Chapter 2 - The human genome

John Bell

Introduction

Every 50 years, a breakthrough in basic science feeds through to provide a major change in the practice of clinical medicine. Some of these advances have emanated from basic biological science, such as the revolution in microbiology at the end of the nineteenth century that eventually led to the development of effective antimicrobial therapy and the successful assault on many infectious diseases. On a more modest scale, basic advances in immunology have led to significant changes in our understanding of and therapy for immune-mediated disease. In other cases, developments in the basic science had been in non-biological areas. No one would argue about the impact of the silicon chip and microprocessors on modern western medicine in the second half of the twentieth century. Progressing many basic science developments to the bedside has sometimes proved slow, with major advances in science impacting on clinical practice only after 20-50 years. Few would question that recent developments in genetics will prove to be the most significant scientific advances of the current era and it is inevitable that this will eventually feed through to affect many, if not all, of clinical medicine. As with infectious disease, genetics is fundamentally involved with disease pathogenesis, but in addition it is responsible for many aspects of response to therapy and disease outcomes. As a result, the impact of genetics on medicine may well prove to be greater than any previous scientific advance.

The timing of advances in human genetics coincides with a crisis in clinical practice and health-care systems that have evolved in the last 50 years. Financial, demographic and philosophical factors will all precipitate dramatic changes in medicine in the next decade. The limitations in our current practice of medicine are many and they will require substantial changes in our approach to patients with disease if medicine is to continue to thrive into the next millennium. The solution to many of the problems may be found as the impact of human genetics is felt in the clinical sector.

The limitations of current clinical practice

Although clinical medicine has made tremendous progress over the last 50 years, there are clearly substantial limitations to medicine as it is currently practised. Not only has the overall burden of disease increased due to demographic changes, but the extent of chronic disease and our ability to manage it to reduce disability is clearly limited. Contrasted with the considerable success we have had in managing some acute illnesses, such as the major infectious diseases, little progress has been made in preventing or treating diseases such as diabetes, asthma, athritis, or cancer. The explanations for this lack of success are many but they include the major problems discussed below.

Late application of therapy

Most disorders of adult life are preceded by a prolonged presymptomatic period. Vascular disease is known to begin early in life, long before the disease becomes symptomatic, while symptomatic Type II diabetes may be preceded by many years of abnormal glucose tolerance. Much of current medical practice is applied to resolving the problems associated with symptomatic disease, late in the natural history of most of these disorders. Only in a few circumstances have drugs been applied to treat presymptomatic individuals in an attempt to prevent the end-stage sequelae of these abnormalities. The treatment of hypertension and hyperlipidaemia are two examples of where such early therapy has had a profound impact on the management of vascular disease. These problems are also highlighted by the problems associated with screening programmes for diseases such as breast or cervical cancer. These attempts to identify individuals early in the course of disease have been hampered by the difficulties of screening a large segment of the population without stratification. The ability to treat disease early to alter its natural history, rather than treating end-stage disease, ultimately requires methods for early presymptomatic diagnosis and early targeted and safe therapy.

Diseases are defined by the phenotype not their mechanisms

The failure to define diseases and their mechanisms has profound implications on the ability to anticipate the natural history and adequately treat diseases, utilising information about their biochemical basis rather than simply their phenotype. For example, diseases such as diabetes may result from many distinct mechanisms that raise blood sugar while asthma may occur as the result of multiple different causes of bronchospasm or atopy. Understanding and defining the mechanism of disease will permit a clearer definition of these disorders and will have implications for a more precise prediction of their natural history, and more precise targeting of appropriate therapies. At the present time the tendency is to "lump" disorders together based on phenotype criteria, which greatly limits the ability to adequately manage each patient. In the future it will be necessary to split chronic disease phenotypes into different mechanistically defined categories to facilitate the appropriate application of clinical practice.

Therapy is used in heterogeneous groups of patients, resulting in a wide range of therapeutic responses

At the present time therapeutic interventions are used in chronic disease with the knowledge that patients may respond beneficially to particular drugs, but also may either fail to respond or may show serious toxicity to the agent. It is not possible, at present, to predict which category individual patients will fall into. A range of different factors are likely to contribute to this variation in response to therapy. The first and most important variation will be that patients treated for the same disease may have fundamentally different pathophysiological mechanisms (see above). This will mean they will respond differently to therapy and appropriate therapy may be different among the different groups. A second possibility for variation in response to therapy relates to differences in drug metabolism or the biotransformation of drugs in different individuals.

Finally there are a cascade of proteins involved in drug action, each of which may be variable in a population and which may determine differential response to a particular therapeutic agent. These three factors, referred to as pharmacogenomics, will be clearly defined in the near future, and will provide important information for those managing patients with chronic disease.

The common theme in all of these factors that limit current clinical practice is that we know too little about the variation in population that leads to differential disease susceptibility, disease mechanisms, and drug response, and it is not possible to accurately identify those at risk before symptoms occur. An understanding of the basics of variation will inevitably lead to an ability to predict disease risk and individualise therapy, with very considerable improvements in outcome and cost-benefit ratios. These developments will arise from the application of modern molecular genetics to the problem of human disease. The remainder of this chapter will discuss how modern human molecular genetics is likely to rapidly provide us with the information we need to implement changes and correct the deficiencies outlined above.

Genes, polymorphisms and mutations

There are about 100 000 expressed genes in the human genome, many of which have now been sequenced and are available to those studying disease genetics. These genes are made up of strands of deoxyribonucleic acids (DNA). The bases cytosine, thymine, guanine, and adenine are linked to five carbon sugars that carry phosphate groups and these nucleotides are joined in tandem to form strands of DNA. Within a code generated by nucleotides in a DNA strand, there lies the instructions for the synthesis of proteins from amino acids. The sequence of each triplet of nucleotides in a coding sequence encodes a single amino acid, this code is transmitted through messenger RNA to ribosomes where polypeptide chains are created from amino acids. When folded, these produce the proteins that are responsible for most cellular structures and functions. Variation in most aspects of human biology is therefore associated with changes in triple code of DNA and this arises predominantly by mutation. These mutations can lead to the transition of one nucleotide to another, changing the code and the amino acid, or can lead to the insertion or deletion of base pairs, leading to a break in the frame of the triplet code and hence an abnormal or shortened protein. Some of these modifications are silent, while others have important functional effects on the encoded protein. Where function is changed, selection can lead to increased frequency in the population or to an elimination of detrimental variants. The mutation and selection of DNA variants, because of a conferred advantage in the host, underlies all our current concepts of evolution. The diversity generated by such DNA variance accounts for much of the variation seen in human populations, and accounts for much of the variation in disease susceptibility and response to therapy in the population.

In addition to DNA variance that alters coding sequence, variation in non-coding regions of DNA can also have important effects in mediating changes in physiology or disease pathogenesis. Sequences upstream of coding regions are responsible for the regulation of transcription of proteins and variation in these regions can determine the levels of proteins present in individual cells or tissues. Similarly, variation in non-coding regions can alter the splicing pattern of cells and produce significant variance in protein structure. Again these DNA variations can have profound functional effects and can in some cases be selected because of their beneficial effects in individuals and in populations.

Against a background of mutations occurring throughout the genome, particular DNA variants are selected because of beneficial effects. One of the best examples of such selection of DNA variation occurs in molecules responsible for determining the immune response to infectious pathogens. The highly variable HLA alleles have been extensively selected for their ability to present pathogen peptides to the immune system. The HLA allele HLABw53, for example, protects West African children from severe malaria, and has hence been periodically selected in that population where it has high allele frequency. It is less common in regions where malaria selection has not been present.

Such DNA variants are referred to as polymorphisms. Their frequency is usually controlled by a balance between beneficial effect and detrimental effect of these polymorphisms in populations. Often the detrimental effect of such alleles may be an increased liability of another disease. This is important in our understanding of polymorphism and its role in human disease, as many of the DNA variants that account for disease susceptibility in populations are polymorphisms that have been selected because of some beneficial effect they confer on individuals. The concept of "bad" genes in the context of common human disease is an erroneous one. Most, if not all of the DNA variants that account for disease susceptibility carry with them beneficial effects and hence have been important for survival in another set of environmental conditions.

Genes and disease

Lessons from single genes

It is clear that the understanding of genetic factors responsible for fully penetrant single gene disorders has enhanced our ability to detect and eventually treat these disorders. The majority of these diseases fall within the domain of the clinical genetics community and the early successes in defining the genetic basis of diseases such as muscular dystrophy, Huntingdon's disease, myotonic dystrophy, fragile X syndrome and cystic fibrosis has already led to a revolution in our ability to predict, screen and understand these diseases. The power of human genetics in defining the molecular basis of these diseases could have been relatively easily predicted because of their simple inheritance patterns. The impact of this body of knowledge on preventative medical care is likely to be small, however, largely due to the infrequency of these disorders in the population. Nevertheless, a few highly penetrant single gene disorders do occur at a sufficiently high frequency to have a substantial public health implication. One such example is hereditary haemochromatosis. The gene mutation responsible for this disease has been characterised and involves the mutation of cysteine involved in an intrachain disulphide bond in the α3 domain of an HLA-like molecule on chromosome 6. The cloning and characterisation of this DNA variant and the subsequent characterisation of the molecule mutated in this disorder has provided remarkable insights into the disease. These observations, likely to provide an understanding of the molecular basis of iron loading in such patients, will provide an opportunity to screen for individuals at risk for this disease and provide early intervention in the form of venesection to prevent the long-term sequelae, and finally provide evolutionary insights as to why an allele responsible for iron loading will have been selected to such high frequency in human populations.

Genetic contributions to common disease

Compared to the infrequent single gene disorders, understanding the genetic contribution to common human disease provides a considerably greater opportunity to impact on health care in the future. While the mapping of single gene disorders provided the basic paradigm for tracking genes, the greatest challenge was to come from attempting to identify genes involved in diseases where multiple genetic factors were involved and interact with environmental determinants. The range of diseases which can be tackled using such genetic approaches is extremely broad, accounting for most of the major causes of morbidity and mortality in human populations (Table 2.1). Many common diseases are well recognised to have a strong genetic susceptibility component while other diseases, such as the infectious diseases, have only recently emerged as disorders where susceptibility and response to infection is heavily determined by genetic susceptibility factors.

Table 2.1 Examples of common diseases with strong genetic susceptibility
Cardiovascular	Ischaemic heart disease†
	Peripheral vascular disease*
	Hypertrophic cardiomyopathy†
	Dilated cardiomyopathy
	Long Q-T syndrome†
	Rheumatic vascular disease*
	Pulmonary thromboembolism†
Respiratory	Asthma†
	Chronic obstructive airways disease
Gastroenterology	Inflammatory bowel disease*
	Chronic active liver disease*
	Coeliac disease†
	Haemochromatosis†
Endocrine	Type 2 diabetes†
	Type 1 diabetes†
	Autoimmune thyroid disease†
	Polycystic ovary syndrome*
	Osteoporosis*
Rheumatology	Rheumatoid arthritis†
	Ankylosing spondylitis†
	Reiter's syndrome†
	Osteoarthritis*
	SLE*
Nephrology	Glomerulonephritis*
	Renal stone disease†
	Gout†
	Renal tubular disease†
Oncology	Breast cancer†
	Colon cancer†
	Ovarian cancer
	Prostate cancer*
Neurology	Alzheimer's disease†
	Myasthenia gravis†
	Multiple sclerosis*
	Epilepsy†
	Migraine†
	Motor neuron disease†
Psychiatry	Schizophrenia*
	Manic depressive psychosis*
	Anxiety disorders*
	Alcoholism†
* Locus identified by linkage analysis. † Gene and DNA variant cloned.

Several fundamental differences between this form of genetics and that seen in many single gene disorders have emerged. Most importantly, it is clear that the genetic susceptibility responsible for most common diseases are genetic factors which have been selected in populations for their beneficial effects. So, for example, a host of immune response genes and immunologically important polymorphisms clearly dictate susceptibility to infectious pathogens as well as to autoimmune diseases. These polymorphisms have largely been selected for their beneficial effects against particular pathogens. Similarly, it is likely that many of the genes involved in metabolic disorders such as diabetes, obesity and cardiovascular disease may have been selected because of their abilities to provide individuals with an advantage during times of famine (i.e. "thrifty genotype"). These polymorphoric genetic variants are therefore not truly disease genes, but often have some beneficial pheno- typical characteristic associated with their expression, hence they have been selected in high frequency in the population.

There have already been considerable advances in the characterisation of genes involved in susceptibility to many common human diseases. A pattern has emerged which suggested many of these diseases, such as breast cancer, hypertension and diabetes, represent multiple different disorders, each resulting in the phenotype which is the basis for the clinical diagnosis. This highlights one of the previously highlighted limitations in current clinical practice, that diseases are commonly defined using simple phenotypic criteria with no understanding of mechanism. Genetics is rapidly revealing the difficulties with this phenotypic approach, in that most common diseases appear to have several distinct mechanisms, each with distinct, natural features and each with different optimal therapies.

Almost all of the major common disorders include a subset of patients in whom genetic susceptibility is dominant. This sort of genetic disease is relatively tractable and many such genes have now been localised. They often provide significant examples of how a particular phenotype can be created from a range of different mechanisms, and are easier to understand mechanistically because environment appears to play little role. The breast cancer genes BRCA1/BRCA2 are examples of highly penetrant disease susceptibility loci, as are the MODY loci for Type 2 diabetes, MODY 1, 2, and 3. In general, however, these highly penetrant disease loci contribute little to the overall burden of disease, accounting for between 5% and 10% of disease frequency in the population. Nevertheless, for extremely common diseases, such as colon cancer, the ability to detect individuals at risk by screening APC genes and HNPCC genes would allow for detection of up to 10% of colon cancer in the population. Given the frequency of this disease, this would have a very significant impact on the approach to this disease in the future.

Perhaps the most important outcome of genetic mapping studies in common disease to date has been that they provide a much clearer understanding of the biological events that may lead to distortions in physiology that produce morbidity and mortality in the population. In Type 2 diabetes, for example, there is now clear evidence for abnormalities in the glucokinase glucostat being responsible for some forms of the disease, while other forms of the disease are the result of insulin receptor mutations or mutations in transcription factors which may be expressed in the beta cell or the liver. These have provided whole new insights into the mechanisms of these disorders and, ultimately, this is likely to have an impact on our understanding of the natural history of the various subtypes of disease, their optimum therapy, and potentially the complications associated with each of them.

Molecular genetics and a new taxonomy for common disease

What is emerging from the current studies of common disease genetics is that extensive heterogeneity exists under the umbrella of current diagnostic categories. Such "lumping" has clearly impeded our ability to provide optimum medical care to many individuals. Individuals have been treated as if they have the same disease when in fact, biologically speaking, they have totally different disorders, but simply share some particular phenotypic outcome. In the past it has proved extremely difficult to dissect individual disease subtypes in a precise fashion. Histopathology has been used as a rather limited tool to define pathological subtypes of disease, but the constraints on such morphological assessments are clear. Only in the infectious diseases has a century of microbiology and virology provided us with clear insights into these mechanisms. For example, hepatitis has in the past been a clinical phenotype that encompassed a range of patients suffering from non-obstructive forms of jaundice but who had widely differing natural histories, prognoses, and for whom no clear therapy has been identified. The ability to dissect the mechanisms responsible for hepatitis, particularly the various viruses involved (hepatitis A, B, C, E) has helped different clinical syndromes with differing natural histories. Perhaps most interesting is that therapies which are now believed to have some therapeutic effect in one subtype or another (i.e., alpha interferon in hepatitis C) would have been shown to be ineffective if all forms of hepatitis had been lumped together without a clear understanding of their mechanisms. A similar scenario confronts us in most of the other non-infectious diseases, whereby normal therapies are doomed to fail because of the heterogeneity underlying a particular disease phenotype, and where a clinician's ability to predict outcome or manage a patient reliably is severely impeded by the lack of diagnostic precision.

It is clear that an improved understanding of the various mechanisms of disease will provide huge opportunities for better patient diagnosis and management, and that is likely to arise from the substantial molecular genetic efforts to define disease pathways in many common disorders. This is particularly interesting in the context of the apparent variation in disease frequency and pattern in different ethnic populations. Hypertension, for example, is well known to be particularly severe in black populations and is often ACE inhibitor resistant. Similarly, Type 2 diabetes is found at dramatic frequencies of 15-50% in particular ethnic populations, particularly those in Asia and the South Pacific. These variations are likely to arise from differences in gene frequencies and genetic polymorphisms responsible for these phenotypes in different populations, and may have profound implications on how such individuals should be managed clinically. Overall, an understanding of the biochemical pathways involved in a disease process will allow us to move to a new disease taxonomy, with disease being diagnosed and managed rationally, based on mechanism rather than on haphazard notions based on phenotypic criteria.

Genetics and new opportunities for therapy

Targets

A clear understanding of genetic susceptibility could provide the information necessary to contract to development and targeting of new therapies. Genetics is very likely to provide a range of new targets which will allow a more rational and rapid approach to drug design. Genetic approaches have provided the pharmaceutical industry with a wealth of new targets against which to design drugs. Genetics may provide the industry with the means to identify targets implicated in major biochemical pathways involved in disease pathogenesis and will also identify control points on biochemical pathways that may be particularly amenable to interventions. For example the discovery of leptin, using genetic techniques, has opened up a whole pathway with many targets for the development of drugs for obesity. Whether individual disease genes emerge as useful targets in the future may be less important than the fact that the pharmaceutical industry will be pursuing new therapies based on careful and rational understanding of disease processes, rather than by allowing them compound screening.

Pharmacogenomics

Perhaps the first widespread important clinical application of human molecular genetics will be in identifying individuals capable of responding to particular therapeutic interventions, or identifying those unlikely to respond or likely to suffer toxicity due to a particular agent. The wide diversity of response to individual drugs is well recognised by clinicians and pharmaceutical companies alike. In the course of drug development many therapeutic agents provide promising phase 2 clinical data, only to fail at phase 3 clinical trials. Such results are likely to be due to the fact that the patient populations used for these studies are heterogeneous. The inability to select homogeneous patient populations has meant the drug development process has become both risky and expensive. Similar experiences are common in a clinic. Patients respond with substantial variability to most antihypertension medications, and the inability to predict responsiveness leads to poor blood pressure control or treatment with several agents in many patients. Similar problems exist with the use of most therapies in patient populations, all arising because of genetic variation. In addition, drug toxicity has proved to be a major limiting factor in modern medicine. Some effective agents have toxicity in individuals that prevent their use in a responding population, and who cannot be identified from those suffering toxicity. Similarly, relatively rare but important but sometimes fatal toxic effects are important for drugs that are widely used. The ability to identify individuals at risk of such severe toxicity would greatly enhance the practice of clinical medicine.

There are three general ways in which our understanding of genetics is likely to impact on our use of therapeutics in the future. The first area relates to polymorphisms of enzymes involved in the biotransformation of drugs. These enzymes play a central role in modifying compounds we are exposed to in the environment and have therefore been selected extensively in human populations and many contain substantial degrees of genetic polymorphisms. The effects of such biotransformation enzyme polymorphisms have been recognised for many years and have had a significant impact on a range of drugs evaluated as early as the 1960s when debrisoquine toxicity was shown to be the result of polymorphism in the cytochrome P450 enzyme CYP2D6, and acetylator polymorphisms were shown to account for the variation in handling agents such as isoniazid or hydralazine. Molecular techniques have demonstrated that many such enzymes are polymorphic (Table 2.2). Toxicity from drugs is often related to drug metabolism and its variation in the population and hence the ability to identify the DNA variants responsible for these effects, and to predict metabolism and clinical outcomes, is likely to have a profound effect on the way new drugs are both developed and applied in practice.

A second important mechanism for the more precise and accurate application of therapy will be the definition of disease based on genetic mechanisms rather than phenotypes. Individuals who might suffer from totally different forms of hypertension (i.e., one associated with abnormal salt handling versus another associated with catecholamines) are likely to be differentially responsive to therapies directed at such mechanisms. Understanding, genetically and mechanistically, the factors responsible for the disease susceptibility will undoubtedly lead to substantial further precision in the use of drugs.

Table 2.2 - Allele frequencies for certain polymorphic CYP450 enzymes in Caucasians
Enzyme polymorphism		Frequency
CYP2A6	Consensus	78%
CYP2A6v1		17%
CYP2A6v2		7%
CYP2C9*1	Consensus	79%
CYP2C9*2		12%
CYP2C9*3		9%
CYP2D6	Consensus	65%
CYP2D6A		4%
CYP2D6B		10%
CYP2D6C		4%
Adapted from C.W. Wolf and `Pharmacogenomics' (Scrip, 1998)

A third and very important application of genetics, clinical pharmacology, will be the identification of variation in drug targets and molecules associated with them. It is already recognised that many of the common drug targets such as beta adrenoceptors, 5HT receptors, and angiotensin receptors, and in many cases their associated signalling molecules, show polymorphism in the population. Some of these variants have been demonstrated to have important effects on the response or toxicity associated with drugs that utilise them as targets. It is likely that many such polymorphisms will be found and will explain a range of responses in a population to every individual therapy currently available. If such polymorphisms can be shown to predict response then they are likely to be used as an adjunct in everyday clinical practice for the selection of appropriate therapy for different individuals. This important area has only recently been accessible for study through the availability of large sequence databases and extensive characterisation of simple nucleotype polymorphisms in such genomes. It is likely to dictate both drug development and clinical practice extensively in the future.

Genetics and clinical practice

Individualisation of medical practice

One of the clear limitations in current practice is that patients are grouped together because of a particular phenotypic feature, and are then managed using standard protocols. Recently this has been institutionalised in the form of practice guidelines for individual complaints. Beneficial as these approaches have been in the past, they clearly fail to address the central issue in modern medical practice, and that is that very few patients are likely to be truly identical either in the nature of their disease, the presence of particular disease modifiers that alter the response to particular disease insults, or in their response to therapy. As a result, clinical practice has proved to be crude and only sometimes effective. The huge variation in individuals has made the task of applying medical care efficiently across large cohorts of patients difficult because of the wide variation in the genes that relate to or modify a disease phenotype.

Although large randomised controlled trials have become the gold standard for validation of effective therapeutic interventions, the lack of diagnostic precision at entry to such trials means that patient subpopulations, with different disease mechanisms or pharmacogenetic features, will be obscured by the heterogeneity of the overall patient population. Valid active therapy is undoubtedly overlooked by such protocols but they remain the only robust mechanism for currently evaluating therapy. Their utility would be greatly improved and refined if the populations studied at least suffered from the same disease.

As medical care advances, it will be necessary to increasingly recognise this genetic variation, characterise it, and then utilise it by increasing the individualisaton of therapy. This need not provide excessive costs to any health-care system. Given that genetic variants will be easily and inexpensively detected in the relatively near future (see below), it is likely that the information will be available to take such decisions on an individual basis. The resources that are wasted by applying therapies which have no effect or are toxic in some individuals, and the efforts made attempting to treat disease using one mechanistic paradigm when the patient is in fact suffering from a phenocopy of the disease should provide ample leeway to introduce such individualisation of therapy without substantially adding cost. The current system is ineffective because of lack of precision, and if that precision can be gleaned effectively without substantial added cost then the economic benefits are likely to be real.

An additional commercial benefit will be that the cost of drug development is likely to fall. New therapies will be available because of the targets created by genetic technologies. Drug development programmes are more likely to achieve success because of appropriate patient selection and stratification. The utilisation of such genetic definition is likely to be actively encouraged by health-care providers, who are unlikely to be willing to apply drugs in populations where they may be beneficial in only a minority of patients.

Diagnostics as an adjunct to therapy

One of the important aspects of the field of genetics and clinical medicine is that genetic diagnosis is likely to be used in the future as an extremely important tool in the management of all patients. If therapy is going to be utilised in a rational way and if patients are going to be properly categorised before being provided with appropriate advice and therapy, then rapid and efficient diagnostic methods will need to be available. A revolution is occurring in this particular aspect of medical instrumentation in that it is now possible to detect relatively large numbers of single nucleotide polymorphisms simultaneously, using a range of new technologies.

There is now an increasingly powerful international programme to detect the variation that exists within the human genome. The DNA sequence of an entire human genome is likely to be available within the next 5 years, but more importantly as many as 100 000 single nucleotide polymorphisms are likely to be detected over the same timeframe. Some of these polymorphisms will be in non-coding sequences but may be in strong linkage disequilibrium with polymorphisms in coding sequences that are important functionally. This set of polymorphisms is likely to provide a handle on many of the important genetic determinants of disease susceptibility and response to therapy that exist in the population. At the present time, techniques for typing such polymorphisms are slow and laborious. If genetic diagnostics is likely to be widely applied in clinical practice, it is essential that relatively low cost technology is available for typing large numbers of polymorphisms in parallel. The number of DNA variants that will need to be tested in any individual is likely to be large over their lifetime of health care and as a result a systematic in parallel approach to typing polymorphisms may prove to be the most effective. At present single nucleotide polymorphisms (SNPs) are being typed using conventional slab gel technologies or more recently capilliary electrophoresis arrays. These permit a relatively high throughput of polymorphisms but have nowhere near the capability that will be required to detect thousands or tens of thousands of polymorphisms simultaneously. Two new technologies provide the possibility of this in the future. The first of these are microarrays which use oligonucleotides attached in arrays to glass slides or chips. Hybridization of amplified DNA sequences to these oligonucleotides provide information on single nucleotide variation with sufficient redundancy to provide accurate results. Many tens of thousands and eventually hundreds of thousands of oligonucleotides will be available on a single chip, allowing for many single nucleotide polymorphisms to be typed simultaneously for an individual. This raises the possibility that an individual will have their genotyping performed on cord blood and the data on these polymorphisms stored for all future use by medical practitioners. Given the potential impact of genetics on health care over a lifetime, such an approach is likely to be very cost effective. This technology is progressing rapidly with substantial commercial support and it is likely that it will be available for systematic use within the next 2 years. A similarly powerful technology involves mass spectroscopy which should permit the detection of large numbers of single nucleotide polymorphisms with a high degree of accuracy, again in parallel. This technology has not been as well developed but may prove to be as powerful as microarrays in the long term.

If genetic diagnostics are to be used at all stages of medical activity, i.e. for identifying individuals who would benefit from screening, for refining a diagnosis based on mechanisms, and for identifying response for toxicity to a range of therapies, such systematic approaches to genetic typing will be necessary. If genetics is used to identify populations that are particularly responsive to therapy by the pharmaceutical industry prior to the licencing of such agents, it is likely that such genetic testing will accompany regulatory approval for new drugs, and if this is the case it will force the implementation of genetics very rapidly into the clinic.

Genetic stratification

The use of genetics in either selective populations or widely across the population will in the future be necessary to ensure that health care is being directed at those at most risk of individual diseases and outcomes. At present most population screening programmes are acquired generically across particular age groups in an attempt to identify individuals with early disease. The success of programmes such as the breast cancer screening and cervical cancer screening programmes has been limited, because individuals at a particularly high risk of these disorders are managed in the same way as those with no risks. The lack of patient stratification means that significant numbers of low-risk individuals are exposed to the possibility of false-positive results, while high-risk individuals with disease may be identified. Similar criticisms could be made of virtually all existing screening programmes, which are pursued with considerable expense and effort within most health-care systems.

In addition to providing improved methods for identifying individuals at risk of diseases which are subjected to screening tests, genetics may also provide an opportunity to identify relatively small subsets of very high-risk individuals who may benefit from early treatment. The concept of treatment in high-risk individuals has been pioneered through the management of diseases such as hypertension and hypercholesterolaemia. These interventions have validated the concept that early intervention may have important therapeutic benefits if individuals at sufficiently high risk of disease can be identified. The great power of genetic testing is that it identifies such individuals before they have become symptomatic, thus permiting early therapy. This is a central issue in modern clinical practice, as most of our interventions occur at a late stage in disease, often after disability has begun. The ability to intervene early in a large number of diseases may dramatically alter our abilities to reduce disability in the population, and reduce our dependence on intensive resuscitation in individuals with end-stage disease.

One can imagine already how genetics might be applied to further refine our ability to use presymptomatic therapies effectively. It is recognised that the risk of myocardial infarction falls throughout the entire range of cholesterol. Cholesterol-lowering agents are utilised predominantly in individuals with extremes of the phenotype, as it is in this population that the benefits are most cost effective. Individuals with cholesterol levels in the normal range provide a poor cost-benefit ratio unless individuals can be identified with particularly high risks of developing myocardial infarction in this group. Secondary prevention studies have shown that reduction of cholesterol in individuals who had a myocardial infarction provide some benefit. Similarly, one would predict that primary prevention studies of individuals who have significant risk factors, either environmental (smoking) and/or substantial other genetic risk factors for ischaemic heart disease are likely to benefit considerably from cholesterol-lowering therapy, even if their cholesterols are not elevated. Many such opportunities to intervene in presymptomatic patients are likely to arise, particularly if the pharmaceutical industry is confident that risk factor prediction using genetics will allow them to intervene effectively in early stage disease. Diseases such as Alzheimer's disease will be optimal targets for this sort of approach.

The ethics and practice of genetic testing

Predictive testing versus risk factor detection

Predictive testing

There is considerable confusion about the relative role of genetic testing in highly penetrant genetic disorders compared to the identification of genetic factors that confer increased (or reduced) risk of particular diseases. The conventional paradigm established by clinical geneticists is around highly penetrant single gene disorders where the identification of a particular DNA variant is a highly reliable predictor of outcome. Rules for testing have developed around this model, with particular concern about ethical and consent issues. Counselling has played an important role in the management of patients undergoing this form of predictive testing which can relatively accurately provide information about individual risk. This area of genetic testing is likely to undergo substantial changes in the near future. The reason for these changes is that the magnitude of the problem has grown considerably. In particular, with the identification of highly penetrant genes in diseases such as breast cancer and colon cancer, there will be an urgent requirement for the development of testing services for high-risk populations, and perhaps eventually for entire population cohorts, for the detection of these DNA variants. This is particularly true with colon cancer where prophylactic measures are feasible, and where the burden of disease can be considerably reduced using such approaches. Because of the dramatic forms of interventional therapy available to those with breast cancer genes, this is unlikely, in the UK, to be utilised widely. The situation may change, however, should agents such as tamoxifen be shown to be therapeutic with benefits in individuals with this predisposition. Similarly, other genetic variants, such as those responsible for haemochromatosis, can be relatively easily screened at present and will provide an opportunity to significantly reduce the morbidity and mortality from this disease, using simple forms of early intervention such as venesection. It is unlikely that all individuals undergoing predictive testing in this substantially expanded arena are likely to enjoy the benefits of genetic counselling. Screening, even for these relatively highly penetrant single gene disorders, is likely to occur outside of clinical genetics units, most probably in the hands of subspecialists interested in the disease itself. The counselling available for individuals entering such genetic testing programmes is likely to be no more than those entering existing non-genetic screening programmes.

Risk factor detection

There are fundamental differences between predictive testing using DNA variants to detect genetic abnormalities that are highly penetrant and can be used to predict, with high degrees of certainty, the risk of disease in an individual, and those DNA variants which provide evidence of risk in populations but are considerably less useful in individual patients. Most of the genes responsible for common disease will be the latter, that is they will account for some of the risk of developing an individual disease but, because of the complex interplay between multiple genetic determinants in the environment, cannot in any way be used as predictive factors in an individual patient. There has been much confusion about this in clinical circles as the concept of genetic testing has been mistakenly transferred from highly penetrant single gene disorders to low penetrance risk factors defined by DNA testing. This has extremely important implications for the management of such testing programmes and, in particular, the ethical considerations involved in such genetic analyses. There is already extensive experience in evaluating risk factors of a variety of kinds in populations. Hypertension or high cholesterol are both examples of important risk factors. Indeed, both these risk factors are probably accounted for substantially by a range of genetic factors and the phenotype that is measured is a culmination of those genetics factors and some environmental determinants. The screening of individuals for such risk factors is now well evaluated and the ethical issues associated with such presymptomatic testing have been widely and carefully considered. There is little or no difference between this form of risk factor detection and that associated with the identification of risk factors at the level of DNA.

The importance of this distinction is that it should be feasible to do large scale analysis of risk factors to provide information and risk stratification across a wide number of diseases as the disease susceptibility genes are increasingly cloned and characterised.

There are already some important examples of susceptibility genes which may in the future be used to predict risk and possibly also aid in early therapy. The best single example is the ApoE 4 allele, an allele at the ApoE locus with a frequency of 15%. This allele has been shown to contribute substantially to the risk of developing Alzheimer's disease in many different studies. Individuals with one ApoE 4 allele are at increased risk of developing Alzheimer's disease. The risk is even greater in those who are ApoE 4 homozygous. Interestingly, new therapies for Alzheimer's disease have been shown to be differentially effective in ApoE 4 positive and negative populations with this clinical syndrome. It is possible therefore that this genetic information may be used to identify people early in the course of disease and possibly also predict those likely to respond to therapy or even particular environmental alterations, such as change in diet.

Applications of genetic testing

As a host of disease susceptibility genes are characterised, testing for these is likely to be commonplace in all forms of clinical medicine. Two different groups of individuals are likely to benefit from such testing. These include presymptomatic patient populations, and those who have been diagnosed as having a particular condition and who need to be most appropriately managed.

Presymptomatic testing

The use of presymptomatic testing will be driven by the availability of therapeutic agents that can be used early in the course of disease for the identification of environmental factors which have a significant effect on disease initiation or progression. Many of these opportunities are likely to arise as genetics permits a more accurate disease definition and, hence, a more precise mechanism for defining disease subtypes and their relevant environmental factors. Early exposure to allergens may prove to be an important factor in susceptibility in asthma and, hence, genetic prediction for those at risk of particular forms of this disease would be of some benefit, even if only used to reduce the allergen exposure. Early interventions already exist for many forms of vascular disease and the availability of genetic susceptibility factors that could be used to identify individuals at high risk can be used in conjunction with the identification of other environmental risk factors, such as smoking, to identify subsets of the population who would benefit from early forms of therapy. In Type 1 diabetes it is clear that the autoimmune process that eventually damages and destroys the pancreatic beta cells begins many years before patients present with the disease. This provides an opportunity to intervene with immunomodulatory therapy at an early stage in this disorder if individuals at high risk of the disease could be identified. It is likely that genetics will play an important part in identifying the subset of the population most likely to benefit from such screening programmes.

Testing in symptomatic patients

The other major application of genetic testing will be in individuals who present with symptoms. In this event, genetics is likely to be used for a wide number of purposes in the diagnosis and management of individuals. First, genetic testing is likely to help establish more accurate diagnoses in such patients by providing insight into particular mechanistic disease subtypes. This will have an important implication for predicting the natural history of the disease, as will the presence of particular disease modifier polymorphisms, or genes responsible for conferring risk of complications in diseases such as diabetes. In this population of patients, decisions about management of patients is likely to be closely allied to genetic information with all aspects of pharmacogenomic technology being applicable as appropriate therapies are chosen. Information about disease mechanisms may also provide opportunities for environmental variation, which may also prove to be effective therapy for the disease. On balance, therefore, it is likely that genetic testing will provide essential information on almost all parts of the diagnostic and therapeutic process in the future. It is likely to be sufficiently widespread that systematic testing of individuals may provide sets of genetic data which reside as a database for practising clinicians.
Table 2.3

Table 2.3 - Pharmacogenetic applications - examples

Cancer

Thiopurine S-methyltransferase deficiency

Dihydropyrimidine dehydrogenase deficiency

Cyclophosphamide metabolism

Ataxia telangiectasia

Neurological and psychiatric

Apo E and Alzheimer's disease

Anaesthesia

Succinylcholine sensitivity

Malignant hyperthermia

Cytochrome P450 effects and psychotropic drugs

Clozapine and 5-HT receptors

Migraine

Drug addiction

Cardiovascular

Debrisoquine

N-acetylation polymorphism: procainamide and hydralazine

Cholesterol ester transfer protein

Long Q-T syndrome

Anticoagulation (Factor V, Warfarin)

β adrenoreceptors

Infectious diseases

Identification of pathogens

Drug resistance

Limitations and time lines

Much of what has been discussed in this chapter can be predicted by information which we currently have available. The overall implications of genetics to all aspects of medicine are now inevitable. What is much less certain is over what time these changes are likely to be introduced and what implications this has for the training of health-care professionals and the education of the public at large. The pace at which human genetics is moving is continually accelerating, driven largely by new technologies and large commercial resources. This is likely to mean a rapid development of maps of the genome containing large numbers of single nucleotides and a fairly rapid progression, through linkage and association studies, to the detection of polymorphisms that contribute to disease susceptibility and therapeutic response. The use of genetics for the development of novel therapeutic agents, and the combined use of diagnostics and therapeutics together, is likely to rapidly drive the requirement for genetic testing into the health-care system. This may be further accelerated by demands by drug regulators that new therapeutic agents are used only in particular subtypes of the population that are likely to receive benefit as defined by genetic factors.

Implementation of genetics and health care will, however, require several other factors. Much of the genetic information will need to be tested in large patient cohorts to establish the real contribution of disease susceptibility variants to risk, and to determine the real effects of genes and their interactions with other genes and environment in the population. Additional complexity will come from the study of varying ethnic groups, particularly with the DNA variants being studied are likely to be in linkage disequilibrium with functional variants. As with all new innovations, implementation should rely on solid evidence of benefit.

The following colleagues were invited to act as commentators on early drafts:
Dr Eric Meslin Executive Director, National Bioethics Advisory Commission, Rockville, Maryland, USA
Professor William Cookson Wellcome Trust Senior Clinical research Fellow, Nuffield Department of Clinical Medicine, John Radcliffe Hospital, Oxford
Dr Jonathan Flint Wellcome Trust Senior Clinical Fellow & Honorary Consultant Psychologist, Institute of Molecular Medicine, Headington, Oxford
Professor Diane Cox Professor of Human Genetics, University of Alberta, Hospital for Sick Children, Toronto, Ontario, Canada
Dr Mark Edwards Scientific Director, Oxagen, Abingdon, Oxfordshire, UK

[Return to Clinical Futures index]