| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vol. 49, Issue 4, 369-380, December 1997
Department of Pharmacology, University of Toronto, Toronto, Ontario, Canada
I. Introduction
II. The Word "Pharmacogenetics"
III. The Nature of Examples
A. Human Data
1. Pharmacokinetic compared with pharmacodynamic variation.
2. The example of debrisoquine hydroxylase deficiency.
3. Epidemiological implications of kinetic variants.
B. Toxicant Resistance in Arthropods and Bacteria
1. Arthropods.
2. Bacteria.
IV. Comparing Intoxication and Infectious Disease
V. The Biological Cost of Variation
VI. Quantitative Aspects of Pharmacogenetic Variation
A. Monogenic Variation
B. Gaussian Variation
VII. Pharmacogenetics and Evolution
VIII. Summary and Conclusions
Acknowledgments
References
 |
I. Introduction |
|---|
 Top NextReferences
|
|---|
Pharmacogenetics is a field of growing interest in medicine and within the pharmaceutical industry. To physicians whose patients do not respond to drug therapy as expected, pharmacogenetics is increasingly a worry or a nuisance. Numerous drugs that tended to produce variable and sometimes serious responses had to be withdrawn from the market and thus became a loss to the industry. Today, many pharmaceutical companies attempt to minimize such losses by pretesting their products for metabolism, using genetically variable enzymes that may lead to response variations. These practical considerations are valuable but tend to produce a narrow view of pharmacogenetics.
This article provides a brief overview of pharmacogenetics, emphasizing those features that make it a part of population biology rather than merely an item that is from time to time of medical concern. Pharmacogenetic variation is truly important with respect to the adaptability and survival of populations. Pharmacogenetic diversities are a precondition for Darwinian selection.
To present pharmacogenetics from a biological point of view, we start by avoiding confusion with a linguist's look at the word "pharmacogenetics." This is followed by a bird's-eye view of examples of pharmacogenetic variations in humans, insects, and bacteria. Next is a comparison between responses to toxicants and to carriers of infections. We raise the question of the biological cost of pharmacogenetic variation. Section VI. includes a discussion of some quantitative aspects of pharmacogenetics, contrasting monogenic and gaussian variation and recommending a new way to estimate heritabilities. In the final discussion, we consider pharmacogenetic variation in the context of evolution.
| |
II. The Word "Pharmacogenetics" |
|---|
|
TopPreviousNextReferences
|
|---|
The broad applicability of pharmacogenetic principles to all forms of life (e.g., bacteria, insects, mammals, and plants) is sometimes missed for linguistic reasons; to ancient Greeks, the word "pharmakon" meant magic charm, drug, or poison (Webster's). Hence, a meaningful translation of the word could be "xenobiotic," indicating a biologically active material formed outside the host's body (in contrast to an endobiotic, e.g., a hormone). In modern life, however, the meaning of the prefix "pharmaco-" is often equated in a narrow sense with medicine or drug. Hence, geneticists and other scientists sometimes referred to "ecogenetics" when concerned with variable response to environmental chemicals. Other terms used are "toxicogenetics" or "environmental genetics." Agriculturists concerned with insecticide resistance or herbicide resistance, or microbiologists concerned with bacterial resistance to antibiotics, are indeed dealing with a pharmacogenetic phenomenon, but in their communications, they tend to refer to response variations without invoking pharmacogenetics.
Another restriction in customary use of the word "pharmacogenetics"
lies in the fact that, at present, it almost always refers to monogenic
variants. The old twin studies of Vesell and colleagues (1968
, 1992)
gave the best experimental indication of strong genetic components in
drug elimination, although no specific components of variability were
known; Vesell calculated heritabilities. Such heritability data are
clearly part of pharmacogenetics. Thus, in the present context,
pharmacogenetics refers to any kind of inborn variation in any group of
creatures in response to xenobiotics. This definition excludes the
induction of mutations by xenobiotics and the research by modern
industry that uses DNA sequences of receptor proteins to develop new
drugs.
| |
III. The Nature of Examples |
|---|
|
TopPreviousNextReferences
|
|---|
When humans apply chemicals to bacteria, insects, or other "pests" such as rats and mice, it is usually with the purpose of killing them. The usual purpose of applying chemicals to humans is to support life or well-being. There is no reason to assume that such differences in purpose will affect the principles that govern the interactions between biological systems and chemicals in the form of drugs or toxicants. Human data provide the most detailed measurements and records of pharmacogenetic differences between individuals and also between populations, but they are biological snapshots. Toxicological data in bacteria cover thousands of generations, in insects, dozens of generations. Therefore, they are better suited than human studies to reveal principles that relate to survival of a population or of a species, as distinct from survival of the individual.
A. Human Data
1. Pharmacokinetic compared with pharmacodynamic variation.
Numerous physicians and investigators have devoted efforts to the study
of therapy-related clinical pharmacogenetics. Genetic deviations from
the common response to a given chemical may be due to alterations of
the drug target [e.g., receptor systems (Propping and Nothen, 1995;
Vernier et al., 1995)], or abnormalities of the drug's fate in the
body [e.g., drug metabolism (Kalow, 1992b; Pacifici and Fracchia,
1995)]. In most cases so far, genetically controlled exceptional
reactions to commonly used drugs were due to the variation of
drug-metabolizing enzymes (Daly et al., 1993; Kalow, 1993; Price Evans,
1993; May, 1994). It is not clear whether this prominence is incidental
to the development of analytical techniques; the chemical methods to
determine drug metabolism are older than the molecular techniques used
in the study of drug receptor variations. However, it is possible that
receptor variations tend to be rarer because they are often associated
with pathology, a factor that would reduce their frequencies in a
population.
). In humans, approximately 3 dozen of the enzymes that
metabolize foreign compounds have been shown to be genetically
variable. Most variants show deficiency but some show excessive
activity (Kalow, 1992b; Pacifici and Fracchia, 1995).
; Nebert, 1979)
and still under study in humans, is variability in response to enzyme
inducers via the Ah receptor (Okey, 1992). Because the Ah receptor
controls several enzymes, it is interesting to think that a single
variant could control several products.
Ziegler (1991), in his B. B. Brody Award Lecture, emphasized the
general principle that the "enzymes of detoxication" are more
variable than the "enzymes of biosynthesis". This is a shared impression (Kalow and Grant, 1995), but objective counts of variability are not yet available.
Genetically determined variations in drug-metabolizing enzymes are of
obvious importance in the presence of drugs that are substrates for
these enzymes. It is not always evident, however, whether or not such
variation has important biological consequences in the absence of these
drugs (see Section VI.). This is true of some of the best-studied
pharmacogenetic variations, such as the example of
CYP2D6b
(debrisoquine hydroxylase) given below.
2. The example of debrisoquine hydroxylase deficiency.
An
illustrative example from human pharmacogenetics is the deficiency of
the cytochrome P450 CYP2D6 (Balant et al., 1991; Meyer et al., 1992;
Eichelbaum and Gross, 1992). This example is chosen because of the
extent and the intensity of investigations devoted to it. At least six
different mutations lead to a sufficiently faulty gene so that the
enzyme is not formed and thus is absent (Kalow and Grant, 1995). CYP2D6
participates in the metabolism of over 40 drugs. The consequences of
deficiency depend on the metabolized chemical; the deficiency may cause
an exaggerated, even fatal, response because of failure of drug
elimination (e.g., perhexiline, sparteine), or it may cause a lack of
response because of a failure of pro-drug activation (e.g., codeine not
converted to morphine), or the effect may be negligible because there
are alternative processes that compensate for the genetically altered ones. Recessive enzyme deficiency is like the tip of an iceberg (Kalow
and Bertilsson, 1994): The heterozygotes average half of the wild-type
enzyme activity.
). There
are differences not only in enzyme deficiency. Besides variation within
each population, differences of enzyme activity between populations are
evident. The average enzyme activities of both Africans and Chinese are lower than in both European populations, but the enzyme variants are
not identical (Kalow and Bertilsson, 1994; Johansson et al., 1991;
Masimirembwa et al., 1993; Masimirembwa, 1995). The main CYP2D6
difference between Spaniards and Swedes is a more frequent gene
duplication in the former (Dahl et al., 1995), leading to more cases
with very high enzyme activity; this almost certainly represents North
African admixture (Aklillu et al., 1996).
|
, using polymerase chain reaction-single-strand conformation polymorphism analysis, screened 672 subjects of European descent and identified 48 point mutations, 29 of which were novel. Because of doubling or tripling of some point
mutations, there were 53 enzyme variants. The functional significance
of any of the new variants has been tested by phenotyping the carriers
with either debrisoquine, sparteine, or dextromethorphan. Sixteen
genotypes were found to be poor metabolizers, and three cases displayed
genotype-phenotype discrepancies. These results illustrate the power of
the new technique; we have to count on prominent variation of many
enzymes and other proteins, and genotyping and phenotyping of
individual subjects may never completely agree. Today, it is
technically easier to identify a new point mutation than to investigate
its function. Among functional tests, absence of enzyme activity is
much more readily demonstrated than Km variation, which may alter only the enzyme specificity.
Nothing is known about any particular advantage or disadvantage of any
CYP2D6 variant. Because CYP2D6 occurs not only in liver but also in
brain (Tyndale et al., 1991), it might affect some personality traits
(Bertilsson et al., 1989; Llerena et al., 1993), and this might
conceivably affect fitness and, thereby, frequency of the variants.
Most pharmacogenetic differences currently known have not yet been
shown to have any nonpharmacological effects that could affect their
prevalence and their biological importance. To gain some insight into
this question, it is necessary to look at other examples for which the
biological consequences have been more clearly defined.
3. Epidemiological implications of kinetic variants.
The
consequences of drug metabolism as terminator of the action of
medicines is a matter of daily observation by patients and physicians.
The need for drug-metabolizing enzymes with functions that are
unrelated to drug therapy may be less obvious. However, the need is
clear in the case of the exposure to natural toxins, which can be
catastrophic. Exposures of populations to some mycotoxins (fungus-produced toxins; World Health Organization, 1979; Jelinek et
al., 1989; Bhatnagar et al., 1992; Kwon-Chung and Bennett, 1992) have
been incisive medical events of epidemiological proportions in human
history. Examples are the different epidemics of ergot intoxication
(Bettmann, 1956); one was characterized by gangrene of hands and feet
and known as "St. Anthony's Fire" in the Middle Ages in Europe,
the other by high frequency of abortions. It is not known which human
drug-metabolizing enzymes would be able to attack the ergot alkaloids,
but the variability of such enzymes must have affected survival in
these epidemics.
). Drug-metabolizing enzymes play a role in both the
bioactivation and destruction of aflatoxins (Guengerich and Kim, 1990).
Depending on the balance between these two processes, increased
drug-metabolizing activity may actually increase the risk of
aflatoxin-induced cancer.
The mycologist Stormer (1992) cited various periods of high mortality
and low fertility in England since 1541 that could be ascribed to mold
poisoning. He concluded that the sudden increase in the size of the
English population after 1830 cannot be explained by an improvement of
sanitation or medical care. There was a dramatic increase in potato
consumption in the diet of the poor. The increase has been
traditionally ascribed to improved nutrition by the introduction of
potatoes. The modern mycological explanation is that the potatoes did
not provide more calories than did the customary grains, but that
potatoes contain fewer fungi than untreated grains always do. (In
modern times, fungal growth is prevented by predrying the grains before
storage.) The implication is that the toxins produced by fungi present
in the cereal grains in England during the last 400 years sometimes
killed enough people to significantly diminish population size and
thereby, in the long run, impaired population expansion.
We do not know specifically the toxic chemicals nor the
drug-metabolizing enzymes that must have been involved; however, there is no doubt that human biochemical defense systems must have been called into action. If so, it is an example to show that
drug-metabolizing enzymes are not merely serving one drug-exposed
patient, but that they serve a population. One may speculate that some
people had better defense systems than others, but there is nothing to
indicate any development of tolerance to the fungal toxicants in the
population; perhaps the toxic impacts were not strong enough for such
an effect to become visible.
B. Toxicant Resistance in Arthropods and Bacteria
In summarizing the principal aspects of resistance by bacteria to
antibiotics and by arthropods to pesticides, Graham-Bryce (1987) wrote,
"Untreated populations of organisms can be assumed to be so large
that they will contain, as a result of mutation, individuals capable of
withstanding toxicants applied at a rate which will kill the great
majority (i.e., the resistance is preadaptive)".
1. Arthropods.
Arthropod resistance to pesticides poses
serious problems in agriculture (Georghiou and Saito, 1983; Ford et
al., 1988; Roush and Tabashnik, 1990). To retain susceptibility to
pesticides in agricultural practice, the times and kinds of chemical
exposure and measurements of resistance may follow sophisticated
schedules (Forrester et al., 1993). The large variety of types and
kinds of pesticides used against the large variety of living targets is
associated with many different modes of resistance that range from
alterations of toxicant metabolism to changes of toxicant target.
Moderate degrees of resistance may represent enzyme induction, a
reversible process without heritable consequences (Okey, 1992). However, the agriculturally important high resistance levels of whatever cause must be due to multiplication of initially uncommon, naturally insensitive individuals who survive insecticide exposure and
who reproduce in spite of continuous exposure.
). On continued exposure of houseflies to DDT, some
individuals who were able to rapidly metabolize DDT survived and became
the parents of a resistant strain. In many cases, the resistance was
due to the flies' rapid metabolic dechlorination of DDT to form
dichlorodiphenyl-dichloroethylene. In 1971, Brown made the summarizing
statement, "In laboratory strains of houseflies that are not yet pure
for DDT-resistance, the susceptible minus-variants outgrow the
resistant plus-variants because they complete their larval and pupal
developments slightly faster". In other words, if DDT is absent, this
retardation of development in the individuals carrying the gene for
resistance causes the resistance to disappear from a fly population
within a few generations. However, after 30 generations of laboratory
breeding in the continuous presence of DDT, the DDT resistance
persisted in spite of termination of the DDT exposure.
Roush and McKenzie (1987) indicated that some of these traditional data
have to be viewed with reservation. In their review they state that
resistance evolved in the laboratory tends to be multifactorial and is
often seen to be clearly associated with decreased fitness, fitness
being defined in terms of fertility by counting the offspring produced.
After relaxation of the selection pressure, resistance tends to revert
because of the fitness deficit. Resistance developing in field strains
typically involves selection from much larger numbers than in the
laboratory. This tends to produce a higher specificity of resistance
that will often be monogenic. If termination of exposure is followed by
loss of resistance, it is mostly by dilution, i.e., by immigration of
unexposed and therefore nonresistant individuals, a factor that is
difficult to control and quantify under field conditions.
A modern set of field data is shown in figure
2. It illustrates the development of
resistance of an Australian moth (Helicoverpa armigera) to
two different pesticides over a period of 12 years in three different
regions (Forrester, 1996, personal communication). Pyrethroids are
complex organic chemicals derived from a natural plant-produced
insecticide; endosulfan contains a fully chlorinated ring structure.
The data reflect closely controlled, seasonal applications of the
insecticides and precise measurements of resistance by collecting moth
eggs from the sprayed plants, rearing the larvae, and testing their
responsiveness to a lethal dose in the laboratory; the tests are timed
to always target a new generation of moths. The dashed-line portions of
the graphs indicate the times of seasonal interruption of insecticide
spraying; the loss of resistance during this period mostly reflects
immigration of moths from unsprayed areas. This immigration tends to
fade toward the end of the recording, indicating that the abundance of
unaffected moths capable of migrating into the sprayed area is
gradually diminishing.
|
2. Bacteria.
The development of resistance of bacteria to
antibiotics over the past 30 years has been a process of "stunning
effectiveness" (Bennett, 1995). The resistance has reached
proportions that put in question the future utility of antibiotics for
treating infections. Resistance has become a problem that calls for new
pharmaceutical solutions (Service, 1995). For many years already, there
has been bacterial resistance to all clinically used antibiotics (Levy and Novick, 1986), although some resistant strains fortunately are
still isolates (Levin, 1995).
), there is a
fundamental distinction between resistance by chromosomal mutation
(changes in the base sequence of DNA) and resistance carried by
plasmids (extrachromosomal genetic elements that replicate independently of the chromosome; Levin, 1995). Chromosomal mutation may
or may not be associated with a decrease of general fitness; it varies
from case to case. LeClerc et al. (1996) found hypermutation in several
strains of nonlaboratory Escheria coli and salmonellas collected in the wild; hypermutation provides special opportunities for
developing chromosomal antibacterial resistance and for adapting to new
hosts. Bennett (1995) raised the interesting point that most
antibiotics are native or modified products of bacteria or fungi that
must have been naturally resistant to their own products. DNA of some
of these resistant genes was found in preparations of antibiotics. If
this resulted in a DNA transfer to bacterial targets of therapy, it
could enrich a plethora of resistance mechanisms.
Plasmid-determined antibiotic resistance is the most common, although
it comes regularly with a modest biological cost, as reflected by lower
rates of multiplication of the plasmid carriers. A bacterium may carry
several plasmids (Bennett, 1995). Plasmids can be transferred even
between different bacterial species so that clinically important
resistance tends to persist as long as there are some carriers of
appropriate plasmids (Russell and Chopra, 1990).
The current bane of therapy are the R-plasmids conveying multiple
antibiotic resistance. Levin (1995) writes "it seems almost certain
that the ancestors of contemporary R-plasmids and the transposable
elements responsible for the movements of resistance genes to and from
plasmids existed before the human use of antibiotics". It seems that
mechanisms of antibiotic resistance in bacteria are often more complex
and of a greater variety than is pharmacogenetic variation in the more
complex forms of life.
| |
IV. Comparing Intoxication and Infectious Disease |
|---|
|
TopPreviousNextReferences
|
|---|
"[I]t is an advantage to a species to be biochemically
diverse
For the biochemically diverse species will contain at least some members capable of resisting any particular pestilence"
(Haldane, 1949). If we replace the word "pestilence" with
"toxicant," we have a case description of what has been called
pharmacogenetics, that is, the lore of inborn dissimilarities in
response to xenobiotics. Let us compare infectious diseases and
intoxications.
Different life spans of bacterial pathogens and mammalian hosts result
in a complex pathogen-host coevolution (Brunham et al., 1993).
Bacterial pathogens are able to modify their virulence and will change
defensively with time to accommodate different host antigens. Thus, the
timing of the most visible antipathogen defenses of the host is geared
to the life span of the pathogen; the immune system has to adapt
rapidly to the entry of an infectious agent (Langman, 1989; Schultz and
Lerner, 1995).
In addition to the anti-infection defenses controlled by the immune system, there may be genetic variations in the host that have consequences that are relevant to the host's life span. In many aspects, this latter type of variation is comparable to pharmacogenetic variation of the host. Four examples can be offered.
The best known example is the protective effect of genetic deficiency
of glucose-6-phosphate dehydrogenase against malaria (Beutler, 1993;
Luzatto and Mehta, 1995). This particular deficiency happens to be
relevant also to pharmacogenetics; it was discovered in the attempt to
explain ethnicity-related primaquine hemolysis. Approximately 400 genetic variants of that enzyme are known. Also, a particular human
lymphocyte antigen determined at locus A is associated with survival of
infection with malaria (Hill et al., 1991).
Against tuberculosis, there is host protection by a genetic variant
affecting monocytes in different mammals, which is effective before the
immune system comes into play (Vidal et al., 1993). It is not yet clear
to what extent this protection determines different rates of
tuberculosis in different populations.
In mice, the gene for cystic fibrosis has been shown to convey
resistance to cholera (Gabriel et al., 1994); it is possible that the
high frequency of cystic fibrosis in human populations is a consequence
of resistance of the heterozygous carriers of the gene to intestinal
infections that have often been fatal in childhood.
Approximately 1% of Caucasians are immune to infection with the HIV
virus and thereby are protected against acquired immunodeficiency syndrome (AIDS; Huang et al., 1996; Dean et al., 1996). The cause of
immunity is homozygosity for a 32-nucleotide deletion within the
chemokine receptor 5 gene. Infected heterozygotes show a delay in the
development of AIDS. There is also human immunodeficiency virus
resistance in African populations but by an obviously different (yet
unidentified) mechanism.
It seems that one can think in terms of a parallelism between pharmacogenetic variation and genetic host variation against infectious diseases. However, although the immune system may have to cope with bacterial adaptations, this is not necessary when the host is exposed to a xenobiotic that is a clearly defined chemical. Hence, pharmacogenetic variability is principally simple when compared with the events that may follow the entrance of a pathogen into a host's body.
| |
V. The Biological Cost of Variation |
|---|
|
TopPreviousNextReferences
|
|---|
Does pharmacogenetic variation come with a biological cost? Cost in biology is measured in terms of reductions of fitness, whereby fitness is assessed by the number of offspring. It is instructive to raise this question, because many elements of pharmacogenetics have to be considered when looking for an answer.
The answer requires, in the first place, a specification of what is meant by pharmacogenetic variation. The question can apply usefully only to monogenic variation. There it is appopriate to make a distinction between rare and polymorphic variants. The distinction means, in principle, a dealing with two sources of variants. Rare variants may be the result of fresh mutations. Polymorphic variants occur by definition more frequently than could be expected from fresh mutations; their frequency is arbitrarily defined as >1%. In other words, there must be some factor that helps to maintain polymorphic variants in a population.
Balanced polymorphism (Ford, 1940; Cavalli-Sforza and Bodmer, 1971) is
a concept that was introduced to explain the high frequency of some
genetic variants in a population. It means that a variant gene may be
detrimental in homozygous double dose, whereas it may increase fitness
in heterozygotes. The gene frequency in a population is determined by
the balance between heterozygote advantage and homozygote disadvantage.
The classical example is sickle cell hemoglobin; the homozygotes tend
to die young from sickle cell disease, and the heterozygotes survive
malaria better than do nonaffected persons. The sickle cell gene stays
in a population only in areas with malaria.
The general principle displayed by this example means that the benefit or disadvantage of a given allelic gene does not represent an unequivocal or single value. In heterozygous form, the variant may be beneficial while disadvantageous in homozygous form (an independent experience resulting from this situation is hybrid vigor). When considering this principle, one might also think of the Hardy-Weinberg law that the proportion of heterozygotes in a population will always be larger than that of homozygotes. This means that, in any population, the number of alleles in heterozygous form will always be larger than those in homozygous form (except when the gene frequency is 50%).
Pharmacogenetic variants are preadaptive in the sense that they occur before there has been any exposure to the drug. It is a different matter that each variant is noticed only after exposure to a xenobiotic; most exposures are unpredictable events. Because there are innumerable xenobiotics, pharmacogenetic variability must represent a multiplicity of variants. The majority of these variants must be present in heterozygous form and thus may tend to increase rather than decrease the fitness of a population.
Kimura (1968) introduced the now important concept of neutral mutations
to indicate that not all polymorphic variants must represent
"balanced polymorphism". If a new mutation is disadvantageous, it
upsets an existing equilibrium and tends to be eliminated by the forces
of selection. If a new mutation survives in an organism, it is likely
more or less neutral, perhaps causing slightly decreased fitness
(Cooper et al., 1995). The multiplicity of variants that represent
pharmacogenetic variability in a population must be expected to be near
neutral, besides the fact that they mostly occur in heterozygous form.
Their presence should not mean much fitness reduction of the
population.
In conclusion, pharmacogenetic variability will not be expected to mean any substantial biological cost to a population. This fits with the observations that most variants have no visible effect if the xenobiotic is absent and that many variants are polymorphic. Furthermore, any variant if rare and if somewhat disadvantageous would not strongly affect a population. However, this does not mean that pharmacogenetics cannot be costly under some circumstances.
Let me consider in this context a case of a clear-cut association
between monogenic toxicant resistance and reduced fitness. After
exposure to organophosphates, reduced fitness of insects has been
associated with protective overactivity of carboxylesterases, which is
often due to gene amplification (Roush and Daly, 1990). In clones of
some aphids highly resistant to organophosphates, as much as 3% of
body protein was devoted to esterases; the homozygous resistant
genotypes appeared to be only half as fit as susceptible genotypes,
producing only half the normal numbers of offspring. This is one
clearly documented case in which maintenance of a resistance gene is
biologically expensive for the population. However, in this case, the
gene has become frequent in a population because it did provide
protection against a toxicant. Therefore, the variant will be quite
present in homozygous form and in duplicated form, in which it may be
biologically costly.
A kind of fitness disadvantage that has been seen in insects represents
"pleiotropy," that is, multiple phenotypic effects of a gene that
is responsible for a toxicant resistance (Roush and Daly, 1990). In
principle, pleiotropic effects can occur in any living subject,
including humans. However, any disadvantage would depend on the
specific effects of the gene. Sometimes, the only indication of a
disadvantage in insects has been the loss of resistance after
termination of exposure.
Childs (1995), writing about DNA, sums up the situation; it is likely
that some of the mutants that are functionally neutral or nearly
neutral when formed may assume a defensive function when called upon.
What is neutral in one setting may be adaptive or maladaptive in
another. It may constitute a reserve of variation against unforeseen
need.
| |
VI. Quantitative Aspects of Pharmacogenetic Variation |
|---|
|
TopPreviousNextReferences
|
|---|
There are two kinds of variation in drug response: gaussian variation and all-or-none (monogenic) variation. Concurrent contributions of multiple genetic and environmental factors lead to random (gaussian) variation. Monogenic (often polymorphic) variation is based on a controlling effect of a single gene product that may lead to all-or-none responsiveness; pharmacogenetics is often understood to deal only with monogenic variation. There are always transitions or overlaps between these kinds of variation, because random variations are pervasive.
Because of frequent social and economic costs of differences in
response to drugs, toxicants, and carcinogens, pharmacology uses some
otherwise uncommon graphic representations designed to detect
deviations from gaussian variation by monogenic influences (Probit
plots, Normal-Test-Variable plots) (Endrenyi and Patel, 1991).
A. Monogenic Variation
In genetic terms, product deficiency of any autosomal gene will divide the population into three groups: one homozygous for the wild-type gene, one homozygous for the variant, which may mean complete absence of the gene product, and the third group being the heterozygotes. If the variant gene product means the absence of the only drug-metabolizing enzyme, the drug-metabolizing capacity of the heterozygotes will average half that of the wild-type homozygotes. Because there will be gaussian variation in each subgroup, the reduction of the drug-metabolizing capacity in the heterozygotes may be overlooked under clinical conditions.
Under these conditions, the genetically controlled absence of a particular protein may allow a population to be divided into two groups, one group whose members show a particular response to a given chemical and the other group whose members do not. Examples are the "Poor and Extensive" metabolizers of debrisoquine and other drugs, or the "slow and fast acetylators," referring to N-acetyltransferase type 2. If genetic variation of an enzyme does not merely determine its presence or absence but its function (e.g., a Km deviation), the effects tend to vary with the substrate.
B. Gaussian Variation
Gaussian variation is often viewed as environmentally determined,
rather than as part of pharmacogenetics. Most variation in drug
response is of the gaussian variety. Gaussian variation usually
contains hereditary elements that can be defined in terms of
heritability (Vesell, 1992). It is important to note that the edges of
gaussian distribution curves are often of greater clinical interest
than their means (Kalow, 1992a; Kalow and Bertilsson, 1994).
Gaussian variation is a mathematically defined variation,
systematically used for the determination of the median effective or
lethal dose of a drug (ED50 or
LD50; Trevan, 1927). In animal studies with 1000 or more individuals, there is often amazing precision with which the
distribution curves adhere to the gaussian rules; an example among
others is the dose-effect study by Morrell and Chapman (1933) who
tested neoarsphenamine in 1331 rats and whose data were reanalyzed and
reported by Clark (1937). If standard deviations are precisely
measurable parameters, then small differences between them also become
important and susceptible to biological analysis.
Clark (1937) called the size of the standard deviation of a drug
response in different people or animals the "characteristic" of the
drug. He thereby emphasized the behavior of different drugs in the same
population, disemphasizing any differences in standard deviations of
the same drug in different populations. In any case, all such
differences become susceptible to genetic analysis with measurements of
heritability. Questions arising are: What determines the magnitude of
the standard deviations or, in more general terms, the coefficients of
variation (standard deviation as percent of the mean). Is the number of
factors contributing to gaussian variation decisive? To what extent do
differences between these coefficients reflect different mutant
frequencies in the different systems that might control drug effects?
Do we see variation of the drug's target or of the forces determining
its absorption or elimination? There have been very few attempts to
come close to answering such questions.
I was excited as a young scientist (1949) by a paper by Lands et al.
(1948) that related the slopes of logarithmic dose-effect curves of 12 catecholamines to their structures and their patterns of activity; the
drugs were given to mice by intraperitoneal injection, and results were
reported in terms of LD10,
LD50, and LD90. Compounds with the same N- or C-substitutions had the same slopes (i.e., the same
standard deviations) and identical or similar patterns of activity; for
instance, adrenaline and epinine had the same slope in spite of a
100-fold difference in LD50. In this case, it
must be variation of the drug target that determined variability of the
drug response.
There is a recent observation that reveals a rule that affects the
magnitude of variation. Hellriegel et al. (1996) conducted a
meta-analysis of 143 suitable publications and concluded that "the
lower a drug's bioavailability, the greater the intersubject variability in bioavailability". This is perhaps not surprising; low
bioavailability could indicate, for example, excessive first-pass metabolism in gut or liver, or active counter-transport by
p-glycoprotein (Lown et al., 1994; Wacher et al., 1996). Nothing is
known about the genetics of transport proteins.
Perhaps the rule of Hellriegel et al. (1996) is reflected by the
following observation. Lindahl et al. (1996) compared the absorption of
fluvastatin, antipyrine, metoprolol, and atenolol in nine subjects by
testing plasma level kinetics after jejunal perfusion. The coefficients
of variation were 35.1, 41.5, 47.3, and 78.1, respectively. Atenolol
shows the largest variation and has a far lower octanol/water partition
coefficient than the other drugs. Perhaps, the absorption of atenolol
involves variability of active transport, whereas the other drugs
depend mostly on diffusion.
It is clear that gaussian variation may express variability of a drug
target, drug metabolism, or drug transport. When considering variability, we usually do not know which of these factors are involved. Nor do we usually know how much of a given variation is due
to heritability (Vesell, 1992) or environmental factors. However, this
last question should be answerable more often than is currently the
case.
In all the examples discussed so far, the kind of variation almost
automatically considered is between-subject variation. Within-subject
variation is formally considered only when conducting bioequivalence
studies (Midha and Blume, 1993). However, a comparison of between- and
within-subject variation can usually substitute to some extent
for a twin study in sorting environmental and genetic factors. Twin
studies are a necessity if one wants to determine the heritability of
durable characteristics like body size or blood pressure. However, for
studies of the heritability of drug effects, it seems that repeated
applications of the same drug to the same person can often substitute
for the use of twins.
The ratio of between- and within-subject variances of different
caffeine metabolite ratios could clearly show the genetic control of
N-acetyltransferase type 2 and the environmental control of xanthine
oxidase; the activity of the P450 cytochrome CYP1A2 appeared to be
under mostly environmental, but still mixed, control (Kalow, 1996).
Such comparisons are less expensive than twin studies, and they should
be reliable guides particularly in cases of low heritability.
| |
VII. Pharmacogenetics and Evolution |
|---|
|
TopPreviousNextReferences
|
|---|
Pharmacogenetic diversity provides a good illustration of Darwinian principles. Pharmacogenetic variation can be the saving grace for a population by creating survivors of most kinds of toxic impacts. If a population is exposed to a toxicant that destroys a majority of its members, survivors able to withstand the toxic effects become the fittest individuals under the new circumstances, and their propagation replaces the original population. However, diversity cannot give the kind of evolutionary direction that the Darwinian selection process provides by favoring the survival of the fittest individual. Diversity of a population and Darwinian selection are different milestones; the former makes the latter possible.
Diversity of a population is advantageous for its defense not only
against chemical- or pathogen-produced adversities, but for its defense
against all kinds of environmental dangers. With this realization, we
come very close to thoughts expressed and critically discussed by
Williams (1974). We have the advantage of being guided by the
relatively simple and straightforward experiences from
pharmacogenetics.
Here we remember that in humans, dealing with intoxication is not only a matter of biology and pharmacogenetics but of a brain that is capable of acquiring and using medical knowledge. Mental capacities directing behavior are vital in coping with environments. This means that useful variability in a population may be enhanced by individuals who are beyond procreational age.
| |
VIII. Summary and Conclusions |
|---|
|
TopPreviousNextReferences
|
|---|
What have we learned? Pharmacogenetics, heritable variation in response to xenobiotics, is present in all forms of life. Initially, human data perhaps have created the most excitement, and they provide much biochemical detail. However, if we look at pharmacogenetic variation of insects and bacteria, we see it as a characteristic of populations; individuals with inborn resistance to various toxicants can cause the survival of a population by the process of Darwinian selection. Diversity of a population and Darwinian selection are different milestones serving population survival.
Variation of drug response may represent variation of drug targets, drug metabolism, and probably drug transport. Metabolic variation appears to be the most prominent; at present, it is not clear whether this prominence has historical or biological causes.
It is an interesting exercise to compare pharmacogenetic resistance with intoxication and resistance to infection by invasion of disease-carrying bacteria or other pathogens. The big difference is that pathogens tend to show variabilities that drugs do not have. The immune system is made to deal with the genetic variabilities linked to the short life span of most pathogens. However, there are, besides the immune system, several cases of genetic host resistance associated with the long life span of mammalian hosts. Such genetic host resistances are factors equivalent to pharmacogenetic variation. Current data pertain to resistances against malaria, tuberculosis, cholera, and AIDS.
Most pharmacogenetic variants within a population are preadaptive, that is, they are established before xenobiotic exposure. Hence, one must postulate a multiplicity of variants in a population capable of resisting a multiplicity of drugs. The persistence of this multiplicity suggests that most variants are either present in heterozygous form and are thereby advantageous for their carriers, or they are selectively neutral mutants. It means that the biological cost of pharmacogenetic diversity, measured in terms of reduced fertility, should be low in a population.
The frequencies of variant genes are usually not the same in different populations. Also the nucleotide substitutions in a variable gene often differ between populations. In other words, pharmacogenetic differences between populations are typical events.
Pharmacogenetics is usually thought of as the study of a situation in which a single gene product exerts control over a given drug response so that a failure to respond, or an excessive response, may result. However, one should not forget that random variation is always present, probably reflecting the randomness of mutations plus variation of any environmental factors that might contribute. This underlying randomness of variation will always affect the picture of any all-or-none variation. Future pharmacogenetics must deal with both random and monogenic variation.
| |
Acknowledgments |
|---|
|
TopPreviousNextReferences
|
|---|
I owe special thanks to Dr. Harold Kalant in the Department of Pharmacology for his thoughtful critiques. I am grateful for stimulating discussions and criticism to Professor S. Pfeiffer of the University of Guelph and, in Toronto, to Dr. Richard Collins, Department of Genetics, to Dr. Spencer C.H. Barrett, Evolutionist in the Department of Botany, and in Pharmacology to my colleagues Dr. Laszlo Endrenyi and Dr. Rachel Tyndale. Thanks for special advice and for the permission to use unpublished data are due to Dr. Neil W. Forrester of the New South Wales Agricultural Research Station in Australia.
| |
Footnotes |
|---|
a Address for correspondence: Dr. W. Kalow, Department of Pharmacology, Medical Sciences Building, University of Toronto, Toronto, Ontario M5S 1A8, Canada.
| |
Abbreviations |
|---|
CYP2D6, debrisoquine hydroxylase; DDT, dichlorodiphenyltrichloroethane.
| |
References |
|---|
|
TopPrevious |
|---|
the determination of the characteristic curve for rats.
J. Pharmacol.
48: 391-409, 1933.
)-cocaine and nucleotide sequence identity to human hepatic P450 gene CYP2D6.
Mol. Pharmacol.
40: 63-68, 1991[Abstract].
0031-6997/97/4904-0369$03.00/0
PHARMACOLOGICAL REVIEWS
Copyright © 1997 by The American Society for Pharmacology and Experimental Therapeutics
| ||||||||||||||||||
