Emerging Technologies in Drug Development

Important Points Highlighted by Individual Speakers

•   Understanding complex biological networks may require complete genome sequencing of many millions of people.

•   In-depth sequencing may be required to find relevant variants with both known and unknown effects.

•   Genomic technologies are improving rapidly and soon will be able to provide very rapid sequencing with high sensitivity.

•   Complete genomic information may require new regulatory approaches—for example, when considering the approval of new drugs and diagnostics targeted at rare mutations.

•   Phase I clinical trials may be enhanced by requiring that all participants have their genomes sequenced so that they can be assigned to different trials.

New paradigms in drug discovery and development rely on new technologies. These include not only DNA sequencers but a wide variety of new tools for gathering, analyzing, and disseminating genetic and genomic information. Three speakers at the workshop discussed technologies now being used and under development for genomic-based drug discovery and development. As they noted, these technologies collectively have the potential to reshape drug development in fundamental ways.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 29
4 Emerging Technologies in Drug Development Important Points Highlighted by Individual Speakers Understanding complex biological networks may require com- plete genome sequencing of many millions of people. In-depth sequencing may be required to find relevant variants with both known and unknown effects. Genomic technologies are improving rapidly and soon will be able to provide very rapid sequencing with high sensitivity. Complete genomic information may require new regulatory approaches--for example, when considering the approval of new drugs and diagnostics targeted at rare mutations. Phase I clinical trials may be enhanced by requiring that all participants have their genomes sequenced so that they can be assigned to different trials. New paradigms in drug discovery and development rely on new tech- nologies. These include not only DNA sequencers but a wide variety of new tools for gathering, analyzing, and disseminating genetic and genomic information. Three speakers at the workshop discussed technologies now being used and under development for genomic-based drug discovery and development. As they noted, these technologies collectively have the poten- tial to reshape drug development in fundamental ways. 29

OCR for page 29
30 GENOME-BASED THERAPEUTICS LARGE-SCALE WHOLE-GENOME SEQUENCING Almost all genetic variants have contextual expression and meaning that depend on other genomic sequences and environmental factors inter- acting through complex regulatory networks. Given the importance of con- text, the accurate interpretation and effective use of genetic instructions are impossible with only partial access to genetic codes, said Radoje Drmanac of Complete Genomics. Furthermore, each person has 10,000 to 100,000 family-specific genetic variants along with approximately 100 de novo personal variants in addition to a few million population variants. No comprehensive predefined genetic variant chips can be designed to detect such a wide range of variability. Whole-genome sequencing provides a maximum level of strictly genetic information, Drmanac said. By providing greater understanding of disease, it has the potential to produce greater efficacy, safety, and overall success in drug development. Whole-genome sequencing has two main areas of application: biologi- cal understanding and genomic medicine. Understanding the molecular and genetic bases of thousands of human diseases, developing better targeted drugs and other therapies, including those for disease prevention, and devel- oping personal genome interpretation software will require the sequencing of millions of genomes, Drmanac said. Today perhaps only 10,000 genomes have been sequenced, so developing a true understanding of disease biology requires sequencing on a much larger scale. The world contains billions of people, Drmanac noted. "If we're really serious and want to take whole-genome sequencing as the basis for medi- cine, we need to sequence billions of genomes." But sequencing on that scale will require that the process be industrialized. Small processes may have benefits for specialized applications, but large-scale massively parallel whole-genome sequencing is needed and this must be designed and opti- mized to achieve high quality and low cost. Complete Genomics has been developing a turnkey service that enables customers to outsource whole-genome sequencing. Customers send samples to the company and receive data in return. Furthermore, the data received are not just sequences but the fully assembled genome with an annotated list of informative sequence variants, with each base marked as reference, variant, or a no-call. Each variant has a confidence score to balance sensitiv- ity and specificity, and variations in known protein coding and regulatory gene sequences are identified. Drmanac acknowledged that for medical applications it is necessary to include an interpretation of the data, and Complete Genomics is currently working on developing a system for this purpose.

OCR for page 29
EMERGING TECHNOLOGIES IN DRUG DEVELOPMENT 31 Increasing the sequencing coverage of the DNA makes it possible to call bases for 96 to 97 percent of the genome with very high confidence. How- ever, there are certain regions of the genome that are difficult to sequence, Drmanac said, and within the majority that can be sequenced there are still about 4,200 errors. However, by eliminating the 5 percent of calls that have lower confidence, the error rate for the remaining calls is just 366 per genome, or 1 for every 7 megabases. "We can get to clinical quality, even with our standard process," he said, "and we have other processes that will further improve [quality]." Complete Genomics expects to be able to sequence 2,000 genomes per month by the end of 2012, up from 800 at the time of the workshop. Better instruments could boost that rate to 100,000 genomes per year, and the company has the goal of being able to do millions of genomes per year. "That's coming," Drmanac said. "We don't need to wait for new inven- tions. It's just regular improvements." The company has a diverse base of more than 100 users, including organizations from government, industry, academia, and private industry. These organizations have been using the technology to investigate cancer, de novo mutations, genomic variation and disease, and the potential of translational medicine. "It's a golden age where we can sequence millions of genomes in cancers, in unknown disease, and in frequent disease," Drmanac said. "It can be done in the next 5 to 6 years." Drmanac proposed the idea of a million-genome project, or perhaps a 10-million-genome project, to understand both health and disease. "Hope- fully we can reduce cost at the same time," he added. "It's very difficult and not guaranteed, but I think with the genome we hopefully can do that." Sequencing capabilities will continue to grow exponentially, both in terms of capacity and accuracy. In a few years, 1,000 times coverage may be routine. A genome sequence could be done at birth--or even preimplantation--to provide a foundation for health care and research throughout life. Understanding of biological function is exploding, and sophisticated software to interpret genomes is undergoing rapid develop- ment. Given this rapid growth in understanding and capability, perhaps all people enrolled in clinical trials should have their genomes sequenced to minimize risks and maximize efficiency, Drmanac suggested. Genomic data will need to be integrated with complete transcriptome and epigenome analysis for the relevant tissues of each subject, but whole genome sequenc- ing can provide a foundation to integrate over other kinds of data. In this way, drugs could be developed for distinct genomic states and individual DNA sequences, such as genetically engineered stem cells or drugs that attack only cancer cells with specific DNA signatures.

OCR for page 29
32 GENOME-BASED THERAPEUTICS THE VALUE OF CLINICAL NEXT-GENERATION SEQUENCING TO DRUG DEVELOPERS For a variety of reasons, the current model of drug development is not sustainable, observed Gary Palmer of Foundation Medicine. Limited amounts of tissue biopsies are available to search and test for informative biomarkers, and screening patients for rare markers is inefficient. Prospec- tive studies are limited by turnaround times for analysis, and retrospective studies can have difficulties securing the appropriate samples. Finally, com- plex biology requires a better knowledge of disease pathways and thorough interpretation, not just raw data. These problems can be further exacerbated with clinical cancer sam- ples, Palmer said. Cancerous cells may be only a small fraction of the total sample. Multiple sub-clones of cancer may be present in one sample, and chromosomal gains and losses may modify the abundance of a mutation. As a result, relevant mutations may be rare in a pool of sequenced DNA. Next-generation sequencing can address many of these issues, Palmer said. In particular, Foundation Medicine was established to help clinicians and pharmaceutical companies screen patients for targeted therapies that are already on the market or that soon will become available. The company works with very small (40 micron) samples of paraffin-embedded tissues from pharmaceutical companies, oncologists, or pathologists. At the time of the workshop, it was analyzing 182 genes that are known to be somati- cally mutated in cancer, with plans to increase that number to 250 in the near future. "We sequence the heck out of them," Palmer said, with cover- age rates averaging close to 900. The analysis is optimized for accuracy and is designed to produce annotation, interpretation, and a clinical report in 14 to 21 days. The analysis identifies point mutations, short insertions and deletions, copy number alterations, and a select number of genomic rearrangements. The interpretation is particularly important for oncologists, Palmer said. It includes linkage to FDA-approved drugs, either on- or off-label, and connections to clinical trials that are available. In the company's initial stages, interpretation often has involved genetic alterations that need to be curated for the first time, which requires figuring out an alteration's possible effects. "As we're seeing more and more things, that's going to be a more efficient process," he said, "and the turnaround time will get less and less." The service provided by Foundation Medicine is situated somewhere between whole-genome sequencing and "hotspot testing," which looks for a specific abnormality or group of abnormalities. The company does in- depth sequencing on its panel of selected genes where abnormalities may be of importance. In the process it finds not only expected abnormalities but also other abnormalities that show up repeatedly in different tumors

OCR for page 29
EMERGING TECHNOLOGIES IN DRUG DEVELOPMENT 33 in varying percentages, which is why the same panel is run for all tumors. Furthermore, many of these abnormalities are actionable, in that they are the target of an FDA-approved drug either on- or off-label. Variants of unknown significance are also identified, although they are not reported. Still, Foundation Medicine is studying these abnormalities to elucidate their role in cancers. The company has been doing several types of studies with pharmaceuti- cal partners. It is running single-agent clinical trials, longitudinal studies, multiple simultaneous Phase I trials in which individual samples are tested for multiple biomarkers, and analyses of samples from failed Phase III trials that did not meet clinical endpoints but produced evidence of responders for whom a drug might be valuable. "This is very attractive to pharma companies that don't want to let a drug die . . . and see if, in fact, there is a reason that they can move forward," Palmer said. Pharmaceutical partners have several requirements that they need Foundation Medicine to meet. They want the ability to work with paraffin- embedded tissues so they can retrospectively analyze samples. For prospec- tive studies, they want clinically relevant turnaround times. They want deep sequencing coverage, so that relevant alterations will not be missed. They also want help with genomic insights: What does a particular altera- tion mean? Finally, they want assistance with computational biology either because they are not able to do it themselves or because they have made the decision to outsource for the technology. The provision of these services can provide great benefits for the pharma ceutical industry, Palmer said. It can aid in biomarker identifica- tion, help stratify patients for clinical trials, help determine resistance markers, enable combination therapies, and assist in resurrecting clinical trials. What next-generation sequencing cannot do is increase the enroll- ment rate, meaning that very rare but actionable alterations will still require that many patients be screened. Similarly, it cannot overcome problems with statistical power. Policy makers need to understand the stakes behind this approach to genomic medicine, Palmer said. For example, a separate validation for each marker is not feasible. Next-generation sequencing does not test for a spe- cific marker because thousands of results are possible. As the testing meth- odology that is being employed at Foundation Medicine could potentially be used as a companion diagnostic for the development of a therapeutic, a clear pathway is needed for these newer technologies. It is possible that next-generation sequencing will be required in Phase I of all clinical trials, Palmer said. Potential patients could then be placed in the appropriate trial through next-generation sequencing, including combi- nation therapy trials and "case report" trials, so that label extensions could be based on multiple N-of-1 reports.

OCR for page 29
34 GENOME-BASED THERAPEUTICS Finally, oncologists cannot keep up with the deluge of new information that is being generated. For example, even today oncologists are split on whether to try a drug off-label when they find abnormalities in different tumor types. Currently, many patients are not receiving the testing they need, which means that they are not getting proper therapies, and this situ- ation will become worse over time. It will be of critical importance, Palmer said, to have initiatives to educate providers, patients, regulators, payers, and other relevant stakeholders. "There is a lot of potential improvement in patient care that is available by just technology and I think also would lead to improvement in pharma recruitment and [patient outcomes]." THE USES OF GENOMIC INFORMATION The development of personalized healthcare approaches depends sub- stantially on the strength of the hypothesis being developed, said Jane Fridlyand of Genentech. Where a strong diagnostic hypothesis exists, a strong scientific rationale allows for patient selection through all stages of development. This situation can provide a relatively fast development path and a relatively straightforward path to approval. Without a strong diagnostic hypothesis, patient selection is much more difficult. In that case, retrospective data exploration and planning for future data collection need to be emphasized, Fridlyand said. In many cases, a biomarker and drug have been developed, but it is not clear in which population the drug will work. In this situation the three key challenges are label-enabling trial design and analysis, biomarker cut- off and refinement, and multiple biomarkers and multi-marker tests. The developers of drugs and diagnostics need to decide which patient popula- tions should be tested. These decisions ultimately depend on the scientific rationale and clinical context, Fridlyand said. Key questions include What is the clinically meaningful treatment benefit level? What is the magnitude in benefits between diagnostic-positive and diagnostic-negative patients? What is the riskbenefit evaluation in diagnostic-negative patients, and what are their unmet medical needs? Biomarker indicators are often continuous, requiring that a threshold be set for selection and refinement of patient groups. The result is a tradeoff between population size and effect size. Often there is no clear best thresh- old. Instead, the appropriate threshold depends on the clinical riskbenefit, the scientific rationale, and the distribution properties of the biomarker (such as whether it is bimodal). A useful strategy, Fridlyand said, is to

OCR for page 29
EMERGING TECHNOLOGIES IN DRUG DEVELOPMENT 35 base a threshold on limited data from a Phase II study and then adjust the threshold based on a much larger volume of data from Phase III. Fridlyand reiterated a point made by Palmer, which is that full clinical validation of each biomarker may not be feasible. Infrequent mutations may be too rare to be validated. For multiplex biomarker signatures, there is no expectation that individual biomarkers will be predictive. A particular pathway activation that modulates the activity of a drug may be activated only when expression of multiple genes in other pathways are on or off, and expression might occur at different prevalences. Instead, it will be necessary to consider alternative clinical and analytical validation, depending on the situation. "It's going to be really hard to show that each of these members of your companion diagnostics are clinically validated," Fridlyand said. "Rather, we have to think about it as a clinical validation for a summary measure." Analytical validation also raises issues. Testing multiple biomarkers will use up tissues, and eventually patients will not be willing to give more, Fridlyand said. Furthermore, physicians do not want to put patients in trials where they do not think the patient will benefit. Overcoming these issues will require extensive drug and technology pipelines, effective collaborations, cross-functional teams, and strategic information gathering to provide robust datasets for better-informed and efficient drug development. In addition, said Fridlyand, the development strategy needs to include feedback from development to research in order to improve disease molecular classification and biomarker hypothesis generation.

OCR for page 29