patient, given patient '
`
`population pharmacokinetic data and the individualsmeasurable characteristics.
`
`
`
`
`
`Thus far, the assumption has been made that all people are alike. True, as a species, humans
`
`
`are reasonably homogeneous, but differences among people do exist including their re-
`
`
`sponsiveness to drugs. Accordingly, there is a frequent need to tailor drug administration
`
`to the individual patient. A failure to do so can lead to ineffective therapy in some patients
`and toxicity in others.
`This section of the book is devoted to individual drug therapy. A broad overview of the
`subject is presented in this chapter. Evidence for and causes of variation in drug response,
`and approaches toward individualizing drug therapy are examined. Subsequent chapters
`deal in much greater detail with genetics (Chap. 14), age and weight (Chap. 15), disease
`(Chap. 16), interactions between drugs within the body (Chap. 17), and monitoring of
`plasma concentration of a drug as a guide to individualizing drug therapy (Chap. 18).
`Before proceeding, a distinction must be made between an individual and the popula—
`tion. Consider, e.g., the results of a study designed to examine the contribution of an acute
`disease to variability in drug response. Suppose, of 30 patients studied during and after
`recovery, only 2 showed a substantial difference in response; in the remainder the differ—
`ence was insignificant. Viewed as a whole, the disease would not be considered as a sig—
`nificant source of variability, but to the two affected patients it would. Moreover, to avoid
`toxicity, the dosage regimen of the drug may need to be reduced in these two patients
`during the disease. The lesson is clear: Average data are useful as a guide; but ultimately,
`information pertaining to the individual patient is all—important.
`On a similar but broader point, substantial differences in response to most drugs exist
`among patients. Such z‘nten‘ndividual variability is often reflected by a variety of marketed
`dose strengths of a drug. Because variability in response within a subject (immindividual)
`is generally much smaller than inte'rindividual variability, once well—established, there is
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`203
`
`Page 3
`
`
`
`
`Page 3
`
`

`

`204
`
`VARIABILITY
`
`CHAPTER I 3
`
`usually little need to subsequently adjust an individuals dosage regimen Clearly, il‘intmin-
`dividual variability were large and unpredictable, trying to titrate dosage for an individual
`would be an extremely difficult task, particularly for drugs with narrow therapeutic win—
`dows. Stated differently, a drug that exhibits a high intmindividual variability in pharma—
`eokinetics can be prescribed only if it has a wide therapeutic window.
`
`EXPRESSIONS 0F INDIVIDUAL DIFFERENCES
`
`Evidence for interindividual differences in drug response comes from several sources.
`Variability in the dosage required to produce a given response is illustrated in Figure 17 5
`(Chap. 1), which shows the wide range in the daily dose of warfarin needed to produce a
`similar degree of anticoagulant control. Variability in the intensity of response with time
`to a set dose is seen with the neuromuscular agent doxacurium (Fig. 13—1). As illustrated
`in Figs. 13—2 and 13—3, which show frequency distribution histograms of the plateau plasma
`concentration of the antidepressant drug nortriptyline, to a defined daily dose of the drug
`and the plateau unbound plasma concentration of S—warfarin required to produce a similar
`degree of anticoagulant control, variability exists in both pharmacolcinetics and pharma—
`codynamics. Variability in pharmaeokinetics was also illustrated by the wide scatter in the
`plateau plasma concentration of phenytoin seen following various daily doses of this drug
`(see Fig. 1—6, Chap. 1).
`
`The Need for Models
`
`The magnitude and relative contribution of pharmacokinetics and pharmacodynamics to
`variability in response to a given dosage within a patient population vary with the drug and,
`to some extent, the condition being treated. For example, with a nonsteroidal anti—inflame
`matory drug, the relative contribution of pharmacodynamic variability may be different
`when the endpoint is the relief of a headache than when it is the relief from chronic aches
`and pains associated with inflamed joints. In clinical practice, attempts to assign the relative
`contribution to pharmacokinetics and pharmacodynamies may be made based on direct
`observations of plasma concentration and response. The assignment could be strongly in--
`fluenced, however, by the timing of the observations and the magnitude of the response,
`
`
`
`
`
`NumberofPatients
`
`
`
`Fig. 13—1. The degree of neuromus-
`cular block with time after an iv. bolus
`dose of 0.01 mg/kg doxacurium to pa-
`tients varies widely. (1 mg/L = 0.97 HM)
`(Modified from Sehmith, V.D., Fiedler-
`Kelly, l, AboueDonia, M., Huffman,
`(3.8., and Grasela, T.H.: Population
`pharmacodynamics of doxacumin. Clin.
`Pharmacol. Ther., 525287536, 1992.)
`
`_,
`
`
`
`PercentBlock
`
`I
`
`80
`
`F
`
`160
`Minutes
`
`‘I
`
`240
`
`,
`
`320
`
`Page 4
`
`Page 4
`
`

`

`CHAPIER 13
`
`CHAPTER 1 3
`
`VARlABlLITY
`
`205
`
`as illustrated in Fig. 13—4. Here, a drug that displays little inteipatient variability in C"m,
`tum and in maximum effect, but large variability in half—life and concentration needed to
`produce 50% maximum response, is given orally at two doses, one that achieves close to
`maximal response in all patients and one that does not. At the higher dose, observations
`made at Cum would suggest little variability in either concentration or pharmaeodynamicS,
`with perhaps a greater assignment of variability to the former, as variation in plasma con—
`centration produces relatively little change in response. At later times after this higher
`dose, substantial variability is observed in both concentration and response. In contrast, for
`
`[BI]. Clearly, ifz'ntrain-
`sage For an individual
`.‘I'OVV therapeutic Win~
`variability in pharma—
`V.
`
`From several sources.
`
`rstrated in Figure 1—5
`needed to produce a
`)l‘ response with time
`. 13—1). As illustrated
`
`of the plateau plasma
`:laily close ol‘ the drug
`d to produce a similar
`kinetics and pharma—
`16 Wide scatter in the
`
`Lily doses of this drug
`
`)harmaeodynamies to
`try with the drug and,
`isteroidal anti—inflam—
`
`lity may be different
`if from chronic aches
`
`; to assign the relative
`iade based on direct
`
`could be strongly in—
`:udc of the response,
`
`
`
`
`
`g
`g
`E
`Z
`
`,
`
`
`
`
`
`
`11
`
`
`W
`
`[j
`,
`0.3
`0.2
`0.1
`0
`Plasma Nortriptyline Concentration (mg/L)
`
`A
`
`99 —
`
`A 90 —
`
`.'
`
`.
`
`.x
`
`B
`
`.
`
`_ '
`
`.
`
`°
`
`g 50 -
`g
`“g 10 1
`8
`5‘:
`
`M
`
`1
`
`1
`I
`0.1
`0.05
`0.01
`Plasma Nortrlptyline Concentration
`(mg/L, log scale)
`
`r
`0.5
`
`Fig. 13—2. A, The plateau plasma concentration of nortriptyline varies widely in 263 patients receiving a regimen
`of 25 mg nortriptyline orally three times daily. B, The concentrations are logenormally distributed, as seen from
`the straight line, when the percentiles of the cumulative number of patients are plotted on probit scale against
`the logarithm of the concentration. (1 mg/l, = 3.8 MM) (Redrawn and calculated from Sjoqvist, F., Borga, 0.,
`and Orme, M.L.E.: Fundamentals ofclinical pharmacology. In Drug Treatment. Edited by CS. Avery. Edinburgh,
`Churchill Livingstone, l976, pp. 1—42.)
`
`
`
`
`
`0
`
`2
`
`4
`
`6
`
`Unbound Plasma S-Warfarin
`Concentration (ug/L)
`
`240
`
`320
`
`y active S—warfarin associated with a similar
`Fig. 13—3. The unbound plateau concentration of the predominatel
`ely among a group of 38 patients receiving racemic warfaiin. (1 mg/L = 3.3
`degree of anticoagulation, varies Wid
`uM) (Adapted from Chan, E., McLachlan, A.]., Pegg, M., Mackay, AD, Cole, R. B., and Rowland, M.: Disposition
`of‘ warfarin enantiomers
`and metabolites in patients during multiple dosing. Br. ] Clin. Pharrnacol., 37563—569,
`l 994.
`
`/ .
`
`/
`
`tes
`
`
`
`Page 5
`
`Page 5
`
`

`

`VARIABIl ITY
`
`CHAPTER l 3
`
`
`
`PlasmaDrugConcentration
`
`53 m
`
`.0 M on
`
`O '__.L 01
`
`O ;A
`
`
`
`Response/(MaximumResponse)
`
`18
`
`24
`
`12
`Hours
`
`18
`
`24
`
`12
`Hours
`
`therapy. To illustrate this statement consider the frequency distributions in clearance oil
`
`
`Fig. 13—4. The interindividual variability in concentration and response varies with dose and time of observation.
`Shown are plasma concentrations (left) and responses (right) following large and small doses of a drug that displays
`little intcrpatient variability in Cmax, tum and maximum response, but large inter-patient variability in halfelife and
`concentration needed to produce 50% maximum response. High dose (top): at tum, the maximum response in all
`patients is produced with little variability in either Cum or response. Greater variability in concentration and
`response is seen at later times. Low dose (bottom): at tmm, variability in CW“. is still low, but that in response is
`now considerable.
`
`the lower dose, at tum there is still little interpatient variability in Cum, but now there is
`considerable variability in response. This dependence on dose and time in the assignment
`of variability is minimized by expressing variability not in terms of observations but rather
`in terms of the parameter values defining pharmacokinetics and pharmacodyniaxnics, that
`is, in F, kn, CL, and V for pharmacokinetics, and in maximal response, concentration to
`achieve 50% of the maximum response, and the factor defining the steepness of the con—
`centration~response relationship for pharmacodynamics (Chap. 20, Pharmacologic Re—
`sponse). Once variability in these parameters is defined, the expected variability in con
`centration and response within the patient population associated with given dosage
`regimens can be estimated. The accuracy of the models defining pharmacokinetics and
`pharmacodynamics is obviously critical to an understanding of variability in patient re—
`sponse. Where appropriate, these models should incorporate such factors as protein bind—
`ing, active metabolites, and tolerance.
`
`DESCRIBING VARIABILITY
`
`Knowing how a particular parameter varies within the patient population is important in
`
`Page 6
`
`Page 6
`
`

`

`
`CHAPl [R l 3
`VARIABILITY
`207
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`CHAPTER 13
`
`\ LargeDose
`
`18
`
`'
`
`24
`
`SmaH
`Dose
`
`l//
`
`18
`
`24
`
`and time of observation.
`scs of a drug that displays
`variability in half—life and
`maximum response in all
`ity in concentration and
`v, but that in response is
`
`u, but now there is
`e in the assignment
`arvations but rather
`
`nacodynamics, that
`:e, concentration to
`
`eepness of the con—
`Pharmacologic Re—
`i variability in con—
`with given dosage
`armacokinetics and
`
`aility in patient re—
`ors as protein bind—
`
`the three hypothetical drugs shown in Fig. 13—5. The mean, or central tendency, for all
`three drugs is the same, but the variability about the mean is very different. For Drugs A
`and B, the distribution is unimodal and normal; here the mean represents the typical value
`of clearance expected in the population. As variability about the mean is much greater for
`Drug B than for Drug A, one has much less confidence that the mean of Drug B applies
`to an individual patient. For Drug C, distribution in clearance is bimodal, signifying that
`there are two major groups within the population: those with high and low clearances.
`Obviously, in this case, the mean is one of the most unlikely values to be Jfound in this
`population.
`Generally, distributions of pharmacokinetic parameters or observations are unimodal
`rather than polymodal, and they are often skewed rather than normal, as seen, e.g., in the
`frequency distribution of plateau plasma concentrations of nortriptyline (Fig. 1372A). A
`more symmetrical distribution is often obtained with a logaiitlnnic transformation of the
`parameter; such distributions are said to be log—normal. A common method of examining
`for log—normal distribution is to plot the cumulative frequency, or percentile, on a probit
`scale against the logarithm of the variable The distribution is taken to be log—normal it the
`points lie on a straight line, As can be seen in Fig. 13—2B, this is the ease for the plateau
`plasma concentration of nortriptyline. In such cases the median, or value above and below
`which there are equal numbers, (litters from the mean. For nortriptyline, examination of
`Fig. 13—213 indicates that the median concentration is 0.05 mg/L, which is less than the
`average value of 0.069 mg/L.
`A comment on the quantitation of variability is needed here. Variance is a measure of
`the deviations of the observations about the mean; it is defined as the sum of the squares
`of these deviations. While useful to convey variability within a particular set of observations,
`variance does not allow ready comparison of variability across sets of observations of dif—
`ferent magnitude. Suppose, e.g., clearance in an individual is 50 mL/min and the mean is
`
`100 mL/min; the squared deviation is 2500 (1nL/min)2. If instead clearance had been
`
`quoted in L/min, the squared deviation would be (0.05 — 0.1)2, or 0.0025 (L/min)2. Coeffie
`
`cient of variation, which expresses variability with respect to the mean value, overcomes this
`
`problem, Specifically, it is the square root of variance (the standard deviation) normalized to
`
`the mean. In the example above, the deviation normalized to the mean is 0.5 and is independ-
`ent of the units ofclearance. Furthermore, a large coellieient oi‘variation now always signifies
`
`a high degree ot'variability. Subsequently, in the book, high and low variability refer to dis—
`
`tributions that have high and low coefficients of variation, respectively.
`
`
`
`A
`Fig. 13—5. As the frequency distributions for the
`clearance of three hypothetical drugs (A, B, C) show,
`
`
`it is as important to define variability around the mean
`and the shape of the frequency distribution curve as
`
`it is to define the mean itself.
`
`
`
`
`Frequency
`
`
`
`
`
`tion is important in
`ons in clearance of
`
`
`
`Clearance (arbitrary units)
`
`
`Page?
`
`Page 7
`
`

`

`the country.
`
`,
`,
`Noncomplionce
`Route of administration
`7', H
`7
`,
`,
`;
`’y
`'
`
`,3
`
`Food 5
`'
`I
`,
`,
`'
`_
`r
`'
`VPOllD’to'nts'
`I
`,
`,r
`,
`Time/of day an seoson
`'
`" :
`" '
`' " '
`.
`,,
`,
`
`208
`
`VARlABlLI iY
`
`CHAPTER l 3
`
`WHY PEOPLE DIFFER
`
`The reasons why people differ in their responsiveness to drugs in medicinal products are
`manifold and include, in general order of importance, genetics, disease, age, drugs given
`concomitantly, and a variety of environmental factors. Although inheritance accounts for a
`substantial part of the differences in response among individuals, much of this variability
`is largely unpredictable. increasingly, however, this source of variability, particularly that
`related to drug metabolism, is being understood and made more predictable using the tools
`of molecular biology (Chap. 14, Genetics).
`Disease can be an added source of variation in drug response. Usual dosage regimens
`may need to be modified substantially in patients with renal function impairment, hepatic
`disorders, congestive cardiac failure, thyroid disorders, gastrointestinal disorders, and other
`diseases. The modification may apply to the drug being used to treat the specific disease
`but may apply equally well to other drugs the patient is receiving. For example, to prevent
`excessive accumulation and so reduce the risk of toxicity, the dosage of the antibiotic
`gentamicin used to treat a pleural infection of a patient must be reduced if the patient also
`has compromised renal function. Similarly, hyperthyroidic patients require higher than
`usual doses of digoxin, a drug used to improve cardiac efficiency. Moreover, a modification
`in dosage may arise not only from the direct impairment of a diseased organ but also from
`secondary events that accompany the disease. Drug metabolism, e. g., may be modified in
`patients with renal disease; plasma and tissue binding of drugs may be altered in patients
`with uremia and hepatic disorders.
`Age, weight, and concomitantly administered drugs are important because they are
`sources of variability that can be taken into account. Genderdinked differences in hormonal
`balance, body composition, and activity of certain enzymes manifest themselves in differ-
`ences in both pharmacokinetics and responsiveness, but overall, the effect of gender is
`small.
`
`Table 13—1 lists examples of additional factors known to contribute to variability in drug
`response. Perhaps the most important factor is noncompliance. Noncompliancc includes
`the taking of drug at the wrong time, the omission or supplementation of prescribed dose,
`and the stopping of therapy, either because the patient begins to feel better or because of
`development of side—effects that the patient considers unacceptable. Whatever the reason,
`these problems lie in the area of patient counselling and education. Occasionally, plasma
`concentration data are used as an objective measure of noncompliance.
`Pharmaceutical formulation and the process used to manufacture a product can be
`important as both can affect the rate of release, and hence entiy, into the body (Chap. 9).
`
`Table I 3-1 . Addilionul Facials Known to Coniribuie Io Vuriubilily
`in Drug Response
`
`FACTORS OBSERVATIONS AND REMARKS
`
`'
`
`.
`
`,
`
`,
`
`,
`
`.
`,
`tQCOtion ,
`_
`'
`,
`
`f
`
`,
`
`,
`
`,
`
`,
`
`.1
`
`xi
`,
`
`.'
`
`,
`
`'
`
`,,
`
`A motor problem in clinical practice; solution lies in ootient education.
`Patient response con vary on changing the route of administration. Not only
`phormocokinetics of drug but also metabolite concentrations can chonge.
`Rote ond occosionolly extent of absorption ore effected by eating. Effects
`depend on composition of food. Severe protein restriction may reduce the
`rote of drug metabolism.
`Drug effects ore often less in smokers and workers occupationally exposed to
`pesticides; o resultolenhoncecl drug metabolism.
`Diurnol voriotions ore seen in phormocokinetics ond in drug response. These
`effects have been sufficiently important to lead to the development of 0 new
`subiect, chronophormocology.
`Dose requirements of some drugs differ between patients living in town and in
`
`Page 8
`
`Page 8
`
`

`

`
`
`
`
`
`
`CHAPTER 13
`
`VARIABILITY
`
`
`209
`
`A well—designed formulation diminishes the degree of variability in the release character—
`istics of a drug in nine. Good manufacturing practice, with careful control of the process
`variables, ensures the manufacture of a reliable product. Drugs are given enterally, topi—
`cally, parenterally, and by inhalation, Route of administration not only can affect the con—
`centration locally and systemically but also can alter the systemic concentration of metab—
`olite compared with that of drug (Chap. 21). All these factors can profoundly affect the
`response to a given dose or regimen.
`Food, particularly fat, slows gastric emptying and so decreases the rate of drug absorp
`tion. Oral bioavailability is not usually affected by food, but there are many exceptions to
`this statement. Food is a complex mixture of chemicals, each potentially capable of inter—
`acting With drugs. Recall from Chap. 9, e.g., that the oral bioavailability of tetracycline is
`reduced when taken with milk, partly because of the formation of an insoluble complex
`with calcium. Recall also that a slowing of gastric emptying may increase the oral bioavail—
`ability of a sparingly soluble drug, such as griseofulvin. Diet may also affect drug metab—
`olism. Enzyme synthesis is ultimately dependent on protein intake. When protein intake
`is severely reduced for prolonged periods, particularly because of an imbalanced diet, drug
`metabolism may be impaired. Conversely, a high protein intake may cause enzyme induc—
`tion.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`CHAPTER l3
`
`dicinal products are
`se, age, drugs given
`tance accounts for a
`
`Ch of this variability
`ity, particularly that
`table using the tools
`
`ial dosage regimens
`mpairment, hepatic
`disorders, and other
`the specific disease
`example, to prevent
`ge of the antibiotic
`id if the patient also
`'equire higher than
`over, a modification
`organ but also from
`may be modified in
`2 altered in patients
`
`t because they are
`:rences in hormonal
`1emselves in differ,
`
`effect of gender is
`
`0 variability in drug
`ompliance includes
`of prescribed dose,
`)etter or because of
`'hatever the reason,
`tceasionally, plasma
`
`= a product can be
`the body (Chap. 9).
`
`Iv
`
`itieni education.
`ministration. Not only
`nirotions con Change,
`i by eating. Effects
`then may reduce the
`
`Jpotionolly exposed to
`
`drug response. These
`development of or new
`
`its living in town and in
`
`Chronopharmacology is the study of the influence of time on drug response. Many
`endogenous substances, e.g., hormones, are known to undergo cyclic changes in concen—
`tration in plasma and tissue with time. The amplitude of the change in concentration varies
`among substances. The period of the cycle is often diurnal, approximately 24 hr, although
`there may be both shorter and longer cycles upon which the daily one is superimposed.
`The menstrual cycle and seasonal variations in the concentrations of some endogenous
`substances are examples of cycles with a long period. Drug responses may therefore change
`with time of day, day of the month, or season of the year. Particular note of this phenom—
`enon is taken in cancer chemotherapy. Many chemotherapeutic agents have very narrow
`margins of safety and are given in combination. Appropriate phasing in the timing of ad—
`ministration of each drug during the day can improve the margin of safety.
`Cigarette smoking tends to reduce clinical and toxic effects of some drugs, including
`chlordiazepoxide, diazepam, and theophylline, The drugs affected are extensively metab—
`olized by hepatic oxidation; induction of the drug~metabolizing enzymes is the likely cause.
`Many environmental pollutants exist in higher concentrations in the city than in the country;
`
`they can also stimulate synthesis of hepatic metabolic enzymes.
`
`
`identiiying the Sources of Variability
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`In practice, all the above-mentioned factors can contribute to observed variability in re—
`sponse, and care must be taken to ensure an appropriate conclusion is reached when trying
`to assign causes of variability. Consider, e.g., the data displayed in Fig. 13—6 which show
`the half—lives of phenylbutazone, a once widely used drug, in healthy subjects and in patients
`with hepatic disease (primarily cirrhosis). initially, no difference was revealed between the
`two groups, except for a greater variability in the half—life among the patients with hepatic
`disease (Fig. 13—6A). When, however, both groups were further subdivided on the basis
`of whether they received other drugs, a clearer picture emerged (Fig. 1376B). Of those
`receiving no other drugs, patients with hepatic disease handled phenylbutazone more
`slowly than did healthy subjects. Evidently, some of the other drugs received hasten phen-
`ylbutazone elimination.
`Various strategies can be employed to identify sources of variability in response. The
`Classic design is one in which as many of the variables as possible are fixed, apart from the
`one of interest. For example, to test if renal disease affects the pharmacokinetics of a drug,
`
`
`
`Page 9
`
`Page 9
`
`

`

`VARIABILlTY
`
`CHAPTER l 3
`
`Normal
`No Drugs Drugs
`
`B
`
`Hepatic
`Disease
`No Drugs Drugs
`
`Hepatic
`Disease
`
`. 197
`,- 162
`144
`ago 0
`
`
`
`PhenylbutazoneHalf-life
`
`(hf)
`
`U‘I O
`
`aspire-5;:
`1:0..0‘9'00
`
`o
`
`3:»
`
`
`
`PhenylbutazoneHalt-lite
`
`(hr)
`
`01 O
`
`
`
`I
`
`siderable overlap in the plasma concentrations among the groups. Thus, individuals from
`
`
`0.09
`
`(O01
`
`36
`
`19
`
`34
`
`61
`
`Number of Subjects
`
`Number of Subjects
`
`Fig. 13—6 A, No difference is seen between the average half—life of phenylbutazone (horizontal line) in normal
`subjects (black points) and that in patients with hepatic disease (colored points). 13, After separating those who
`take other drugs (light points) from those who do not (dark points), the prolonged elimination of phenylbutazone
`in patients with hepatic disease becomes evident. In both groups the half—life tends to be shorter when other
`drugs are taken concurrently. (Redrawn from Levi, A.]., Sherlock, S., and Walker, D.: Phenylbutazone and iso—
`niazid metabolism in patients with liver disease in relation to previous drug therapy. Lancet, 1127571279, 1968.)
`
`all other factors such as age, gender, other drugs, and diet should be held constant. The
`ideal would be a longitudinal cross—over design in which each patient acts as his or her own
`control. This design is often not possible, however. The patient with renal disease is gen-
`erally not available for study prior to the disease, and renal disease is generally irreversible.
`The penalty for deviating from such a design is greater variability with loss of efficiency,
`such that many more patients are needed to allow a firm conclusion to be made about the
`contribution of a factor to variability. The benefit of loosening the design, however, is that
`many patients who might otherwise be excluded can be a part of the study. In this category,
`e.g, are elderly patients suffering from several diseases and requiring many drugs, including
`the one of interest. Care must still be taken, however, to ensure that a sufficient number
`of patients are included with each of the attributes or conditions of interest.
`
`DEFINING THE DOSE—RESPONSE RELATIONSHIP
`
`Variability has an important bearing on the estimation of doseeresponse relationships in
`clinical trials. A common procedure is to divide patients into several groups, each group
`receiving a different dose of drug such as 5, 10, or 20 mg. An attempt to establish a dose;
`response relationship is then made on the mean data for each group, using variability within
`groups to test for levels of statistical significance. A problem arises when much of the
`variability between dose and response resides in pharmacolcinetics such that there is con—
`
`Page 10
`
`Page 10
`
`

`

`
`
`
`
`
`
`
`c; tAP’IER 13
`VARIABiLITY
`21 1
`
`
`
`the high- and low—dose groups can have the same plasma concentration (and response),
`namely, those in the low—dose group with a low Clearance and those in the high—dose group
`
`
`With a high clearance of the drug. The overall effect, by increasing variability within each
`
`
`group, is to weaken the ability to detect a dose—response relationship.
`
`
`7 One solution is to increase the number of subjects in each group to reduce the uncer—
`
`
`tainty of estimating the mean response at each dose level. Here, the problem is often one
`
`
`of not knowing in advance how many subjects would be needed in the trial, as well as the
`
`
`added expense of an increased number of subjects. Another solution is to expose each
`
`
`patient to several dose levels of the drug. This last solution has the distinct advantage of
`
`
`not only increasing the chances of establishing a dose—response relationship, but also of
`
`
`providing an estimate of interpatient variability in the relationship. Unfortunately, in prac—
`
`
`tice, this design is not always possible, especially for drugs for which the full effect only
`
`
`occurs after several months or longer into drug administration. A third solution is the
`concentration—controlled clinical trial. In this approach, the pharmacokinetics of the drug
`is first evaluated in the patient cohort and then, based on this information, doses are
`adjusted so that the plasma concentration in each patient lies in one of several tightly
`defined bands. This more elaborate, and sometimes more expensive, design enables much
`
`
`clearer statements to be made about the concentrationresponse relationship and about
`
`
`interpatient variability in pharmacokineties. However, it may have limited utility for dose
`
`
`recommendations, if a poor correlation is found between plasma drug concentration and
`
`
`response. Many other designs, varying in complexity, each with advantages and disadvan—
`tages, can be envisaged. In all cases, variability is a central issue.
`
`
`iiiNETIC MANIFESTATIONS
`
`
`Considerable variability in enzymatic activity and, to a lesser extent, in plasma and tissue
`binding exists even among healthy individuals. How such variability manifests itself, in
`
`
`pharmacokinetic parameters and in such measurements as plateau plasma concentration,
`
`
`depends on the hepatic extraction ratio and route of administration of the drug. For ex—
`
`
`ample, the large interindividual variability in half—life of theophylline (Fig. 13—7) can be
`
`
`explained primarily by variations in hepatic enzyme activity, probably associated with var—
`
`
`iations in the amounts of the enzymes responsible for metabolism of this compound. This
`conclusion is based on theophylline being predominantly metabolized in the liver, having
`
`
`a low extraction ratio, and being only moderately bound to plasma and tissue components.
`
`
`In contrast, such a high degree of variability in enzymatic activity is expected to be masked
`
`
`in the clearance of a drug having a high hepatic extraction ratio, because clearance tends
`
`
`to be perfusion rate—limited and hepatic blood flow is relatively constant among healthy
`
`
`individuals. Moreover, unless plasma and tissue binding are highly variable, volume of
`
`
`distribution, and hence disposition kinetics, of such a drug are much the same for all healthy
`
`
`individuals. This is so for propranolol (Fig. 13—8) a drug of high hepatic clearance.
`
`
`As described in Chap. 11., when considering induction and inhibition, changes in hepatic
`enzyme activity result in variations in oral bioavailability for a drug with a high hepatic
`
`
`extraction ratio. Accordingly, with subsequent disposition being controlled by hepatic per—
`fusion, a series of similarly shaped plasma drug concentration-time profiles, but reaching
`
`
`different peak concentrations, should be seen among individuals with varying enzyme ac—
`tivity receiving