Pharmacogenomic-guided drug development: regulatory perspective
|L J Lesko1 and J Woodcock2|
1Office of Clinical Pharmacology and Biopharmaceutics, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, MD, USA
2Office of the Center Director, Center for Drug Evaluation and Research, Food and Drug Administration, Rockville, MD, USA
Correspondence to: L J Lesko, PhD, Director, Office of Clinical Pharmacology and Biopharmaceutics, HFD-850, Center for Drug Evaluation and Research, Food and Drug Administration, 5600 Fishers Lane, Rockville, Maryland 20852, USA. Tel: +1 301 594 5690 Fax: +1 301 480 8329 E-mail: email@example.com
The Pharmacogenomics Journal (2002) 2, 20–24. DOI: 10.1038/sj/tpj/6500046
Pharmacogenetics (PGt) and now the more global term, pharmacogenomics (PGx), have come to the forefront after an evolutionary period of more than 30 years. Several transforming events in the past 5 years, not the least of which was the completion of the human genome sequence in 2001, have created an expectation that genetic and genomic information will produce sweeping changes in the practice of medicine and the prescribing of drugs. The almost daily press reports of new gene discoveries lend credence to the argument that personal genetic/genomic profiles will have a tremendous impact on health by the year 2010. This vision has not been without naysayers; however, we believe that the central issue is not whether PGt- or PGx-guided drug prescriptions will happen, but when and how.
Genomic information has the potential to revolutionize pharmacologic therapies at many levels. The process of drug discovery may be transformed by this knowledge. Extensive genetic data will promote understanding of the molecular genetic contribution to many diseases. Genes and gene products suspected of being involved in disease pathogenesis will become new targets for intervention, and will stimulate new drug discovery programs. Conversely, gene expression profiling is being used currently to gain new insights into the molecular mechanism of drug actions, and the drug–disease interaction. Taken together, these techniques are expected to yield major advances in identifying drug candidates.
Genomic information will be increasingly used in the preclinical phases of drug development. There is great interest in using gene expression profiling to develop markers for both desired pharmacologic actions and toxic effects. Batteries of markers will then be used to characterize drug candidates and to aid in selection of those with optimal properties for further development, thus improving the effectiveness of drug development.
At the clinical level, the hope is for true individualization of therapy, which would maximize benefit and minimize toxicity. Currently, clinicians have few tools for predicting who will respond to a drug, or who will suffer ill effects. Although such differential responses have long been characterized as ‘idiosyncratic’, clearly there are underlying reasons for them, and many have a genetic component. It is believed then most chronic diseases represent a heterogeneous group of disorders at the molecular level. This heterogeneity is one of the reasons that not all people with a disease respond to a given drug. One contribution of genomic science could be to provide a much more precise diagnosis, based either on underlying genotype, or on gene expression profiles. Similarly, some differences in drug efficacy response, and some toxicities, are based on variability in exposure or in pharmacodynamic response, caused by genetic differences. The ability to predict and account for such differences could markedly improve the therapeutic index of many drug interventions. Finally, it is hoped that genetically-based mechanisms of toxicity can be elucidated, and adverse effects avoided, by application of pharmacogenomic information.
The explosion of interest in PGt and PGx has raised concerns that the regulatory environment could inhibit progress. While drug discovery and preclinical studies are not likely to be significantly affected, there is concern about the use of PGt/PGx in clinical trials. This article provides an overall regulatory perspective on the clinical study issues and considerations, many of them presently unresolved, that PGt and PGx present to the drug development and regulatory decision making processes. It does not extensively discuss the development, validation and usage of diagnostic kits, although this is an important issue to FDA. We acknowledge that there are also many other stakeholders (eg, managed health care agencies, insurance companies) and issues (eg, privacy, ethics) in the debate about PGt and PGx, but those domains will not be part of this article. We hope that we can provide a greater understanding of the issues and an agenda of topics that will need resolution through effective communication among scientists in academia and the industry, and those in the FDA and other regulatory agencies.
We are not aware of any consensus on the definition of PGt and PGx, and in fact there are many different definitions in the scientific literature. Occasionally, these terms are used interchangeably. For the purposes of this article, we will consider PGx to be the global science of using genetic information from an individual or population for the purpose of: (1) explaining interindividual differences in pharmacokinetics (PK) and pharmacodynamics (PD); (2) identifying responders and non-responders to a drug; and (3) predicting the efficacy and/or toxicity of a drug. Also, we will consider PGt to be a scientific subset of PGx in which there are genetic variations (eg, polymorphism in cytochrome P-450 metabolizing enzymes) to drug doses and dosing regimens that result in different systemic drug exposure patterns (PK) in individuals or populations.
PGx technologies are still evolving but in the past 5 years there has been a tremendous growth rate in gene databases, high-throughput DNA microarray methodologies and SNP analysis tools. Methods in PGx/PGt are becoming more cost-effective and widespread throughout the pharmaceutical industry, and with the co-development of bioinformatics software and computerized decisional analysis tools, these methods are becoming much more informative as well.
In 1998, the Secretary’s Advisory Committee on Genetic Testing (SACGT) was chartered to advise the Department of Health and Human Services (DHHS) on the medical, scientific, ethical, legal and social issues raised by the development and use of genetic tests. One of FDA’s concerns is to ensure the quality of PGx/PGt tests that might be used in clinical drug development, and public access to quality PGx/PGt tests. While not directly bearing on the drug development process, the major issues, overarching principles and final recommendations of SACGT can be extrapolated to issues and questions related to PGx/PGt tests used in drug development. The final report of SACGT, issued in July 2000, is publicly available on the Internet (http://www4.od.nih.gov/oba/gtdocuments.html).
Traditionally, drug development is divided into the discovery phase, preclinical phase and three clinical phases (I, II and III). In the past 10 years, the most dramatic changes in drug development have occurred in drug discovery due to the technological advancements in combinatorial chemistry, high throughput screening and molecular targeting. Animal studies continue to provide a fundamental understanding of the pharmacological action of drugs and represent a major preclinical effort to determine the feasibility of continuing with further development in humans. Early clinical studies in Phase I and Phase II (sometimes split into IIA and II are intended to be exploratory in terms of estimating appropriate doses and dosing regimens for later trials, demonstrating proof of therapeutic concept and making decisions to continue with further development. During Phase III, large-scale clinical studies are conducted to confirm efficacy in a target population with a specified indication, and to confirm the relative safety of the drug. Concurrent with Phase III, sponsors frequently conduct so-called registration studies (eg, drug–drug interactions) to provide the additional information necessary to adequately label the product. Under the current system, the drug development process that leads to a traditional review and approval (ie, not accelerated) requires typically that a sponsor conduct at least two independent, adequate and well-controlled clinical trials in Phase III to provide crucial evidence that a drug is effective and relatively safe. Safe, in a regulatory context, means that the Agency has deemed the drug to have an acceptable risk/benefit ratio for a given indication in a statistically significant number of patients in a defined population, when compared to a placebo or to an accepted standard therapy.
The FDA believes that a drug can be determined to be effective and safe only when the relationship of beneficial and adverse effects to a defined exposure (eg, dose or plasma concentrations) is known. Exposure-response relationships define the drug’s safety margin as well as its dose- or plasma level-limiting side effects. It is well known that in drug development clinical trials, traditional medical histories and the usual demographic factors often fail to explain the inter-individual variation in exposure-response relationships. The Agency is very interested in understanding the intrinsic (eg, age, disease state, genetics) and extrinsic (eg, drugs, diet, smoking) factors that may alter the exposure-response relationship in patients. So-called registration studies, ie, definitive PK studies in patients subgroups defined by renal and/or hepatic disease, age, gender, race and drug interactions are routinely conducted in today’s drug development programs. These data are reported in the package insert in either the clinical pharmacology, precaution/warning or contraindication sections and, where appropriate, dose and dosing interval adjustments are provided for these patients in the dosage and administration section. In the past, most of the exposure-response variability was attributed to differences in PK, and as a result, substantial attention has been focused on PGt, especially as it affects the activity of the CYP-450 drug metabolizing enzymes and results in inter-individual variability in the dose-exposure relationship. Despite this focus, PGt and PGx overall have not had a major role in clinical drug development to date.
Over the past 5 years, many have expressed the concern that human clinical efficacy and safety trials in a traditional drug development program are challenging, time-consuming and increasingly more expensive to conduct. The relatively high rate of failure of drug candidates entering the clinical phases of drug development add significantly to these estimated costs. More recently, new drug candidates have been filtered from the discovery and development pipeline because their hepatic metabolism requires CYP-450 enzymes subject to genetic polymorphism. Several experts in the science of drug development perceive the increasing costs, and recent decreasing return on investment, as a significant threat to the viability of the pharmaceutical industry in the next 10 years.
Conventional wisdom suggests that PGx-guided clinical trials would shift the drug development paradigm toward a more efficient and informative process, resulting in a lower attrition rate of new drug candidates, and an overall lower development cost to the sponsor, albeit in the long-run. Furthermore, many believe that drug therapy based on the genetic profile of individuals could provide public health benefits such as better management of post-approval risks, and a decreased incidence of drug-induced morbidity and mortality. PGx could also reduce the incidence of drug product market withdrawals due to serious or fatal adverse events by allowing pre-selection, in advance of prescribing the drug, those patients who will be predisposed to toxicity. That the pharmaceutical industry is investing in this promise is no more evident than by the high level of interest and cooperation within the SNP (single nucleotide polymorphisms) Consortium, a collaboration of pharmaceutical companies, bioinformatic enterprises and academia established to explore SNPs as a drug development tool. SNPs are single-base differences in the DNA sequence that can be discerned between individuals in a population. SNPs are the most frequent type of variation in the human genome and are attractive as biomarkers for the drug discovery and development process. The challenge will be to identify correctly those SNPs that influence or change pharmacokinetics, pharmacodynamics and/or clinical endpoints.
However, pharmaceutical companies appear to be moving forward cautiously in integrating PGx/PGt in their drug development programs. As a result, FDA has limited experience with clinical pharmacology and clinical efficacy and/or safety studies in which sponsors have prescreened subjects using PGx/PGt tests. To increase our awareness, the Office of Clinical Pharmacology and Biopharmaceutics conducted a survey of recent Investigational New Drug (INDs) and New Drug Applications (NDAs) to identify the extent to which PGx/PGt was used in clinical studies. The survey, still being analyzed, found over 15 applications in which PGx/PGt tests were reported with all but one test related to pharmacogenetic variability in CYP enzymes. The tests were used to either phenotype or genotype CYP 2D6 in seven submissions, 2C19 in five submissions and CYP 2C9, 3A4 or other metabolizing enzymes respectively in one submission each. The use of this information as reported by the sponsors was as follows: (1) to define patients subgroups in either early phase PK studies and/or late phase efficacy trials using sparse sample analysis and population PK methodologies to assess the significance of geno-/phenotype as a co-variate; (2) to provide a post hocexplanation for the variability in drug exposure (and response) in some subjects and/or patients (eg, outliers where plasma levels were too high or low) as a basis to exclude such subjects from an analysis; (3) to measure the impact of CYP enzyme polymorphism on plasma drug clearance in subject subgroups defined by PGt; and (4) to use the results of PGx/PGt as an entrance or exclusion criteria for subjects in drug interaction studies. While providing insight into the inter-individual variability in the dose-plasma drug level-response relationship, differences thought to be related to PGx/PGt subgroups were not used as a basis for any specific dosing recommendations for these subgroups in the product labels.
It is hoped that more extensive use of PGx/PGt information will be utilized in future clinical trials. However, there are many unanswered questions about the FDA’s regulation of clinical development programs using various elements of PGx/PGt. The following issues and questions, which need further discussion, are among the major concerns of the industry with regard to drug development:
In addition to these general questions, there are further uncertainties based on new genetic information, which have not been addressed formally to our knowledge. For example, it is well known that there are demographic factors (eg, race) which influence the response to drugs in several pharmacological classes such as beta-blockers and ACE inhibitors.1,2 More recently, it has been reported that patients with high cholesterol who have the B1B1 variant of the CETP gene will have greater reductions in blood lipid levels when treated with certain lipid-lowering ‘statins’.3 Similarly, subsets of women with breast cancer defined by mutations in BRCA1 and BRCA2 susceptibility genes may be more or less responsive to tamoxifen chemoprevention.4 What if these types of drugs were currently under development? Would the Agency expect that PGx/PGt testing be done to determine the cause of these variabilities in response? Would such tests be part of the drug approval process? These are only a few examples of the many issues and questions, which may either be decided on a case-by-case, or lead us to further discussion and the development of an over-arching set of principles to be used in drug development and regulatory decision-making.
Successful use of PGX/PGt-customized medicines will require physicians to: (1) select patients for drug therapy before writing a prescription; (2) exclude patients from drug therapy before writing a prescription (based on predicted toxicity or poor response); (3) select the optimal individual dose and dosing regimen; and (4) evaluate the genetic basis for an adverse event. The safety and effectiveness profile of drugs developed under PGx-guided programs may thus be predicated on successful PGx-guided prescribing.
As part of the drug review and approval process, the FDA will have to assess how well PGx information will be translated into clinical practice. Will physicians and other health care professionals, who are reported to be prone to information overload, be able to learn about, understand and then alter their professional practice to incorporate PGx-directed prescribing? In some cases (eg, Hercept for breast cancer) this has been successfully accomplished. In other instances, information important to successful drug therapy has not penetrated medical practice. For example, scientists have known for many years that individual variations in genes (polymorphism) that encode and determine the activity of Phase I (eg, 2D6) and II drug metabolizing (eg, N-acetyltransferase) enzymes are a major source of interindividual variability in systemic exposure from drug doses. In contrast to the acceptance of PGx/PGt tests to select patients for Hercept therapy in mainstream oncology, PGt-guided tests to identify gene variants in patients that influence the activity of cytochrome P-450 drug metabolizing enzymes have not met with widespread acceptance despite the demonstrated cost-effectiveness of such a strategy. For example, it has been well known for 30 years that phenotyping and genotyping for the activity of certain CYP enzymes (eg, CYP 2D6) can be used to identify a large number of patients in the general population who carry a particular variant of a gene that is an important determinant of exposure. It has been shown that clinical outcomes are related to the metabolic genotypes for drugs such as omeprazole (a 2C19 substrate) in the treatment of H. pylori. In addition, a subset of patients are unable to convert drugs to metabolites (eg, poor metabolizers of CYP 2C9 or 2D6) and are prone to either a higher risk of toxicity, or a lack of efficacy, than extensive metabolizers when administered the usual doses of drugs such as warfarin or codeine respectively. However, identifying these patients in advance of selecting a drug dose is an exception rather than a rule in medical practice.
For these reasons, PGx-directed clinical drug development programs may need to include strategies to evaluate the applicability of genetic/genomic testing to clinical practice. While this may be straightforward in disciplines such as oncology, incorporating genetic/ genomic testing into more general medical practice will undoubtedly be more challenging.
To advance the notion that PGx/PGt information is of value to drug development and regulatory decision making, more prospective clinical trials in which this information is integrated into study design and data analysis need to be planned and conducted. To date, established or unequivocal evidence of the value of PGx/PGt is limited with few real-life examples. Nevertheless, the FDA is very interested in discussing or reviewing drug development program proposals that intend to use PGx/PGt in clinical studies.
Ideally, a regulatory agency should be able to meet its public health mandate without stifling new technology that might lead to better drugs including the ‘customized medicines’ of the PGx/PGt era. It should be noted that the FDA went on record of supporting the basic idea of ‘customized medicines’ for a patient subset, back in 1998, as evidenced by its approval of trastuzumab (Hercept). The approved indication of trastuzumab, a recombinant DNA-derived humanized monoclonal antibody, was for the treatment of only those patients with metastatic breast cancer whose tumors overexpress the protein, HER2, in large amounts. This patient subset represents up to 30% of all women with breast cancer. The FDA has also approved one prognostic PGt assay and two PGx immunohistochemical assays to measure HER2 neu protein overexpression to be used in patient selection before prescribing trastuzumab treatment. It is likely that FDA would not have approved Hercept without the accompanying diagnostic test. Also, the recent approval of Gleevec (imatinib mesylate) in May 2001 for late phase chronic myelogenous leukemia (CML) is another example that the CDER is well aware of, and open to, individualization of drug therapy using PGx or other research strategies. The discovery and development of imatinib, while not strictly PGx-driven, is a good example of the type of molecular targeting to abnormal proteins, in this case in CML cells, that is possible with the help of PGx information. In addition, the Center for Biological Evaluation and Research (CBER) has extensive experience with gene therapy development for over 10 years and has reviewed applications for genetically engineered protein drugs for such indications as sepsis and hemophilia.
The FDA has not been idle in the face of emerging PGx and PGt technologies but is preparing more for the predicted influx of NDAs containing PGx/PGt information. We are carefully taking the following steps to assure that, as a regulatory agency, FDA is prepared to deal with the future influx of PGx/PGt data in submissions from IND and NDA sponsors:
A search of several websites of national regulatory agencies using keywords such as pharmacogenetics, pharmacogenomics and genetic polymorphism resulted in relatively few ‘hits’. For example, in searching the FDA website (http://www.fda.gov/cder/regulatory/default.htm) containing 144 180 documents there were 54, 36 and 25 ‘hits’ for the terms pharmacogenetics, pharmacogenomics and genetic polymorphism respectively. However, several guidances for industry published by regulatory agencies as domestic guidances or as a result of the International Conferences on Harmonization (ICH) contain general concepts or principles related to PGx, including the following:
FDA has also taken notice of the recent explosion in PGt and PGx information and realizes the need to actively engage in internal discussions, and in open dialogue with the pharmaceutical industry and to identify the new implications, questions and issues related to drug (and device) approvals. If necessary, the FDA is prepared to develop new domestic guidances or work through ICH to develop new harmonized guidances with Europe and Japan.
The parallel developments in the areas of PGx and PGt technology and bioinformatics, and the unraveling of the human genome hold much future promise for the development of drugs and dosing instructions that will alter individualization of therapy. But, who will lead the way?
At present, patient genomic/genetic data from prospective clinical and clinical pharmacology studies are necessary to: (1) evaluate the role that PGx can play in drug development; (2) identify issues that will trigger more urgent and extensive discussion between the Agency and industry; (3) focus the regulatory review on the important science/clinical questions and determine what evidence is necessary to support label claims. We continue to be concerned that despite the widespread availability of simple PGx/PGt tests to determine a patient’s phenotype and/or genotype with regard to polymorphism in drug metabolizing enzymes, there has been little use of this information to tailor drug doses and dosing regimens to individual patient subgroups in clinical practice before using the drug. Together, all of the stakeholders in PGx/PGt need to work on ways to assure that this does not happen with the second and third generation of PGx/PGt diagnostic tests. We conclude that the bridge between the current and emerging PGx/PGt research in drug development and regulatory review practices, and related policy, needs to be built systematically and on a sound scientific foundation. The gap between research and the use of PGx/PGt in clinical practice remains very wide, but we are encouraged with the progress that is being made to close this gap.
DUALITY OF INTEREST
|2002, Volume 2, Number 1, Pages 20-24|