Getting Personal

Berwyn Clarke of Lab21 Limited examines the role of the biotech industry in the introduction of genetic data into routine personalised medicine

The process by which new pharmaceutical agents are being discovered and developed is entering a new age. One of the reasons for this is that medicine now has the tools to deliver drugs to individual patients that are tailored specifically to that patient, thus optimising their treatment so that they realise maximum clinical benefit from the drug, and making it no longer acceptable for them to be given a generic medicine. As this new ‘personalised medicine’ becomes widely adopted, the pharmaceutical and diagnostics industries now need to work closely together so that the correct diagnostic tests are available to support the new focused use of these drugs as they are licensed. Although there are many factors that contribute significantly to the ways that individual patients respond to particular drugs, it is no surprise that the genetic background of the patient plays a major role. This article reviews some of the genetic parameters that are already being introduced into personalised medicine, and outlines some of the ways in which both the pharmaceutical and biotech sectors are integrating into this process.

PHARMACOGENETICS AND COMPANION DIAGNOSTICS

There are currently two general areas in which genetic information is being used in personalised medicine. The first of these is the use of pharmacogenetics, and the second is the use of specific new diagnostic assays that are required before a specific drug can be used. These so-called companion diagnostics or theranostics can analyse specific genetic characteristics of a particular pathology so that it can be determined whether the disease will respond to a particular drug. This is a rapidly growing area, particularly in the treatment of a variety of cancers.

Pharmacogenetics is the study of the ways that individual patients respond to different drugs and it is not a new phenomenon (1). In the 19th century, Sir William Osler noted that “no two individuals react alike and behave alike under the abnormal conditions which we know as disease”. Different people often vary in the way they respond to the same medicine; in some people, a drug may have no therapeutic benefit, while in others the drug may actually cause harmful side effects. These parameters of efficacy and toxicity are two of the major drivers for the evolution of personalised medicine. Table 1 indicates that many of the drugs currently used in routine clinical practice may not be being used in the way they should.

Given this information, it is not surprising that the pharmaceutical regulatory authorities in both North America and Europe are now requesting that drug companies integrate genetic information into their clinical trial designs. This requirement has implications across the pharmaceutical industry. For the clinician, it means that when a new drug is licensed there will be available information to determine specifically which patients are likely to respond to the drug and, potentially, those patients who are unlikely to be able to metabolise the drug and who might be susceptible to serious side effects. For the pharmaceutical company, it means that they can define more accurately the patient population in which to conduct the clinical trial, and hence show enhanced efficacy and reduced incidence of adverse drugrelated events. This may mean that the target market for the drug being developed may be reduced but, equally significantly, it means that the attrition rate for drugs entering expensive Phase III trials will be substantially lessened, so the overall cost of development will be significantly reduced. Not only does this mean that the drug development process is more efficient, but it may also avert clinical problems that could arise when the drug is licensed. Since 13 drugs were removed from the market due to adverse drug reactions between 1997 and 2001, this represents a significant problem for the pharmaceutical industry.

Fortunately, advances in molecular diagnostic technologies that have emerged during the last decade have now provided the tools that allow the industry to identify with relative ease the markers that can assist with these problems. The simplest of these has been the analysis of genes that are involved with the metabolism of the drug in the body. Simplistically, if these genes are defective, then the drug may be ineffectively metabolised and so elevated levels of the drug will build up in the body, which may lead to toxicity (2). By the same process, any drug that relies on metabolic enzymes within the body to produce it in its active form will be limited in efficacy in those patients where the enzyme is defective and the active form cannot be produced. Conversely, some patients may have mutant forms of the relevant enzymes which are actually more active than the normal enzyme, and the drug may be metabolised much more quickly than usual. In those patients it may be necessary to use increased doses of the drug in order to elicit the required level of drug in the patient to maximise therapeutic benefit.

The importance of integration of pharmacogenetic information into medicine is now well accepted, and multiple metabolic pathways have been studied. As a result, we have a better understanding of which enzyme families relate to specific drugs and the details of the pharmacogenetic characteristics of that enzyme family. The most well understood area is the cytochrome P450 protein family, which has long been known to be involved in major pathways for metabolism of many different therapeutic agents. This protein family has been extensively studied, and multiple genetic polymorphisms have been identified in specific isotypes within the family that have significant effects on drug metabolism. Perhaps the most extensively studied interaction has been in the metabolism of the anti-clotting drug, warfarin, which is metabolised by the cytochrome P450 2C9 enzyme together with a second enzyme involved in Vitamin K metabolism (VKORC1). Warfarin is a particularly difficult drug to administer to patients because the dose required to control clotting varies significantly between individual patients. Although there are multiple factors involved in warfarin metabolism, it is becoming increasingly clear that the nature of the 2C9 and VKORC1 genes have a major impact on the ability to select the appropriate dose (3,4). By analysing the nature of these two genes in an individual patient, it is possible to identify, first, those patients who may have difficulty metabolising the drug (so would need dose reduction) and, secondly, those patients who metabolise the drug too fast (and hence require an elevated dose). This association between warfarin dosage and genotypic information has now been recognised by the US FDA, who have recommended that genotypic information be incorporated into warfarin administration protocols (5). The CYP450 family is involved in metabolism of many commonly used medicines (see Table 2), and pharmacogenetic analysis of the relevant CYP450 gene will have a significant effect on the correct use of all these drugs.

Apart from the CYP450 system, several other pharmacogenetic markers have also been identified and shown to be important in patient response to particular pharmaceuticals. These include thiopurine methyltransferase (TPMT), involved in the metabolism of the immunosuppressant azathioprine, and UDP glucoronosyltransferase 1A1 (UGT1A1), which regulates metabolism of the cancer drug irinotecan (6).

This growing acceptance of the importance of pharmacogenetics, and the anticipated requirement from the regulatory authorities that new drug submissions will be expected to have associated pharmacogenetic information, is the key to the core concepts of personalised medicine. The requirement to provide this information means that the pharmaceutical and biotech industry will be obliged to undertake detailed analysis of each new drug that they wish to progress to market, and this will be a major driver to the introduction of personalised medicine into routine clinical practice.

As further clinical data accumulates, it is possible that the FDA recommendation for warfarin may at some point become a requirement and that 2C9/VKORC1 analysis would then be legally required before the drug can be prescribed. In this situation, the diagnostic test would become a companion diagnostic, since the information provided by the diagnostic is essential for the clinical use of the drug. Apart from the pharmacogenetic diagnostic requirement, there are a growing number of examples where other types of genetic or other information are now required before certain drugs can be prescribed. Some of these drugs are indicated in Table 3, and it is worthwhile highlighting a few of these examples as they indicate some key principles around the use of companion diagnostics.

One area where molecular diagnostics have been heavily exploited for several years is in antivirals. Many assays are already available to monitor the appearance of antiviral drug resistance in order that drug therapies can be adjusted as specific resistant strains appear. Recently, within this area, Pfizer developed a drug – Maraviroc – which targets the CCR5 receptor, which is used by certain HIV strains to infect cells. This specific tropism is not present in all HIV strains because some of them do not use this receptor to infect cells and have other entry mechanisms. Since Maraviroc targets the CCR5 receptor, it will have no effect on HIV strains which use other routes (7). Pfizer therefore partnered with Monogram Biosciences to develop an assay that could quickly determine the tropism of the HIV strain in a particular patient and thus decide whether prescription of Maraviroc to that patient was appropriate. This is a good example of a companion diagnostic in routine use, but the major use of companion diagnostics incorporating genetic data is in the oncology area. A good example of this is the recent integration of companion diagnostics in the clinical management of patients with colorectal cancer.

Colorectal cancer is the third most common cancer worldwide in men and the second most common in women, and the incidence is increasing, particularly in southern and eastern Europe. Consequently, there are major efforts underway to develop a range of new therapeutics to treat this severe disease. In 2007, Amgen submitted clinical trial data to the Committee for Medicinal Products for Human Use (CHMP) in Europe regarding the efficacy of a monoclonal antibody – Vectibix^® – against metastatic colorectal cancer, but the data were rejected because of an apparent lack of efficacy in many patients. Vectibix^® is a drug which functions by interfering with the epidermal growth factor receptor (EGFR) pathway in the tumour and hence inhibiting its growth. Amgen retrospectively analysed the patients who had been included in the clinical trial process specifically to look at the genetic sequence of a gene called k-ras, which is involved in the EGFR pathway (8,9). They found that about 40 per cent of the patients in their cohorts had specific mutations in the k-ras gene, while the other 60 per cent were normal. Most importantly, they showed that the drug doubled median progression-free survival in patients with non-mutated k-ras. In other words, only patients with nonmutated (wild-type) k-ras responded to Vectibix^®. Following re-examination, in light of this new k-ras data, the CHMP reversed their decision and gave their approval in September 2007. This was the first time that the European Commission licensed a bowel cancer product with the stipulation that a predictive test should be carried out. Subsequently, Merck Serono also launched a monoclonal antibody therapy for the same indication, and they also found a correlation with the kras status. More recently, AstraZeneca has reported a similar story, linking the genetic sequence in the EGFR gene itself to the ability to respond to their lung cancer drug, Iressa. In this case, however, specific mutations in the tumour actually sensitise the tumour towards the drug, so it is the wild-type patients that do not respond to the drug (10). So, in these cases, the genetic information inherent in the tumour is used to determine the specific use of these individual drugs and is a clear indication of how, in addition to pharmacogenetics, the pharmaceutical and biotech industries are leading the way in the use of genetic information in personalised medicine.

Table 3 illustrates a selection of drugs that are now recommended or required to have associated companion diagnostic assays. As this list of drugs grows and the pharmaceutical industry begins to fully integrate detailed biomarker analysis into its clinical development programmes, there will be an increasing accumulation of genetic data either from the patient or the disease tissue into routine medicine. This transformation away from generic medicine also means that the pharmaceutical, biotech and diagnostics industries now need to coordinate their activities very closely, so that drug and diagnostic are both commercially available when the drug first enters the clinic. This also illustrates that the diagnostic is equally as important a factor to the clinician as the drug itself, since high quality detailed diagnostic information is critical to the most effective use of the drug. If the diagnostic assays are not robust and reliable, then the benefit to the patient will clearly be compromised.

CONCLUSION

The drivers to remove adverse side effects and to improve drug efficacy are emanating from multiple sources. These include clinicians who want to see drugs working more efficiently so that they can treat people more effectively, the payers who reimburse healthcare in terms of drug-related costs, hospitalisation, and, most significantly, the pharmaceutical regulators, which increasingly require new information to be associated with clinical development of new pharmaceuticals. All this means that in the medium- to long-term outlook, the majority of drugs that are licensed will have specific diagnostic requirements in terms of basic pharmacogenetic information, or detailed analysis of the genetic, biomarker or other profile of the particular disease. This will have two significant implications. First, it means that fully validated, robust and regulatory approved diagnostic tests will need to be routinely available in all major markets where the drug needs to be licensed. This has major implications on the interface between the pharma/biotech industry and the diagnostic manufacturer/clinical service provider because the diagnostic test is effectively as important as the drug; that is to say, if the test gives false information or is unreliable, then the drug could still be used for the wrong patient at the wrong dose. Similarly, if the test is highly complex or locally unavailable, this could restrict the availability of the new therapeutic. Consequently, these two components of the healthcare industry need to align their business models very closely to deliver the tools and the drugs in parallel. The second implication is that increasing amounts of patient-specific pharmacogenetic information will need to be available for an individual patient to realise optimal healthcare benefits. Since this pharmacogenetic information will be constant during a patient’s lifetime, the logical extension to this process will be to introduce pharmacogenetic profiling in the neonatal setting at the same time as blood group analysis, and screening for genetic diseases such as cystic fibrosis and MCADD. This information can then be linked to a patient’s medical records and, given the appropriate bioinformatic systems, be routinely available for a physician to consult at any time and for any illness that may arise during that patient’s lifetime.

References

1. Gurwitz D and Motulsky AG, Drug reactions, enzymes and biochemical genetics: 50 years later, Pharmacogenomics 8 (11): pp1,479-1,484, 2007
2. Boonsirat U, Angsuthum S, Vannaprasaht S et al, Azathioprine-induced fatal myelosuppression in systemic lupus erythematosus patient carrying TPMT*3C polymorphism, Lupus 17: pp132-134, 2008
3. Higashi MK, Veenstra DL, Kondo LM et al, Association between CYP2C9 genetic variants and anticoagulationrelated outcomes during warfarin therapy, JAMA 287: pp1,690-1,698, 2002
4. Rieder MJ, Reiner AP, Gage BF et al, Effect of VKORC1 haplo-types on transcriptional regulation and warfarin dose, N Engl J Med 352: pp2,285-2,293, 2005
5. FDA approves updated warfarin (Coumadin) prescribing information, Press release of the Food and Drug Administration, Rockville, MD, August 16, 2007; accessed 14 February 2008, at www.fda.gov/bbs/topics/NEWS/2007/NEW01684.html
6. Fargher EA, Tricker K, Newman W et al, Current use of pharmacogenetic testing: a national survey of thiopurine methyltransferase (TPMT) testing prior to azathioprine prescription, J Clin Pharm Ther 32: pp187-195, 2007
7. US FDA, Prescribing information for Maraviroc (Selzentry, Pfizer) www.accessdata.fda.gov/drugsatfda_docs/label/2007/022128lbl.pdf
8. Jimeno A, Messersmith WA, Hirsch FR et al, KRAS mutations and sensitivity to epidermal growth factor receptor inhibitors in colorectal cancer: practical application of patient selection, J Clin Oncol 27: pp1,130-1,136, 2009
9. Allegra CJ, Jessup JM, Somerfield MR et al, American Society of Clinical Oncology Provisional Clinical Opinion: Testing for KRAS gene mutations in patients with metastatic colorectal carcinoma to predict response to anti-epidermal growth factor receptor monoclonal antibody therapy, J Clin Oncol 27: pp2,091-2,096, 2009
10. Rosell R, Moran T, Queralt C et al, Screening for epidermal growth factor receptor mutations in lung cancer, N Engl J Med 361: pp958-967, 2009