Individual approach

One of the benefits of mapping the human genome will be the development of drugs that will maximise the therapeutic effect on the individual while avoiding undesirable side-effects.

Drug discovery has been speeded up significantly in recent years by the power of computers and the introduction of techniques such as combinatorial chemistry and high throughput screening. Lead compounds can now be identified much more quickly by testing libraries of chemicals tailored specifically to have some of the properties required in the final drug molecule.

Recent advances in genetic science are promising another revolution. The unravelling of the human genome has the potential to provide a vast amount of information about disease processes, presenting many new possibilities for drug interventions into disease through the identification of novel targets. The new science of pharmacogenomics looks at how a person's genes affects their body's response to drug treatments, and promises to deliver medicines tailored specifically to particular patients and their genetic make-up.

Many current drugs have significant side-effects in some patients and none in others. Some people suffer severe reactions to penicillins, for example, while others may have no response at all to certain types of antihistamines; and, in some patients, the mechanism for removing a drug from the system may be absent or defective, leading to a dangerous accumulation of the drug within the body.

If the mechanisms by which these side-effects are generated could be tied into the patient's genetic make-up and used to predict responses, then there is great potential for the development of improved drug therapies, saving much time and money as well as making the treatment safer.

As our genes are the codes for the proteins that our bodies produce, and very many drug targets are proteins, so researchers are able to identify the gene or genes responsible for a particular pathway, resulting in huge potential for developing gene-specific drugs.

The current approach to drug therapy essentially relies on developing drugs that have a clinical response in the largest possible number of patients. Statistical analysis is then applied to predict the actual outcome in specific patients.

This ultimately results in an approach which is relatively hit and miss, with doctors having to second-guess the effect of a specific drug in their patient. A tailored approach that matched drug therapy to patient would have many advantages.

Pharmacogenomics should also have an effect on the clinical trials process. If a drug proves fatal in, say, two out of 100 on a trial, then currently it would mean instant failure for that compound — even though many of the other patients on the trial could have benefited greatly from it. If the genetic pathway that causes the adverse reaction could be pinpointed, then it would be safe to continue the research, and even launch the drug, with the proviso that it should not be used in people with a certain genetic make-up.

inherited traits

It has long been known that the genetic make-up of a person has an effect on their ability to recognise and process certain chemicals. In 1932 Snyder reported that some people were genetically unable to taste the chemical phenylthiocarbamide. This taste blindness was shown to be an autosomal recessive Mendelian trait, and inherited.

Earlier studies by Garrod on alcaptonuria, the inability to process the amino acids tyrosine and phenylalanine, had shown that this was also an inherited condition. It transpired that the condition is a result of an absence of the enzyme homogentisic acid oxidase, and Garrod's early hypothesis that enzymes could be involved in the detoxification of chemicals in the body has since been proved right.

Various other examples of drugs that could not be processed in some patients have been reported over the years. The tuberculosis treatment isoniazid led to peripheral neuropathy in many patients, essentially distributed on ethnic and geographic lines. This is now known to be attributable to a slow acetylator phenotype affecting maybe half of all Caucasians. African American soldiers in World War II were much more likely to suffer a reaction to the antimalarial drug primaquine, subsequently attributed to a deficiency in the enzyme glucose-6-phosphate dehydrogenase. And large scale clinical trials on the muscle relaxant suxamethonium led to many cases of prolonged apnoea, found to be the result of a lack of the metabolising enzyme pseudocholinesterase.

gene-based processes

These early examples led to the development of theories based on genetic make-up and its effect on the processing of chemicals within the body; these days many examples are known, and some genetic tests are available. There are four important gene-based processes involved in the body's interaction with drugs, any or all of which could affect the drug's metabolism or action:

Of particular importance in the processing of drugs within the body are the cytochrome P450 enzymes. These pathways are particularly important for lipophilic drugs, whose excretion by the kidneys is minimal. Much work has been carried out on this family of enzymes, and the genes that encode them, and many have been found to play a crucial role in the body's metabolism of drugs.

Patients with a matching pair of the null allele of CYP2C19 have a high sensitivity to a collection of drugs, including propranolol, diazepam, omeprazole and amitriptyline. Around 2–5% of Caucasians, and as many as 23% of Asians are of this phenotype.

The CYP2CP enzyme is involved in processing a wide range of drugs, from tetrahydrocannabinol through piroxicam to the NSAIDs ibuprofen and naproxen, plus the anticoagulant warfarin. As many of these drugs, particularly warfarin, need extremely careful dosing, this enzymatic effect can be very pronounced.

And more than 40 drugs in common clinical use are known to be affected by CYP2D6, so patients homozygous for its null alleles will be poor metabolisers of them, and hence much more likely to suffer from adverse drug reactions; the group includes debrisoquine, nortryptiline and metoprolol, as well as many antipsychotic and cardiovascular drugs.

The majority of pharmaceutical companies now include tests to determine metabolism by important cytochrome P450 enzymes within the discovery process, and many that have side-effects in a significant proportion of the population will be spotted at an early stage. However, this could well mean that a drug that may be of great benefit to the majority of patients might drop out of the development process because of its potential to cause problems in maybe only 5% of patients.

computer power

The recent rapid advances in genomics mean it is now becoming practical to start developing drugs targeted at specific areas of the patient population by identifying the polymorphisms involved. A polymorphism is a variation in a DNA sequence that is found in at least 1% of the population. Most of these are single nucleotide polymorphisms (SNPs), where just one base pair of nucleotides is different.

It is estimated that possibly as many as ten million SNPs exist between a pair of genomes. Only a small proportion of these are likely to have a bearing on the metabolism of drugs, so pinpointing these among three billion or so nucleotides within the human genome is no small task.

Computer power holds the key to the success of pharmacogenomics, along with the development of techniques for pinpointing SNPs within DNA. Without the speed and power of modern computer systems, processing and mining such huge quantities of data would be nigh on impossible. It has become a prime field for small start-up companies with complex proprietary technologies. They can license their techniques or data to the pharmaceutical companies, who gain the advantages of the genomic research without having to invest the huge amounts of time and money that would be necessary to do the work in house.

Millennium Predictive Medicine (MPM), for example, which is a wholly owned subsidiary of Millennium Pharmaceuticals, has a broad ranging agreement with Bristol-Myers Squibb for oncology research. The alliance covers a number of B-MS's marketed anticancer drugs, including Taxol, as well as several that are still undergoing clinical trials, focusing on taxane derivatives and ras farnesyl transferase inhibitors.

MPM is looking to identify the genetic differences between people who respond to these drugs and those who do not, information that would vastly increase the likelihood of a patient receiving an effective chemotherapy regime first time. Understanding the precise genetic nature of different tumours means that the right drug could be chosen to target that malignancy.

A company that specialises in pinpointing SNPs is Incyte Genomics, of Palo Alto in California. It has developed an extensive in silico database — containing more than 100,000 SNPs — by mining its LifeSeq Gold sequence database. It has set up alliances with pharmaceutical companies, including Abbott Laboratories and Eli Lilly, who have access to the company's database and technology.

Incyte's polymorphism detection technique uses a proprietary algorithm called SNooPer to discover common DNA sequence variations. It incorporates a series of filters to differentiate SNPs from a range of different error types, and the company claims it can identify genes based on a single occurrence of a difference in sequence. It says that the algorithm has been shown in laboratory studies to be over 80% accurate.

Incyte is collaborating with Sequenom to give customers access to validated gene-based SNPs with ethnic allele frequencies and associated assays. Its SNP data is processed with Sequenom's MassArray system, which determines the medical relevance of SNPs and genes. It can generate ethnic allele frequencies for these SNPs from its human diversity panel, which is a collection of data from 1,200 people from four different ethnic backgrounds: African-American, Asian, Caucasian and Hispanic. This information improves the predictive skills for the association between SNPs and a patient's response to drug treatment.

pinpointing techniques

Lynx Therapeutics, based in Hayward, California, is another company with expertise in pinpointing SNPs. Its techniques are based on a proprietary cloning procedure, Megaclone, which clones around 100,000 copies of a DNA strand onto the surface of a 5µm bead. This makes the handling of DNA clones much simpler, enabling otherwise cumbersome analyses to be carried out. It uses these beads in its Megatype technology, which it is developing to pinpoint SNPs in a single experiment without the need for individual genotyping. The company says this will be a big advantage over other techniques, which require thousands or even millions of assays to be carried out to pick out individual SNPs. No prior knowledge of either SNP sequences or where they are located on the genome is needed.

Orchid Biosciences, of Princeton, New Jersey, uses an SNP scoring technique. Its SNP identification technology, or SNP-IT system, is a single base primer extension technique, which uses a set of biochemical reactions to isolate the precise location of a suspected SNP, and then is able to determine the precise identity of the SNP using the enzyme DNA polymerase. The company says that, as its technique is a direct method of determining the SNP's identity, it is more accurate and more robust than indirect techniques.

It has a collaboration with the not-for-profit SNP Consortium to confirm and determine the frequency of SNPs in diverse populations. The consortium's members include the Wellcome Trust, many of the big pharmaceutical companies, such as AstraZeneca, Aventis, Bayer, Bristol-Myers Squibb, Hoffman-La Roche, GlaxoSmithKline, Novartis, Pfizer and Pharmacia, plus Motorola, IBM, and a number of academic institutions. Its aim is to place as much SNP data as possible in the public domain, and more than a million mapped SNPs have already been released.

second chance

Orchid's SNP analyses are a step towards the consortium's goal of constructing a publicly available genome-wide SNP map. Orchid is screening 60,000 of the consortium's SNPs, genotyping them on DNA samples from 42 individuals from each of three populations. The project is more proof of the huge numbers involved in genome sciences — the 60,000 SNPs equate to over 7.2 million genotypes to be analysed.

Another potential outcome from pharmacogenomic research is that drugs that have been launched and then withdrawn because of side-effects could be resurrected. This might include some of the high profile drugs that have hit problems recently. Bayer's cholesterol lowering agent cerivastatin (Baycol/Lipobay), for example, was voluntarily withdrawn after being shown to cause the muscle weakness rhabdomyolysis in some patients who had been co-prescribed gemfibrozil.

If at-risk patients could be identified and reliably excluded from treatment, then prescribing cerivastatin, which – side-effects apart – is a very effective lipid lowering agent, could become an option once more. Many other withdrawn drugs could receive a similar second lease of life.

This is also something that will need to be addressed by the regulators — and prescribers. If the cerivastatin prescribing information had been followed to the letter, and the contraindications for co-prescribing with gemfibrozil adhered to, then the problems with cerivastatin would have been unlikely to occur.

new mindset

The potential availability of drugs that are known to cause side-effects in patients with a particular phenotype would make genetic testing for compatibility before prescription essential. A new mindset would have to be developed, by which checking for potential side-effects in each patient became routine.

Ultimately, one can imagine an instant blood type test being taken at the pharmacy counter to determine, say, which otc antihistamine would be the best option for a specific patient. The potential for patients to be given the best treatment, individually tailored to their genetic make-up, is huge. Routine genetic screening for drug effectiveness is still a long way down the line, but in some areas, particularly cancer treatment, pharmacogenomics may well have a direct impact on patients in the foreseeable future.