- Email: [email protected]
Variability in the human drug response
THROMBOSTS RESEARCH,Supplement TV; 3-15, 1983 0049-3848183 $3.00 + .OO Printed in the USA. Copyright (c) 1983 Pergamon Press Ltd. All rights
reserved.
VARIABILITY IN THE HUMAN DRUGRESPONSE G.D. Sweeney, M.B.,
Ch.B.,
Ph.d
Department of Medicine McMaster University L8N 325 Hamilton, Ontario
ABSTRACT Variations in response to drugs may be pharmacodynamic, implying inter-individual differences in the response of receptors in equal that concentrations of drug, or pharmacokinetic, imp1ying individuals receiving the same dose of drug will have different Either type of concentrations of drug in their body fluids. Variations in receptor variation can be inherited or acquired. sensitivity do occur but few instances, inherited or acquired, have response well clinical If the dose documented relevance. relationship for the drug in question is not steep, or if the therapeutic index is low, drug concentration in the region of the receptor will not be critical and causes of kinetic variation are However, it is the many unlikely to be clinically significant. The se causes of kinetic variation which are best described. include effects due to drug formulation and changes in the If a absorption, distribution, metabolism and excretion of drugs. consideration of dynamics suggests that drug concentration will and prediction of determine therapeutic efficacy, analysis Prediction requires variability due to these factors is desirable. an accurate description of the system but commonly used The pharmacokinetic models may fail when prediction is a goal. variables, volume of distribution (Vd) and rate constant. of elimination (K ) are hybrid in that they arise from the interaction of patient anfi drug characteristics. Important events including binding macromol ecu1 ar and altered blood flow be may not represented. More data is required to determine the clinical analyticai significance of pharmacodynamic variation but better tools are required to deal with kinetic variation when this is important. Specifically, pharmacokinetic models should represent physiological variables and levels of unbound drug in body fluids should receive greater emphasis.
Key Words:
pharmacokinetics, binding
pharmacodynamics, 3
response,
variable,
protein
4
VARIABILITY IN HUMAN DRUGRBSPONSE
Suppl.
TV, 1983
INTRODUCTION The double-blind controlled clinical trial is widely regarded as providing reliable evidence that a drug, prescribed for a patient, will act in a predictable manner, provided that indications and usage match those of the trial. is constrained by the statistical However, such predictability evaluation of the outcome of the trial. Reasons for a variable response must to this problem considered. This paper is be examined and solutions concerned with the sources of variation subsequent to administration of compliance and placebo effects are not considered. drugs: Components of individual variation can be subdivided: pharmacodynamic imply drug receptors which respond in an unusual deviations from ‘normal’ to usual levels of drug while pharmacokinetic fashion despite exposure variation implies that normal receptors are exposed to unusual concentrations Both kinetic and dynamic deviations from normal behaviour may be of drug. genetically determined or acquired and objective indices of drug effect are essential if the problem is to be studied and resolved. In clinical practice, non-invasive indices may be lacking and there is a temptation to substitute measurements of drug level for objective indices of effect. this eliminates the receptor from study and thus precludes However, consideration of pharmacodynamic variations. In interpreting clinical trials there is a tendency to regard deviations abnormality may indeed be defined in from the mean as abnormal. Clinically, this sense (e.g. deviation by more than an arbitrary coefficient of variation fran the mean) but it is more important to regard variation in biology as the norm and to define its determinants in order to prevent it from interfering wih therapeutic goals. A number of inherited abnormalities are Inherited Dynamic Variation. However, there known which affect the disposition and metabolism of drugs. are few documented instances of inherited abnormalities of receptor proteins action at this level. Resistance of rats to the which alter drug rodenticidal action of Warfarin [(3- acetonyl benzyl)-4-hydroxycoumarinl was inherited and in 1964 O’Reilly et al. (1) described a human kindred with an The propositus required a daily dose 50 inherited resistance to this drug. This group has reported a similar standard deviations above the mean. they excluded differences in pharmacokinetic finding in a second kindred: behaviour of Warfarin as the basis for the difference, and provided evidence had a reduced affinity for Vitamin K and a that the hapatic receptor relatively greater reduction in receptor affinity to Warfarin (2). Requirements for detection of receptor mutations as a basis for (a) the means to quantitatively assess variations in a drug response are: (b) data concerning at the receptor; effect of drug the biological These circulating drug levels: and, (c) screening of many patients. limited clinical trials and would be conditions cannot be met during It is, extremely costly during later phases of preparation for marketing. however, likely that abnormalities of the receptors for biogenic amines, for and for other agents will be inhibitors, for opiates cycle-oxygenase described in time. A significant number of examples exist in which adverse effects of acute examples are: documented, are well interactions drug-receptor and malignant hyperpyrexia after exposure primaquine sensitivity, porphyria, These effects are dramatic and hard to overlook. to volatile anesthetics.
Suppl.
TV, 1983
VARTARTLTTYTN HUMAN DRUGRESPONSE
5
That the receptors involved mediate undesired effects does not alter the fact that these are examples of genetically determined variations in drug-receptor responses. Acquired deviations from mean of receptor Acquired Dynamic Variation. response are probably more common although data derives mostly from animal to chemical transmission experimentation. Drugs blocking may lead up-regulation of receptors and to hypersensitivity in a manner consistent Possibly covered by this model is with Canon’s law of denervation (3). reduced CNS sensitivity to sedative drugs in subjects who abuse these agents (including alcohol). Reduction in receptor numbers has been documented in insulin resistance, allergic including myasthenia gravis, di sea se states rhinitis and in asthmatics who have over-used agonist drugs for the bronchial In allergic rhinitis and asthma, the situation is beta-adrenoreceptor. It would be consistent with tachyphylaxis or receptor down-regulation. attractive to postulate a similar mechanism for reduced opiate sensitivity in - a dranatic acquired variation in response - but the molecular addiction basis for this state remains unclear (4). Kinetic Variation. In contrast to variations in receptor response, individual differences in pharmacokinetics are many and well documented: clinically the se, however, are not necessarily significant. If the therapeutic index of a drug is high (e.g. most penicillins) large fluctuations in circulating drug level are of no consequence, The shape of the dose-response curve is also important: if this is relatively flat, so that a wide range of drug concentration is not associated with a wide range of biological effect, drug dosage will not be critical. Further, if a drug effect is sought by blocking a receptor (e.g. cimetidine) and this is without known adverse effects, wide ranges in circulating drug levels will be tolerated. In contrast, if drug effects are sought on the steep portion of the dose response curve, and excessive receptor blockade must be avoided as in the case of beta-adrenoreceptor blockade causing cardiac failure, the situation is different. into the circulation from the site of Drug formulation, absorption distribution throughout the body, metabolism to inactive or administration, active products, and finally elimination, are all factors that determine the concentration of drug which actually reaches receptors. Each may be subject to variation and in each instance inherited as well as acquired processes can be considered, but before discussing these kinetic causes of variability, some commonly-used terms require definition. (Vd) is a statement concerning the fraction of Volume of distribution drug circulating in the plasma and is the ratio of the amount of drug in the body to the plasma drug concentration. Thus:
‘d
=
Total amount of drug in the body Concentration of drug in plasma
While the denominator of this fraction is readily measured, the numerator is only accessible via a mathematical mode. Thus, calculation of Vd depends upon the model chosen to represent drug distribution and elimination. Plasma drug clearance (Cl ) is that equivalent volume of plasma completely cleared of drug in unit e’ime. V and Clp, are critical terms because of their relationship to the rated of drug elimination. The rate constant, for elimination of drug (K,) is related to Vd:
6
VARIABILITY IN HUMAN DRUGRESPONSE
K
e
=
Suppl.
IV, 1983
“pl “d
Half-life (T-1/2) is calculated from K thus: T-l /2 = log 2/K . These are statements of a simplified approach toekinetics but they &cur&ely reflect the dependence of widely used variables upon the validity of simple models. Formulation. Formulation of oral drug is a complex process with the tablet containing many ingredients beside active principle and fillers. Around 1970 the issue of equivalence of different brands of generic drugs achieved considerable prominence following the demonstration that digoxin preparations from different sources differed by as much as 4 times in the amount of drug absorbed by the patient (5). Slower drug release from a larger crystalline form of nitrofurantoin is actually used in Macrodantin brand to decrease gastric irritation (6). has been used to describe The term l*bioavailabilityll Absorption. either that drua absorbed from the gastrointestinal tract or drug reaching the systemic circulation. With increasing appreciation of the role of the elimination” gut mucosa itself in inactivating drugs, the term “presystemic is preferable. Brodie (7) stressed the importance of ,physical factors including oil/water partition coefficient and capacity to ionise as acid or base, in absorption from the gut but of overriding importance is the huge surface area of the small intestine available even for the absorption of drugs ionised at the pH prevailing there. Concomitant food intake can be a potent source of variation in drug absorption. While absorption of oxprenolol, for example, is independent of the destruction of penicillin G in the stomach and the food intake (8), The acceleration of this in the presence of food is well recognized (9). complexity of variation in digoxin availability increased with the recent accelerated individuals demonstrate observation that about 10% of inactivation of the drug to cardioinactive reduced metabolites by gut flora (10). sites is of less interest as the Absorption from parenteral Slow and erratic is becoming less fashionable. intramuscular route lidocaine (11 1 and digoxin, absorption of diazepam, chlordiazepoxide, Injection technique may phenytoin from intramuscular sites is acknowledged. have shown that many contribute to this variability: Cockshott et al. intramuscular injections in the gluteal region are probably given into fat (12). When drug is initially introduced to the body it is Distribution. The circulation then distributes drug absorbed to the vascular compartment. High1 y per fused organs to all organs at a rate determined by blood flow. kidney and myocardium) can thus extract disproportionate amounts of (brain, drug during the distributive phase although poorly perfused tissue such as These events are not simulated fat may accumulate larger quantities later. A second important factor determining drug in simple pharmacokinetic models. distribution is the affinity of drugs for macromolecules including plasma proteins. The role of Binding of Drug to Plasma Protein. The role of binding in is far from settled. distri bpaces ution o drugs an in protein binding It is certainly not clear the extent to which alterations
Suppl.
TV, 1983
VARIABTLTTYTN HLJMAN DRUGRESPONSE
are clinically important as opposed to being merely biologically interesting. There is no doubt of the importance of the initial observation of Silverman the increased mortality in babies of mothers receiving a et al. reporting when two prophylactic antibiotic sulfonamide instead of chlor amphenicol regimes were compared (13) and the probable basis for this was kernicterus Similar due to the failure of albumin to transport bilirubin in the blood. arguments were used to explain hypoglycemic attacks induced by sulphaphenzole (14) and the potentiation of in tolbutamide-treated diabetics hypoprothrombinemia by chlorahydrate was explained by Warfarin-induced displacement of Warfarin from binding sites on albumin by the metabolite, trichloracetic acid (15). However, confusion clouds the effects of binding of drug to plasma protein on clearance of drug by kidneys and liver. Clearance at the glomerular filtraion rate (GFR) is of unbound drug only but both renal tubular and hepatic extraction processes may extract bound, as well as free, Bromsulphthalein (BSP) (which circulates extensively drug. bound to serum albumin) is largely excreted during a single pass through the liver (16). Despite this evidence that protein-bound BSP is cleared by the liver, Schmid showed (17) that artificially increasing the serum albumin decreased the elimination rate constant for BSP and he proposed that transport into the hepatocyte depended upon competition between binding sites within the cell and on albumin. Three mechanisms contribute to complex effects of protein binding on the concentration of free drug and the rate of drug elimination. First, free drug concentration may be increased temporarily by displacement from binding sites by a second drug. Second , decreased protein bindng must increase Vd and may thus decrease K . Third, when protein serves to transport drug to the liver (18), decreased protein binding may decrease Kd. Alterations in protein binding are not necessarily due to other drugs; in renal failure there is altered protein binding although this does not affect all drugs (21). An interesting example of variable binding by tissue, rather than plasma proteins, has been suggested in the recently described and clinically significant interaction between digoxin and quinidine. It was established (22) that digoxin in the presence of therapeutic doses of quinidine increased approximately 2.5-fold in the plasma and this was attributed to decreased renal clearance. A subsequent study failed to demonstrate a change in half-life, but confirmed the increase in plasma concentration. The overall effect was related to a decrease in renal clearance, but with displacement of digoxin by quinidine from tissue binding sites. Consequently, the volume of distribution decreased so that the expression for Ke remained relatively unchanged. Variations in albumin concentration over the range 0.3-0.9 mmoles/l occur in practice but have received little attention as a cause of altered drug kinetics. Such variations will have major effects only if the concentration of drug is of the same order of magnitude as that of albumin. For a drug such as dipyridmnole the therapeutic level of total drug is around ratio will be little affected by changes in plasma 4 uM and the bound/free albumin concentration. While there will be an uncertain effect of protein binding on T1/2 and thus dose schedules, dynamics - the receptor response must reflect free drug and this is measured too infrequently. The elegant studies of Piafsky (19) focused attention on similar problems due to cationic drugs being tightly bcouG;d toTiehe acute. phase reactant protein, alpha-l acid glycoprotein . affinity of dipyridamole for AG has been shown to be high (Kd = 1.6 uM, ref. 20).
7
VARIABILITY IN HUMAN DRUG RESPONSE
Suppl. IV, 1983
Subbarao et al. (20) reported that dipyridamole circulates over 98% bound to plasma protein but that this figure did not correlate well with AG concentration. This data suggests that the drug is extensively bound to albumin as well as to AG. Concentrations of dipyridamole and serum protein will be approximately 2 uM and 1 mM respectively: as noted above, binding under these conditions is not sensitive to protein concentration. However, a different situation may exist in the presence of salicylate at a concentration similar to the protein. Variation in kinetics associated with qualitative and quantitative changes in protein binding sites as well as with effects of drug interactions are large. Some have been shown to be clinicaily significant. Unfortunately, there is too little data to assess whether this is generally true. When a drug is r ported “98% bound to plasma protein”, at a tota -7 $ concentration of 1 ug/ml the concentration of free drug is only 20 ng/mland not long ago, technoloiy generally could not resolve a 20% change in this level. This is no longer true but data is needed. Also required are pharmacokinetic analyses able to simulate and (therefore predict) effects of altered protein binding on kinetics. Wilkinson and Shand (23) added an The Hepatic Extraction Ratio. important dimension to our understanding of individual variation when they developed equations to describe hepatic clearance of drugs in terms of blood An important conclusion of this flow and hepatic extraction ratio (HER). work was that for drugs efficiently extracted by liver from blood, clearance (and thus T1/2) for a constant Vd) would be a function of splanchnic blood flow. Variations of splanchnic blood flow under basal conditions in a healthy population have not been documented, but wide variations occur in individuals as circulatory beds adapt to different circumstances (24). Figure 1 demonstrates how such changes may correlate with changes in Variations at least of comparable magnitude can plasma levels of lidocaine. be anticipated in particular. in circulatory disorders and in shock by drugs and such changes may be Splanchnic blood flow is also altered There has recently been associated with important kinetic interactions. hepatic clearance of drugs caused by interest in reduced considerable with reports of cimetidine causing a clinically cimetidine on this basis, and warfarin theophylline of morphine, significant increase in effect clearance of reduced Feely et al. (29) have demonstrated (26-28). propranolol in the presence of cimetidine with evidence (based upon Clearance blood flow; of indocyanine green) that this is due to reduced hepatic Conventional pharmacokinetic this correlation is disputed (30). however, If the hepatic extraction ratio analyses do not represent these events. and blood flow can be equated, but for many drugs approaches unity, clearance While intuition might predict large changes in drug this does not hold. kinetics in cirrhosis, for example, the data does not bear this out except for drugs with high HER.
Suppl.
TV, 1983
VARTABTLTTYTN HUMAN DRUGRESPONSE
MAINTENANCE
INFUSION LIDOCAINE 2.16 mg/min
I
I
1
2
I
3
9
I
J
4
TIME IHOURS)
FIG. 1 A healthy male subject was infused I.V. with lidocaine for 2 hours then exercised at 70% of maximal oxygen uptake for 1 hour. Increasing circulating levels of drug were consistent with reduction in clearance due to reduced hepatic blood flow (from ref. 25 with permission). of drug metabolism occur between Metabolism. Variations in rates individuals and are probably the most significant determinant of variance in Pathways rate constants of elimination when normal volunteers are studied. of drug metabolism are genetically determined and the inheritance of some clinically significant variations has been reported (31). Rates of drug of metabolism may al so change within individual on account a single environmental factors or with other drugs. Conney (32) interactions initially drew attention to the potential of a large number of drugs to induce the activity of hepatic drug metabolizing enzymes and therefore accelerate the metabolism and inactivation of other agents. Initially it seemed likely that this would be both frequent and a clinicaly important cause of interactions and therapeutic drugs. with environmental agents However, these intra-individual effects on drug metabolizing enzymes are probably less than inter-individual differences. genetically-determined Important examples are accelerated inactivation of prednisone or the oral contraceptive pill in patients taking rifampin, phenobarbital or phenylbutazone. Other barbiturates, potent while inducers of drug metabolism, now have little place in therapy. Drugs which affect splanchnic blood flow may alter the rate of metabolism by the liver, but, as described in the preceding sect ion, only if this is flow-limited (i.e. if HER
10
VARIABILITY IN HUMAN DRUG RESPONSE
Suppl.
IV,
1983
approaches unity). 30th propranolol and cimetidine will reduce splanchnic blood flow and reduced clearance of benzodiazepines (33). theophylline (341, propranolol (351, and lidocaine (36) by one of these agents has been demonstrated. an alternative mechanism in the case of cimetidine is However, that the drug competes for metabolism by microsomal cytochrome in liver. Such competition would be a likely mechanism for drugs where the HER is low, for example, theophylline. Excretion. Altered renal function, which may be permanent or temporary, is a common and well documented cause of variation in drug clearance. Fortunately, the changes that occur can be predicted, or party predicted by measuring BUN, creatinine or a timed creatinine clearance. It is standard practice to predict and adjust dose schedules of e.g. aminoglycosides and digoxin along these lines. This is an example of the clinical value of good, predictive algorithms and of models which adequately simulate the events of interest. In the examples cited, GFR is the significant variable and prediction is accurate because of the close relationship between GFR and hSS is known about variation associated with altered renal tubular “Pl* processes. Controversy continues to surround the effects of aspirin and related non-steroid al, anti-inflammatory agents on renal function and renal It seems clear that aspirin does depress renal function clearance of drugs. with elevation of creatinine (37) but whether this is due to an effect on tubular transport or reduced renal blood flow is not resolved. There are many causes of The Clinical Role of Pharmacokinetics. altered absorption beclause variability in drug response and these include: of blood flow; motility: food; etc., altered distribution because of changes sites: alter ed the quality of quantity of macromolecular binding in metabolism because of changes in hepatic and altered extraction ratio; None of these excretion because of altered function. renal tubular physiological events can be represented in the commonly used mathematical altered pharmacokinetics. Under certain circumstances, an models of clearance rate due to changing physiological function related directly to an alteration in K . The most frequently presented pharmacokinetic model is the two compartmen f if tabulated pharmacokinetic open model (381, however, variables are consulted these tacitly assume that the drug is adequately It is contended that these described using a single compartment model. approaches to pharmacokinetics which may be referred to as “abstract” have That is not to say not proven significantly useful in clinical medicine. when volume of distribution and that for a given drug in a given patient, they have failed to predict the elimination rate constant are known, What the models do not do is provide the steady-state drug concentrations. clinician with a tool with which to anticipate kinetic variation in the The reason is not far to seek: each of the response of patients to drugs. variables used to describe compartmental models is a hybrid resulting from the interactions of the properties of the drug with the properties of the Either of these may change and alter the variable but it is patient. exceptional to be able to take a known physiological variation and use it directly to modify a pharmacokinetic expression to predict how the patient To do this we require models which take into account saturable will react. including simulation of protein binding in drug transport and processes, distribution, and which represent flow through various organs in a meaningful Some progress with such models has been made (39,40); mathematically way. are only possible using fairly large they are complicated and solutions but refining what may be called the “classical” models of computers, pharmacokinetics is unlikely to be helpful.
Suppl. TV, 1983
VARJABTLTTY TN HUMAN DRUG RESPONSE
11
In summary, variability in the human drug response is the rule ratner than the exception. It can be disservice to a drug to anticipate optimal responses from fixed dose therapy. It is unlikely that we have much to learn regarding variations in receptor sensitivity but much has been learned regarding variations in kinetic behaviour of drugs both singly and in combination. Our theoretical models for analysis of these events currently lag behind the empiric data available. Laboratories in particular must pay more attention to protein binding and accurate determination of free drug concentrations before it is reasonable to dismiss circulating drug data as unrelated to clinical response. Pharmacokineticists must similarly pay more attention to macromolecular binding events and there is a great need for mathematic solutions which can accept altered physiological parameters instead of working with hybrid parameters where drug and patient are inextricably mixed. REFERENCES 1.
O'REILLY, R.A., AGGELER, P.M., HOAG, M.S., LEONG, L.S., and KROPATKIN, M.L. Hereditary transmission of exceptional resistance to coumarin anticoagulant drugs: First reported kindred. N. Engl. J. Med. 271, 809-815, i964.
2.
BRECKENRIDGE, A. Oral anticoagulant drugs: Seminars in Hematol. 15, 19-26, 1978.
3.
TRIGGLE, D.J. Desensitization. In: Towards Understanding Receptors. J.W. Lamble (Ed.) United Kingdom: Elsevier/North-Holland Biomedical Press, 1981, pp.28-34.
4.
SIMON, E.J. pp.159-165.
5.
SHAW, T.R.D., HOWARD, M.R., and HAMER, J. Variation in the biological availability of digoxin. Lancet (Aug. 12) 303-307, 1972.
6.
KALOWSKI, S., RADFORD, N., and KINCAID-SMITH, P. Crystalline and macrocrystalline nitrofurantoin in the treatment of urinary tract infection. N. Engl. J. Med. 290, 385-387, 1974.
7.
BRODIE, B.B. Physico-chemical factors in drug Absorption and Distribution of Drugs. T.B. Binns London: E. & S. Livingstone Ltd. 1964, pp.16-48.
6.
DAWES, C.P., KENDALL, M.J., and WELLING, P.G. Bioavailability of conventional and slow-release oxprenolol in fasted and nonfasted individuals. Br. J. Clin. Pharmac. 7, 299-302, 1979.
9.
WELLING, P.G. Influence of food and diet on gastrointestinal drug absorption: a review. J. Pharmacokinet. Biopharm. 5, 291-334, 1977.
10.
LINDENBAUM, J., RUND, D.G., BUTLER, V.P., TSE-ENG, D., and SAHA, J.R. Inactivation of digoxin by the gut flora: reversal by antibiotic therapy. N. Engl. J. Med. 305, 789-794, 198-i.
11.
TUTTLE, C.B. Intramuscular injections and bioavailability. Hosp. Pharm. 34, 965-968, 1977.
Opiate
receptors:
some
pharmacokinetic aspec:ts.
recent
developments.
Ibid
absorption. In: (Ed.) Edinburgh &
Am.
J.
12
VARIABILITY IN HUMAN DRUG RESPONSE
Suppl. IV, 1983
12.
COCKSHOTT, W.P., THOMPSON, G.T., HOWLETT, L.J., and SEELEY, E.T. Intramuscular or intralipomatous injections. N. Engl. J. Med. 307, 356-358. 1982.
13.
SILVERMAN, W.A., ANDERSEN, D.H., BLANC, W.A., and CROZIER, D.N. A difference in mortality rate and incidence of kernicterus among premature infants allotted to two prophylactic antibacterial regimens. Pediatrics 18, 614-624, 1956.
14.
CHRISTENSEN, L.K., HANSEN, J.M., and KRISTENSEN, M. Sulphaphenzoleinduced hypoglycaemic attacks in tolbutamide-treated diabetics. Lancet II, 1298-1301, 1963.
15.
KOCH-WESER, J. SELLERS, E-M., and Potentiation of warfarin-induced hypoprothrombinemia by chloral hydrate. N. Engl. J. Med. 283, 827-831, 1970.
16.
BRODIE, V.B., and HOGBEN, C.A.M. Some physico-chemical factors in drug action. J. Pharm. Pharmacol. 9, 345-380, 1957.
17.
GRAUSZ, H., and SCHMID, R. Reciprocal relation between plasma albumin level and hepatic sulfobromophthalein removal. N. Engl. J. Med. 284, 1403-1406, 1971.
18.
WEISIGER, R., GOLLAN, J., and OCKNER, R. Receptor for albumin on the liver cell surface may mediate uptake of fatty acids and other albumin-bound substances. Science 211, 1048-1951, 1981.
19.
PIAFSKY, K.M., BORGA, O., ODAR-CEDERLOF, I., JOHANSSON, C., and SJOQVIST, F. Increased plasma protein binding of propranolol and chlorpromazine mediated by disease-induced elevations of plasma aloha-l acid glycoprotein. N. E&l. J. Med. 299, 1435-1439, 1978.-
20.
SUBBARAO, K., RUCINSKI, B., RAUSCH, M.A., SCHMID, K., and NIEWIARQWSKI, Binding of dipyridamole to human platelets and to alpha-l acid S. glycoprotein and its significance for the inhibition of adenosine uptake. J. Clin. Invest. 60, 936-943, 1977.
21.
REIDENBERG, M.M, ODAR-CEDERLOF, I., VON BAHR, C., BORGA, O., and and SJOQVIST, F. diphenylhydantoin Protein binding of desmethylimipraine in plasma from patients with poor renal functon. -N. Engl. J. Med. 285, 265-267, 1971.
22.
HAGER, W.D., FENSTER, P., MAYERSOHN, M., PERRIER, D., GRAVES, P., Digoxin-quinidine interaction. MARCUS, F.I., GOLDMAN, S. and Pharmacokinetic evaluation. N._Eng. J. Med. 300, 1238-1241, 1979.
23.
WILKINSON, G.R., and SHAND, D.G. A physiological approach to hepatic drug clearance. Clin. Pharmac. Ther. 18, 377-390, 1975.
24.
RGWELL, L.B. Human cardiovascular adjustments to exercise and thermal stress. Physiol. Rev. 54, 75-159, 1974.
25.
SWEENEY, G.D. Drugs - some basic concepts. Exercise 13, 247-253, 1981.
Med. & Sci. in Sports &
Suppl. TV, 1983
VARTARILTTY TN HUMAN DRUG RESPONSE
13
26.
Potentially lethal interaction of FINE, A., and CHURCHILL, D.N. cimetidine and morphine. Can. Med. Assoc. J. 124, 1434-1435, 1981.
27.
FENJE, P.C., ISLES, A.F., BALTODANO, A., MACLEOD, S.M., and SOLDIN, S. Interaction of cimetidine and theophylline in two infants. Can. Med. Assoc. J. 126, 1178-1179, 1982.
28.
KERLEY, B., and ALI, M. Cimetidine potentiation of warfarin action. Can. Med. Assoc. J. 126, 116-117, 1982.
29.
FEELY, J., WILKINSON, G.R., and WOOD, A.J.J. Reduction of liver blood flow and propranolol metabolism by cimetidine. N. Engl. J. Med. 304, 692-695, 1980.
30.
JACKSON, J.E. Reduction of liver blood flow by cimetidine. J. Med. 305, 99-100, 1981.
31.
EICHELBAUM, M. Polymorphism of drug oxidation in man: Trends in Pharm. Sci. 2, 31-34, 1981.
32.
CONNEY, A.H. Pharmacological implications induction. Pharmacol. Rev. 19, 317-366, 1967.
33.
RUFFALO, R.L., and THOMPSON, J.R. Effect of cimetidine on the clearance of benzodiazepines. N. Engl. J. Med. 303, 753-754, 1980.
of
N.
Engl.
novel findings.
microsomal
enzyme
34. WEINBERGER, M.M., SMITH, G., MILAVETZ,G., and HENDELES, L. Decreased theophylline clearance due to cimetidine. N. Engl. J. Med. 304, 670-671, 1981. 35. FEELY, J., WILKINSN, G.R., and WOODS, A.J.J. Reductionof liver blood flow and propranololmetabolismby cimetidine. N. Engl. J. Med. 304, 692-695, 1981. C.F., TURNER, W.M., and JONES, J.K. interactions. N. Engl. J. Med. 304, 1301, 1981.
36. GRAHAM,
Lidocaine-propranolol
37. KIMBERLY, R.P., and PLOTZ, P.H. Aspirin-induced depression of renal function. N. Engl. J. Med. 297, 418-424, 1977. 38. GREENBLATT, C.J., and KOCH-WESER, J. Engl. J. Med. 293, 702-705,1975.
Clinical pharmacokinetics. N_
39. BLOCH, R., SWEENEY, G., AHMED, K., DICKINSON, C.J., and INGRAM, D. 'MacDope': a simulation of drug disposition in the human body: applications in clinical pharmacokinetics. 591-602, 1980.
Br. J.
Clin. Pharmac. 10,
40. HARRISON, L.I., and GIBALDI, M.
A physiologically based pharmacokinetic model for digoxin distribution and elimination in the rat. J. Pharm. Sci. 66, 1138-1142, 1977. SYMPOSIUM DISCUSSION
Dr. B. Smith - There is evidence that aspirin and the active metabolite of sulfinpyrazone act on the megakaryocyte and many of these studies have talked
14
about plasma levels of apply to this situation
VARIABILITY IN HUMAN DRUG RESPONSE
these compounds. How would the such as bone marrow?
Suppl. IV, 1983
pharmacokinetic
data
is, “there isn’t an answer” Dr. Sweeney - I think the answer, of course, until the measurements have been made. Pharmacokinetics does not give information about the actual concentration of a drug outside the plasma space, unless you can extrapolate from the concentration of unbound drug in the plasma. Since bone marrow is a well perfused tissue, it is likely that at ‘least the outside of the megakaryocyte is exposed to a comparable level, but even this assumption may be naive. I would like to make an additional comment regarding variation. I believe that kinetic analysis provides disappointingly few answers dealing with the problem of variation. The best available kinetic parameters are often those computed on the basis of studies in quite small groups of normal volunteers. The physiological determinants of drug levels in a population of real patients are likely to show much larger variance. This is not to say that mean values in a patient population will necessarily deviate from those determined in young healthy volunteers (although they may), it is the variation which is most likely to change significantly. In the elderly, there will be systematic changes due to altered body composition and altered function or organ systems, but again, I would predict that it is increasing variation in the population of elderly patients compared with normal volunteers which is the more significant change. Dr. Brick1 - While I agree with your comment, it depends upon how you use the I think if you look not only at the mean figures but also at the data. or the longest and shortest half-lives, you will have at highest and lowest, least some idea of the reliability you might expect from your pharmacokinetic data. And also you get from your pharmacokinetic data the impression whether Thus ( if you most of the metabolism is due to hepatic or renal clearance. have a drug which is cleared by the kidneys and your patients have a renal when know that you have to be very careful dysfunction, then you will considering the pharmacokinetic data.
Dr. Pat kham - This is
a very naive comment, but I have been puzzled by the I am about historical reason for giving all adults the same dose of drug. half the size as other people and yet in all the clinical trials of antithrombotic drugs that I have read about, the same dosage has been given In terms of the This doesn’t make sense to me. to all individuals. phannacokinetics that we have been discussing here this morning, can anyone give me any historical reason for this, or, is it just because it is the simple thing to do? Historically, much of our practice right. Dr. Sweeney - You are absolutely is based upon an attitude toward drugs different from present attitudes and trials without in quantitative precision. - Randomized controlled 1 acking stratification by dose may not be the best way to prove that a drug is individualizing dose is not effective but the qualification is important, always important. is that the chosen dose is selected by Dr. Buchanan - A second possibility what we think is the dose required to achieve the “optimal effect” based upon you need For example, with aspirin, results in ex vivo and in vitro tests. to block cycle-oxygenase in lo-30 uM aspirin, a very low concentration, the relevance of the assay But this raises another question, i.e. vitro.
Suppl.
TV, 1983
VARIABILTTY IN HUMAN DRUG RESPONSE
system used to determine clinical situation.
the optimal
effect
relative
to its
15
relevance
in the
Dr. Fuster - I think we should go to history again in terms of dipyridamole The doses that have been used in the most recent trials and sulfinpyrazone. were chosen because of their effect on platelet survival. Patients who had shortened platelet survival were given such drugs, and with the given dose platelet survival returned to normal. This is the work done mainly by Dr. at the Harker and Dr. Steele in Denver and we have reproduced their results Mayo Clinic. The advantage of platelet survival is, that it is an -in vivo test. However, there is some controversey about how reliable platelet survival is. While at the present time I agree with Dr. Ranhosky we will be interested in knowing what dose is required for each individual by looking at the effectiveness of the drugs, I don’t think this is possible, and I guess we should start from scratch. I would probably start using these drugs first, by body weight, as suggested by Dr. Packham, and perhaps second, I would decrease the dose of all the drugs that we are giving. We probably will hear more later that we are giving too much dipryidamole, too much sulfinpyrazone and too much aspirin and just make these comments in terms of learning about half-lives. Dr. Tut-pie - I think also one has to consider the stimulus for thrombosis because it is well recognized that in prospective trials with aspirin there is a range from 180 mg to prevent shunt thrombosis to 2.3 mg to prevent thrombosis in knee surgery.
reserved.
VARIABILITY IN THE HUMAN DRUGRESPONSE G.D. Sweeney, M.B.,
Ch.B.,
Ph.d
Department of Medicine McMaster University L8N 325 Hamilton, Ontario
ABSTRACT Variations in response to drugs may be pharmacodynamic, implying inter-individual differences in the response of receptors in equal that concentrations of drug, or pharmacokinetic, imp1ying individuals receiving the same dose of drug will have different Either type of concentrations of drug in their body fluids. Variations in receptor variation can be inherited or acquired. sensitivity do occur but few instances, inherited or acquired, have response well clinical If the dose documented relevance. relationship for the drug in question is not steep, or if the therapeutic index is low, drug concentration in the region of the receptor will not be critical and causes of kinetic variation are However, it is the many unlikely to be clinically significant. The se causes of kinetic variation which are best described. include effects due to drug formulation and changes in the If a absorption, distribution, metabolism and excretion of drugs. consideration of dynamics suggests that drug concentration will and prediction of determine therapeutic efficacy, analysis Prediction requires variability due to these factors is desirable. an accurate description of the system but commonly used The pharmacokinetic models may fail when prediction is a goal. variables, volume of distribution (Vd) and rate constant. of elimination (K ) are hybrid in that they arise from the interaction of patient anfi drug characteristics. Important events including binding macromol ecu1 ar and altered blood flow be may not represented. More data is required to determine the clinical analyticai significance of pharmacodynamic variation but better tools are required to deal with kinetic variation when this is important. Specifically, pharmacokinetic models should represent physiological variables and levels of unbound drug in body fluids should receive greater emphasis.
Key Words:
pharmacokinetics, binding
pharmacodynamics, 3
response,
variable,
protein
4
VARIABILITY IN HUMAN DRUGRBSPONSE
Suppl.
TV, 1983
INTRODUCTION The double-blind controlled clinical trial is widely regarded as providing reliable evidence that a drug, prescribed for a patient, will act in a predictable manner, provided that indications and usage match those of the trial. is constrained by the statistical However, such predictability evaluation of the outcome of the trial. Reasons for a variable response must to this problem considered. This paper is be examined and solutions concerned with the sources of variation subsequent to administration of compliance and placebo effects are not considered. drugs: Components of individual variation can be subdivided: pharmacodynamic imply drug receptors which respond in an unusual deviations from ‘normal’ to usual levels of drug while pharmacokinetic fashion despite exposure variation implies that normal receptors are exposed to unusual concentrations Both kinetic and dynamic deviations from normal behaviour may be of drug. genetically determined or acquired and objective indices of drug effect are essential if the problem is to be studied and resolved. In clinical practice, non-invasive indices may be lacking and there is a temptation to substitute measurements of drug level for objective indices of effect. this eliminates the receptor from study and thus precludes However, consideration of pharmacodynamic variations. In interpreting clinical trials there is a tendency to regard deviations abnormality may indeed be defined in from the mean as abnormal. Clinically, this sense (e.g. deviation by more than an arbitrary coefficient of variation fran the mean) but it is more important to regard variation in biology as the norm and to define its determinants in order to prevent it from interfering wih therapeutic goals. A number of inherited abnormalities are Inherited Dynamic Variation. However, there known which affect the disposition and metabolism of drugs. are few documented instances of inherited abnormalities of receptor proteins action at this level. Resistance of rats to the which alter drug rodenticidal action of Warfarin [(3- acetonyl benzyl)-4-hydroxycoumarinl was inherited and in 1964 O’Reilly et al. (1) described a human kindred with an The propositus required a daily dose 50 inherited resistance to this drug. This group has reported a similar standard deviations above the mean. they excluded differences in pharmacokinetic finding in a second kindred: behaviour of Warfarin as the basis for the difference, and provided evidence had a reduced affinity for Vitamin K and a that the hapatic receptor relatively greater reduction in receptor affinity to Warfarin (2). Requirements for detection of receptor mutations as a basis for (a) the means to quantitatively assess variations in a drug response are: (b) data concerning at the receptor; effect of drug the biological These circulating drug levels: and, (c) screening of many patients. limited clinical trials and would be conditions cannot be met during It is, extremely costly during later phases of preparation for marketing. however, likely that abnormalities of the receptors for biogenic amines, for and for other agents will be inhibitors, for opiates cycle-oxygenase described in time. A significant number of examples exist in which adverse effects of acute examples are: documented, are well interactions drug-receptor and malignant hyperpyrexia after exposure primaquine sensitivity, porphyria, These effects are dramatic and hard to overlook. to volatile anesthetics.
Suppl.
TV, 1983
VARTARTLTTYTN HUMAN DRUGRESPONSE
5
That the receptors involved mediate undesired effects does not alter the fact that these are examples of genetically determined variations in drug-receptor responses. Acquired deviations from mean of receptor Acquired Dynamic Variation. response are probably more common although data derives mostly from animal to chemical transmission experimentation. Drugs blocking may lead up-regulation of receptors and to hypersensitivity in a manner consistent Possibly covered by this model is with Canon’s law of denervation (3). reduced CNS sensitivity to sedative drugs in subjects who abuse these agents (including alcohol). Reduction in receptor numbers has been documented in insulin resistance, allergic including myasthenia gravis, di sea se states rhinitis and in asthmatics who have over-used agonist drugs for the bronchial In allergic rhinitis and asthma, the situation is beta-adrenoreceptor. It would be consistent with tachyphylaxis or receptor down-regulation. attractive to postulate a similar mechanism for reduced opiate sensitivity in - a dranatic acquired variation in response - but the molecular addiction basis for this state remains unclear (4). Kinetic Variation. In contrast to variations in receptor response, individual differences in pharmacokinetics are many and well documented: clinically the se, however, are not necessarily significant. If the therapeutic index of a drug is high (e.g. most penicillins) large fluctuations in circulating drug level are of no consequence, The shape of the dose-response curve is also important: if this is relatively flat, so that a wide range of drug concentration is not associated with a wide range of biological effect, drug dosage will not be critical. Further, if a drug effect is sought by blocking a receptor (e.g. cimetidine) and this is without known adverse effects, wide ranges in circulating drug levels will be tolerated. In contrast, if drug effects are sought on the steep portion of the dose response curve, and excessive receptor blockade must be avoided as in the case of beta-adrenoreceptor blockade causing cardiac failure, the situation is different. into the circulation from the site of Drug formulation, absorption distribution throughout the body, metabolism to inactive or administration, active products, and finally elimination, are all factors that determine the concentration of drug which actually reaches receptors. Each may be subject to variation and in each instance inherited as well as acquired processes can be considered, but before discussing these kinetic causes of variability, some commonly-used terms require definition. (Vd) is a statement concerning the fraction of Volume of distribution drug circulating in the plasma and is the ratio of the amount of drug in the body to the plasma drug concentration. Thus:
‘d
=
Total amount of drug in the body Concentration of drug in plasma
While the denominator of this fraction is readily measured, the numerator is only accessible via a mathematical mode. Thus, calculation of Vd depends upon the model chosen to represent drug distribution and elimination. Plasma drug clearance (Cl ) is that equivalent volume of plasma completely cleared of drug in unit e’ime. V and Clp, are critical terms because of their relationship to the rated of drug elimination. The rate constant, for elimination of drug (K,) is related to Vd:
6
VARIABILITY IN HUMAN DRUGRESPONSE
K
e
=
Suppl.
IV, 1983
“pl “d
Half-life (T-1/2) is calculated from K thus: T-l /2 = log 2/K . These are statements of a simplified approach toekinetics but they &cur&ely reflect the dependence of widely used variables upon the validity of simple models. Formulation. Formulation of oral drug is a complex process with the tablet containing many ingredients beside active principle and fillers. Around 1970 the issue of equivalence of different brands of generic drugs achieved considerable prominence following the demonstration that digoxin preparations from different sources differed by as much as 4 times in the amount of drug absorbed by the patient (5). Slower drug release from a larger crystalline form of nitrofurantoin is actually used in Macrodantin brand to decrease gastric irritation (6). has been used to describe The term l*bioavailabilityll Absorption. either that drua absorbed from the gastrointestinal tract or drug reaching the systemic circulation. With increasing appreciation of the role of the elimination” gut mucosa itself in inactivating drugs, the term “presystemic is preferable. Brodie (7) stressed the importance of ,physical factors including oil/water partition coefficient and capacity to ionise as acid or base, in absorption from the gut but of overriding importance is the huge surface area of the small intestine available even for the absorption of drugs ionised at the pH prevailing there. Concomitant food intake can be a potent source of variation in drug absorption. While absorption of oxprenolol, for example, is independent of the destruction of penicillin G in the stomach and the food intake (8), The acceleration of this in the presence of food is well recognized (9). complexity of variation in digoxin availability increased with the recent accelerated individuals demonstrate observation that about 10% of inactivation of the drug to cardioinactive reduced metabolites by gut flora (10). sites is of less interest as the Absorption from parenteral Slow and erratic is becoming less fashionable. intramuscular route lidocaine (11 1 and digoxin, absorption of diazepam, chlordiazepoxide, Injection technique may phenytoin from intramuscular sites is acknowledged. have shown that many contribute to this variability: Cockshott et al. intramuscular injections in the gluteal region are probably given into fat (12). When drug is initially introduced to the body it is Distribution. The circulation then distributes drug absorbed to the vascular compartment. High1 y per fused organs to all organs at a rate determined by blood flow. kidney and myocardium) can thus extract disproportionate amounts of (brain, drug during the distributive phase although poorly perfused tissue such as These events are not simulated fat may accumulate larger quantities later. A second important factor determining drug in simple pharmacokinetic models. distribution is the affinity of drugs for macromolecules including plasma proteins. The role of Binding of Drug to Plasma Protein. The role of binding in is far from settled. distri bpaces ution o drugs an in protein binding It is certainly not clear the extent to which alterations
Suppl.
TV, 1983
VARIABTLTTYTN HLJMAN DRUGRESPONSE
are clinically important as opposed to being merely biologically interesting. There is no doubt of the importance of the initial observation of Silverman the increased mortality in babies of mothers receiving a et al. reporting when two prophylactic antibiotic sulfonamide instead of chlor amphenicol regimes were compared (13) and the probable basis for this was kernicterus Similar due to the failure of albumin to transport bilirubin in the blood. arguments were used to explain hypoglycemic attacks induced by sulphaphenzole (14) and the potentiation of in tolbutamide-treated diabetics hypoprothrombinemia by chlorahydrate was explained by Warfarin-induced displacement of Warfarin from binding sites on albumin by the metabolite, trichloracetic acid (15). However, confusion clouds the effects of binding of drug to plasma protein on clearance of drug by kidneys and liver. Clearance at the glomerular filtraion rate (GFR) is of unbound drug only but both renal tubular and hepatic extraction processes may extract bound, as well as free, Bromsulphthalein (BSP) (which circulates extensively drug. bound to serum albumin) is largely excreted during a single pass through the liver (16). Despite this evidence that protein-bound BSP is cleared by the liver, Schmid showed (17) that artificially increasing the serum albumin decreased the elimination rate constant for BSP and he proposed that transport into the hepatocyte depended upon competition between binding sites within the cell and on albumin. Three mechanisms contribute to complex effects of protein binding on the concentration of free drug and the rate of drug elimination. First, free drug concentration may be increased temporarily by displacement from binding sites by a second drug. Second , decreased protein bindng must increase Vd and may thus decrease K . Third, when protein serves to transport drug to the liver (18), decreased protein binding may decrease Kd. Alterations in protein binding are not necessarily due to other drugs; in renal failure there is altered protein binding although this does not affect all drugs (21). An interesting example of variable binding by tissue, rather than plasma proteins, has been suggested in the recently described and clinically significant interaction between digoxin and quinidine. It was established (22) that digoxin in the presence of therapeutic doses of quinidine increased approximately 2.5-fold in the plasma and this was attributed to decreased renal clearance. A subsequent study failed to demonstrate a change in half-life, but confirmed the increase in plasma concentration. The overall effect was related to a decrease in renal clearance, but with displacement of digoxin by quinidine from tissue binding sites. Consequently, the volume of distribution decreased so that the expression for Ke remained relatively unchanged. Variations in albumin concentration over the range 0.3-0.9 mmoles/l occur in practice but have received little attention as a cause of altered drug kinetics. Such variations will have major effects only if the concentration of drug is of the same order of magnitude as that of albumin. For a drug such as dipyridmnole the therapeutic level of total drug is around ratio will be little affected by changes in plasma 4 uM and the bound/free albumin concentration. While there will be an uncertain effect of protein binding on T1/2 and thus dose schedules, dynamics - the receptor response must reflect free drug and this is measured too infrequently. The elegant studies of Piafsky (19) focused attention on similar problems due to cationic drugs being tightly bcouG;d toTiehe acute. phase reactant protein, alpha-l acid glycoprotein . affinity of dipyridamole for AG has been shown to be high (Kd = 1.6 uM, ref. 20).
7
VARIABILITY IN HUMAN DRUG RESPONSE
Suppl. IV, 1983
Subbarao et al. (20) reported that dipyridamole circulates over 98% bound to plasma protein but that this figure did not correlate well with AG concentration. This data suggests that the drug is extensively bound to albumin as well as to AG. Concentrations of dipyridamole and serum protein will be approximately 2 uM and 1 mM respectively: as noted above, binding under these conditions is not sensitive to protein concentration. However, a different situation may exist in the presence of salicylate at a concentration similar to the protein. Variation in kinetics associated with qualitative and quantitative changes in protein binding sites as well as with effects of drug interactions are large. Some have been shown to be clinicaily significant. Unfortunately, there is too little data to assess whether this is generally true. When a drug is r ported “98% bound to plasma protein”, at a tota -7 $ concentration of 1 ug/ml the concentration of free drug is only 20 ng/mland not long ago, technoloiy generally could not resolve a 20% change in this level. This is no longer true but data is needed. Also required are pharmacokinetic analyses able to simulate and (therefore predict) effects of altered protein binding on kinetics. Wilkinson and Shand (23) added an The Hepatic Extraction Ratio. important dimension to our understanding of individual variation when they developed equations to describe hepatic clearance of drugs in terms of blood An important conclusion of this flow and hepatic extraction ratio (HER). work was that for drugs efficiently extracted by liver from blood, clearance (and thus T1/2) for a constant Vd) would be a function of splanchnic blood flow. Variations of splanchnic blood flow under basal conditions in a healthy population have not been documented, but wide variations occur in individuals as circulatory beds adapt to different circumstances (24). Figure 1 demonstrates how such changes may correlate with changes in Variations at least of comparable magnitude can plasma levels of lidocaine. be anticipated in particular. in circulatory disorders and in shock by drugs and such changes may be Splanchnic blood flow is also altered There has recently been associated with important kinetic interactions. hepatic clearance of drugs caused by interest in reduced considerable with reports of cimetidine causing a clinically cimetidine on this basis, and warfarin theophylline of morphine, significant increase in effect clearance of reduced Feely et al. (29) have demonstrated (26-28). propranolol in the presence of cimetidine with evidence (based upon Clearance blood flow; of indocyanine green) that this is due to reduced hepatic Conventional pharmacokinetic this correlation is disputed (30). however, If the hepatic extraction ratio analyses do not represent these events. and blood flow can be equated, but for many drugs approaches unity, clearance While intuition might predict large changes in drug this does not hold. kinetics in cirrhosis, for example, the data does not bear this out except for drugs with high HER.
Suppl.
TV, 1983
VARTABTLTTYTN HUMAN DRUGRESPONSE
MAINTENANCE
INFUSION LIDOCAINE 2.16 mg/min
I
I
1
2
I
3
9
I
J
4
TIME IHOURS)
FIG. 1 A healthy male subject was infused I.V. with lidocaine for 2 hours then exercised at 70% of maximal oxygen uptake for 1 hour. Increasing circulating levels of drug were consistent with reduction in clearance due to reduced hepatic blood flow (from ref. 25 with permission). of drug metabolism occur between Metabolism. Variations in rates individuals and are probably the most significant determinant of variance in Pathways rate constants of elimination when normal volunteers are studied. of drug metabolism are genetically determined and the inheritance of some clinically significant variations has been reported (31). Rates of drug of metabolism may al so change within individual on account a single environmental factors or with other drugs. Conney (32) interactions initially drew attention to the potential of a large number of drugs to induce the activity of hepatic drug metabolizing enzymes and therefore accelerate the metabolism and inactivation of other agents. Initially it seemed likely that this would be both frequent and a clinicaly important cause of interactions and therapeutic drugs. with environmental agents However, these intra-individual effects on drug metabolizing enzymes are probably less than inter-individual differences. genetically-determined Important examples are accelerated inactivation of prednisone or the oral contraceptive pill in patients taking rifampin, phenobarbital or phenylbutazone. Other barbiturates, potent while inducers of drug metabolism, now have little place in therapy. Drugs which affect splanchnic blood flow may alter the rate of metabolism by the liver, but, as described in the preceding sect ion, only if this is flow-limited (i.e. if HER
10
VARIABILITY IN HUMAN DRUG RESPONSE
Suppl.
IV,
1983
approaches unity). 30th propranolol and cimetidine will reduce splanchnic blood flow and reduced clearance of benzodiazepines (33). theophylline (341, propranolol (351, and lidocaine (36) by one of these agents has been demonstrated. an alternative mechanism in the case of cimetidine is However, that the drug competes for metabolism by microsomal cytochrome in liver. Such competition would be a likely mechanism for drugs where the HER is low, for example, theophylline. Excretion. Altered renal function, which may be permanent or temporary, is a common and well documented cause of variation in drug clearance. Fortunately, the changes that occur can be predicted, or party predicted by measuring BUN, creatinine or a timed creatinine clearance. It is standard practice to predict and adjust dose schedules of e.g. aminoglycosides and digoxin along these lines. This is an example of the clinical value of good, predictive algorithms and of models which adequately simulate the events of interest. In the examples cited, GFR is the significant variable and prediction is accurate because of the close relationship between GFR and hSS is known about variation associated with altered renal tubular “Pl* processes. Controversy continues to surround the effects of aspirin and related non-steroid al, anti-inflammatory agents on renal function and renal It seems clear that aspirin does depress renal function clearance of drugs. with elevation of creatinine (37) but whether this is due to an effect on tubular transport or reduced renal blood flow is not resolved. There are many causes of The Clinical Role of Pharmacokinetics. altered absorption beclause variability in drug response and these include: of blood flow; motility: food; etc., altered distribution because of changes sites: alter ed the quality of quantity of macromolecular binding in metabolism because of changes in hepatic and altered extraction ratio; None of these excretion because of altered function. renal tubular physiological events can be represented in the commonly used mathematical altered pharmacokinetics. Under certain circumstances, an models of clearance rate due to changing physiological function related directly to an alteration in K . The most frequently presented pharmacokinetic model is the two compartmen f if tabulated pharmacokinetic open model (381, however, variables are consulted these tacitly assume that the drug is adequately It is contended that these described using a single compartment model. approaches to pharmacokinetics which may be referred to as “abstract” have That is not to say not proven significantly useful in clinical medicine. when volume of distribution and that for a given drug in a given patient, they have failed to predict the elimination rate constant are known, What the models do not do is provide the steady-state drug concentrations. clinician with a tool with which to anticipate kinetic variation in the The reason is not far to seek: each of the response of patients to drugs. variables used to describe compartmental models is a hybrid resulting from the interactions of the properties of the drug with the properties of the Either of these may change and alter the variable but it is patient. exceptional to be able to take a known physiological variation and use it directly to modify a pharmacokinetic expression to predict how the patient To do this we require models which take into account saturable will react. including simulation of protein binding in drug transport and processes, distribution, and which represent flow through various organs in a meaningful Some progress with such models has been made (39,40); mathematically way. are only possible using fairly large they are complicated and solutions but refining what may be called the “classical” models of computers, pharmacokinetics is unlikely to be helpful.
Suppl. TV, 1983
VARJABTLTTY TN HUMAN DRUG RESPONSE
11
In summary, variability in the human drug response is the rule ratner than the exception. It can be disservice to a drug to anticipate optimal responses from fixed dose therapy. It is unlikely that we have much to learn regarding variations in receptor sensitivity but much has been learned regarding variations in kinetic behaviour of drugs both singly and in combination. Our theoretical models for analysis of these events currently lag behind the empiric data available. Laboratories in particular must pay more attention to protein binding and accurate determination of free drug concentrations before it is reasonable to dismiss circulating drug data as unrelated to clinical response. Pharmacokineticists must similarly pay more attention to macromolecular binding events and there is a great need for mathematic solutions which can accept altered physiological parameters instead of working with hybrid parameters where drug and patient are inextricably mixed. REFERENCES 1.
O'REILLY, R.A., AGGELER, P.M., HOAG, M.S., LEONG, L.S., and KROPATKIN, M.L. Hereditary transmission of exceptional resistance to coumarin anticoagulant drugs: First reported kindred. N. Engl. J. Med. 271, 809-815, i964.
2.
BRECKENRIDGE, A. Oral anticoagulant drugs: Seminars in Hematol. 15, 19-26, 1978.
3.
TRIGGLE, D.J. Desensitization. In: Towards Understanding Receptors. J.W. Lamble (Ed.) United Kingdom: Elsevier/North-Holland Biomedical Press, 1981, pp.28-34.
4.
SIMON, E.J. pp.159-165.
5.
SHAW, T.R.D., HOWARD, M.R., and HAMER, J. Variation in the biological availability of digoxin. Lancet (Aug. 12) 303-307, 1972.
6.
KALOWSKI, S., RADFORD, N., and KINCAID-SMITH, P. Crystalline and macrocrystalline nitrofurantoin in the treatment of urinary tract infection. N. Engl. J. Med. 290, 385-387, 1974.
7.
BRODIE, B.B. Physico-chemical factors in drug Absorption and Distribution of Drugs. T.B. Binns London: E. & S. Livingstone Ltd. 1964, pp.16-48.
6.
DAWES, C.P., KENDALL, M.J., and WELLING, P.G. Bioavailability of conventional and slow-release oxprenolol in fasted and nonfasted individuals. Br. J. Clin. Pharmac. 7, 299-302, 1979.
9.
WELLING, P.G. Influence of food and diet on gastrointestinal drug absorption: a review. J. Pharmacokinet. Biopharm. 5, 291-334, 1977.
10.
LINDENBAUM, J., RUND, D.G., BUTLER, V.P., TSE-ENG, D., and SAHA, J.R. Inactivation of digoxin by the gut flora: reversal by antibiotic therapy. N. Engl. J. Med. 305, 789-794, 198-i.
11.
TUTTLE, C.B. Intramuscular injections and bioavailability. Hosp. Pharm. 34, 965-968, 1977.
Opiate
receptors:
some
pharmacokinetic aspec:ts.
recent
developments.
Ibid
absorption. In: (Ed.) Edinburgh &
Am.
J.
12
VARIABILITY IN HUMAN DRUG RESPONSE
Suppl. IV, 1983
12.
COCKSHOTT, W.P., THOMPSON, G.T., HOWLETT, L.J., and SEELEY, E.T. Intramuscular or intralipomatous injections. N. Engl. J. Med. 307, 356-358. 1982.
13.
SILVERMAN, W.A., ANDERSEN, D.H., BLANC, W.A., and CROZIER, D.N. A difference in mortality rate and incidence of kernicterus among premature infants allotted to two prophylactic antibacterial regimens. Pediatrics 18, 614-624, 1956.
14.
CHRISTENSEN, L.K., HANSEN, J.M., and KRISTENSEN, M. Sulphaphenzoleinduced hypoglycaemic attacks in tolbutamide-treated diabetics. Lancet II, 1298-1301, 1963.
15.
KOCH-WESER, J. SELLERS, E-M., and Potentiation of warfarin-induced hypoprothrombinemia by chloral hydrate. N. Engl. J. Med. 283, 827-831, 1970.
16.
BRODIE, V.B., and HOGBEN, C.A.M. Some physico-chemical factors in drug action. J. Pharm. Pharmacol. 9, 345-380, 1957.
17.
GRAUSZ, H., and SCHMID, R. Reciprocal relation between plasma albumin level and hepatic sulfobromophthalein removal. N. Engl. J. Med. 284, 1403-1406, 1971.
18.
WEISIGER, R., GOLLAN, J., and OCKNER, R. Receptor for albumin on the liver cell surface may mediate uptake of fatty acids and other albumin-bound substances. Science 211, 1048-1951, 1981.
19.
PIAFSKY, K.M., BORGA, O., ODAR-CEDERLOF, I., JOHANSSON, C., and SJOQVIST, F. Increased plasma protein binding of propranolol and chlorpromazine mediated by disease-induced elevations of plasma aloha-l acid glycoprotein. N. E&l. J. Med. 299, 1435-1439, 1978.-
20.
SUBBARAO, K., RUCINSKI, B., RAUSCH, M.A., SCHMID, K., and NIEWIARQWSKI, Binding of dipyridamole to human platelets and to alpha-l acid S. glycoprotein and its significance for the inhibition of adenosine uptake. J. Clin. Invest. 60, 936-943, 1977.
21.
REIDENBERG, M.M, ODAR-CEDERLOF, I., VON BAHR, C., BORGA, O., and and SJOQVIST, F. diphenylhydantoin Protein binding of desmethylimipraine in plasma from patients with poor renal functon. -N. Engl. J. Med. 285, 265-267, 1971.
22.
HAGER, W.D., FENSTER, P., MAYERSOHN, M., PERRIER, D., GRAVES, P., Digoxin-quinidine interaction. MARCUS, F.I., GOLDMAN, S. and Pharmacokinetic evaluation. N._Eng. J. Med. 300, 1238-1241, 1979.
23.
WILKINSON, G.R., and SHAND, D.G. A physiological approach to hepatic drug clearance. Clin. Pharmac. Ther. 18, 377-390, 1975.
24.
RGWELL, L.B. Human cardiovascular adjustments to exercise and thermal stress. Physiol. Rev. 54, 75-159, 1974.
25.
SWEENEY, G.D. Drugs - some basic concepts. Exercise 13, 247-253, 1981.
Med. & Sci. in Sports &
Suppl. TV, 1983
VARTARILTTY TN HUMAN DRUG RESPONSE
13
26.
Potentially lethal interaction of FINE, A., and CHURCHILL, D.N. cimetidine and morphine. Can. Med. Assoc. J. 124, 1434-1435, 1981.
27.
FENJE, P.C., ISLES, A.F., BALTODANO, A., MACLEOD, S.M., and SOLDIN, S. Interaction of cimetidine and theophylline in two infants. Can. Med. Assoc. J. 126, 1178-1179, 1982.
28.
KERLEY, B., and ALI, M. Cimetidine potentiation of warfarin action. Can. Med. Assoc. J. 126, 116-117, 1982.
29.
FEELY, J., WILKINSON, G.R., and WOOD, A.J.J. Reduction of liver blood flow and propranolol metabolism by cimetidine. N. Engl. J. Med. 304, 692-695, 1980.
30.
JACKSON, J.E. Reduction of liver blood flow by cimetidine. J. Med. 305, 99-100, 1981.
31.
EICHELBAUM, M. Polymorphism of drug oxidation in man: Trends in Pharm. Sci. 2, 31-34, 1981.
32.
CONNEY, A.H. Pharmacological implications induction. Pharmacol. Rev. 19, 317-366, 1967.
33.
RUFFALO, R.L., and THOMPSON, J.R. Effect of cimetidine on the clearance of benzodiazepines. N. Engl. J. Med. 303, 753-754, 1980.
of
N.
Engl.
novel findings.
microsomal
enzyme
34. WEINBERGER, M.M., SMITH, G., MILAVETZ,G., and HENDELES, L. Decreased theophylline clearance due to cimetidine. N. Engl. J. Med. 304, 670-671, 1981. 35. FEELY, J., WILKINSN, G.R., and WOODS, A.J.J. Reductionof liver blood flow and propranololmetabolismby cimetidine. N. Engl. J. Med. 304, 692-695, 1981. C.F., TURNER, W.M., and JONES, J.K. interactions. N. Engl. J. Med. 304, 1301, 1981.
36. GRAHAM,
Lidocaine-propranolol
37. KIMBERLY, R.P., and PLOTZ, P.H. Aspirin-induced depression of renal function. N. Engl. J. Med. 297, 418-424, 1977. 38. GREENBLATT, C.J., and KOCH-WESER, J. Engl. J. Med. 293, 702-705,1975.
Clinical pharmacokinetics. N_
39. BLOCH, R., SWEENEY, G., AHMED, K., DICKINSON, C.J., and INGRAM, D. 'MacDope': a simulation of drug disposition in the human body: applications in clinical pharmacokinetics. 591-602, 1980.
Br. J.
Clin. Pharmac. 10,
40. HARRISON, L.I., and GIBALDI, M.
A physiologically based pharmacokinetic model for digoxin distribution and elimination in the rat. J. Pharm. Sci. 66, 1138-1142, 1977. SYMPOSIUM DISCUSSION
Dr. B. Smith - There is evidence that aspirin and the active metabolite of sulfinpyrazone act on the megakaryocyte and many of these studies have talked
14
about plasma levels of apply to this situation
VARIABILITY IN HUMAN DRUG RESPONSE
these compounds. How would the such as bone marrow?
Suppl. IV, 1983
pharmacokinetic
data
is, “there isn’t an answer” Dr. Sweeney - I think the answer, of course, until the measurements have been made. Pharmacokinetics does not give information about the actual concentration of a drug outside the plasma space, unless you can extrapolate from the concentration of unbound drug in the plasma. Since bone marrow is a well perfused tissue, it is likely that at ‘least the outside of the megakaryocyte is exposed to a comparable level, but even this assumption may be naive. I would like to make an additional comment regarding variation. I believe that kinetic analysis provides disappointingly few answers dealing with the problem of variation. The best available kinetic parameters are often those computed on the basis of studies in quite small groups of normal volunteers. The physiological determinants of drug levels in a population of real patients are likely to show much larger variance. This is not to say that mean values in a patient population will necessarily deviate from those determined in young healthy volunteers (although they may), it is the variation which is most likely to change significantly. In the elderly, there will be systematic changes due to altered body composition and altered function or organ systems, but again, I would predict that it is increasing variation in the population of elderly patients compared with normal volunteers which is the more significant change. Dr. Brick1 - While I agree with your comment, it depends upon how you use the I think if you look not only at the mean figures but also at the data. or the longest and shortest half-lives, you will have at highest and lowest, least some idea of the reliability you might expect from your pharmacokinetic data. And also you get from your pharmacokinetic data the impression whether Thus ( if you most of the metabolism is due to hepatic or renal clearance. have a drug which is cleared by the kidneys and your patients have a renal when know that you have to be very careful dysfunction, then you will considering the pharmacokinetic data.
Dr. Pat kham - This is
a very naive comment, but I have been puzzled by the I am about historical reason for giving all adults the same dose of drug. half the size as other people and yet in all the clinical trials of antithrombotic drugs that I have read about, the same dosage has been given In terms of the This doesn’t make sense to me. to all individuals. phannacokinetics that we have been discussing here this morning, can anyone give me any historical reason for this, or, is it just because it is the simple thing to do? Historically, much of our practice right. Dr. Sweeney - You are absolutely is based upon an attitude toward drugs different from present attitudes and trials without in quantitative precision. - Randomized controlled 1 acking stratification by dose may not be the best way to prove that a drug is individualizing dose is not effective but the qualification is important, always important. is that the chosen dose is selected by Dr. Buchanan - A second possibility what we think is the dose required to achieve the “optimal effect” based upon you need For example, with aspirin, results in ex vivo and in vitro tests. to block cycle-oxygenase in lo-30 uM aspirin, a very low concentration, the relevance of the assay But this raises another question, i.e. vitro.
Suppl.
TV, 1983
VARIABILTTY IN HUMAN DRUG RESPONSE
system used to determine clinical situation.
the optimal
effect
relative
to its
15
relevance
in the
Dr. Fuster - I think we should go to history again in terms of dipyridamole The doses that have been used in the most recent trials and sulfinpyrazone. were chosen because of their effect on platelet survival. Patients who had shortened platelet survival were given such drugs, and with the given dose platelet survival returned to normal. This is the work done mainly by Dr. at the Harker and Dr. Steele in Denver and we have reproduced their results Mayo Clinic. The advantage of platelet survival is, that it is an -in vivo test. However, there is some controversey about how reliable platelet survival is. While at the present time I agree with Dr. Ranhosky we will be interested in knowing what dose is required for each individual by looking at the effectiveness of the drugs, I don’t think this is possible, and I guess we should start from scratch. I would probably start using these drugs first, by body weight, as suggested by Dr. Packham, and perhaps second, I would decrease the dose of all the drugs that we are giving. We probably will hear more later that we are giving too much dipryidamole, too much sulfinpyrazone and too much aspirin and just make these comments in terms of learning about half-lives. Dr. Tut-pie - I think also one has to consider the stimulus for thrombosis because it is well recognized that in prospective trials with aspirin there is a range from 180 mg to prevent shunt thrombosis to 2.3 mg to prevent thrombosis in knee surgery.