Drug Repurposing and Repositioning: Workshop Summary (2014)

3

Enabling Tools and Technology

Repurposing and repositioning have gained new momentum in part because of the development of new tools and technologies, individual speakers at the workshop said. Large databases, genome-based informatics capabilities, and contract research organizations (CROs) have all enabled advances in drug development. These tools and the development of other technologies can aid in repurposing and repositioning efforts.

DISCOVERY USING PUBLICLY AVAILABLE DATA

In 2001 journals began to require that authors using microarrays deposit their data into repositories, said Atul Butte, chief of the division of systems medicine at Stanford University’s Center for Pediatric Bioinformatics. Since then more than 1 million microarray datasets have become publicly available, and the number is doubling about every 2 years (Baker, 2012). Many other kinds of data are becoming publicly available at comparable rates, including molecular, clinical, and epidemiological data. These data could be used to find new uses for existing therapeutics. More data are always better, but data are already plentiful today and should be used even if they are not perfect, Butte said, because using what is currently available is a better option than waiting for perfect data.

For more than 2 years Butte’s laboratory has been analyzing publicly available gene expression data from individuals with diseases and from healthy controls along with data from biological samples treated with drugs and from untreated samples. By comparing the datasets and using a method based on the Connectivity Map strategy, Butte and his colleagues have been able to identify possible drug targets for diseases of interest (Lamb et al., 2006; Sirota et al., 2011). Disease-based computational strategies for drug repurposing have been used by Butte and several other groups as well (Dudley et al., 2011a).

To follow up on promising computational leads, Butte’s laboratory has been contracting out for research services through commercial websites such as assaydepot.com. The CROs also provide access to a wide variety of animal disease models, including a mouse model of diabetes that Butte’s laboratory has used extensively. For example, results from a 16-mouse diabetes drug study that include data on fasting blood sugar, glucose tolerance, insulin tolerance, and other measures can be delivered within weeks for $9,000. Butte and his colleagues design the protocols, but they no longer have to do the experiments themselves. If the robustness of the results needs to be verified, the researchers simply have a test done in more than one laboratory, he said.

Using publicly available data to identify drug targets and then testing these targets by outsourcing the animal work is an extremely fast way to test and reposition drugs, Butte said. As an example from his own laboratory, he described the prediction that the epilepsy drug topiramate could help treat inflammatory bowel disease or Crohn’s disease (Dudley et al., 2011b). Parts of this work were outsourced to other laboratories and to CROs, including a laboratory in Massachusetts that performs rat colonoscopies. Other drug repositioning efforts are also under way, such as an ongoing study of an antidepressant effective against small-cell lung cancer in mice. Just 15 months after the computational prediction was made, institutional review board (IRB) approval for clinical trials was obtained, and two patients at Stanford are already on the trial, Butte said.

A Data-Rich Future

In the future it will be much more common for drug repurposing to be accelerated by findings from existing public data, predicted Butte, who has co-founded a company to commercialize discoveries made in his laboratory. One contributing factor will be the steadily increasing quantities of data available through PubChem, the Library of Integrated Network-Based Cellular Signals (LINCS) of NIH, the Immunology Database and Analysis Portal (ImmPort), and many other databases. Clinical trial data also will become increasingly available, including data from trials that fail but still yield data useful for drug repositioning.

Discoveries of ways to repurpose drugs do not happen automatically, Butte said. Ensuring such discoveries will require the development of a new generation of investigators who “own” their research findings and follow them through to validation, which in turn may require the cultivation of investigators who are interested in this kind of work and the creation of incentives to encourage researchers to share their data openly.

ROLE OF BASIC SCIENCE FOR TRANSLATION

The typical mission of a university researcher is to focus on research, service, and education, which leads to publications, funding, tenure, and, increasingly, a stake in intellectual property, said Larry Sklar, regents professor of pathology at the University of New Mexico. In recent years the translation of research results to clinical applications has also become a significant priority for university professors. The translation process

has many components, including commercialization, technology transfer, and economic development, Sklar said.

The increased emphasis on translation has changed the way that research scientists work. Collaborators, funding agencies, and patients all have become clients in the context of the academic mission, Sklar said. At the University of New Mexico School of Medicine, for instance, cycles of opportunity have occurred for the development of instrumentation, involvement with NIH programs, and the creation of molecular libraries. These opportunities corresponded with a steady stream of new programs at NIH focused on translation, including the biomedical engineering consortia, the Molecular Libraries Program, the National Cancer Institute Experimental Therapeutics Program, and the Clinical and Translational Science Award consortia. These programs provided a motivation for academicians to “move molecules to clinical trials,” Sklar said. Simultaneously, translation was driven by an increase in drug discovery meetings, funding initiatives, compound collections, and screening technologies; the development of new pharmaceutical and biotechnology business models; and activities sponsored by such groups as the Academic Drug Discovery Consortium and the International Chemical Biology Society.

Technology-Enabled Repurposing

The University of New Mexico School of Medicine has used these new initiatives to move aggressively into drug repurposing by using computer modeling and informatics approaches. As one example, Sklar cited the use of a database called DRUGS database, a licensed resource developed at the University of New Mexico Health Sciences Center that contains 4,414 active pharmaceutical ingredients, accumulated from more than 44,000 FDA drug labels and about 58,000 National Drug Codes, which are annotated to 3,117 protein targets.¹ Sklar and colleagues have successfully used this database to map drugs, indications, and targets using chemical structures of drugs and target bioactivity (Oprea et al., 2011b).

Academic research contributes to repurposing in various ways other than simply testing drugs against already known targets and already available screening technologies. In particular, academic research also contributes to the development and application of new technologies. For

______________________

¹Database contents as of January 27, 2014 (personal communication with Larry Sklar). The database is available to academic collaborators and on a fee basis for industry users.

example, a high-throughput flow cytometry technology developed at the University of New Mexico School of Medicine has been used to find new targets for existing drugs, Sklar reported (Kuckuck et al., 2001). Flow cytometric analyses of competitive inhibitors that bind to small molecular weight GTPases (enzymes that hydrolyze guanosine triphosphate) on fluorescent beads have identified new uses for ketorolac (Oprea et al., 2011a). Ketorolac is a nonsteriodal anti-inflammatory drug approved for the treatment of acute pain. By using this flow cytometry approach, it was also discovered that ketorolac may inhibit GTPase signaling to regulate cell growth in ovarian cancer (Agola et al., 2012; University of New Mexico Cancer Center, 2012).

Through the use of computational analysis to model the docking of compounds into metnase, raltegravir, a DNA (deoxyribonucleic acid) repair enzyme that is associated with chemotherapy resistance when overexpressed in malignant cells was found to inhibit metnase activity (Williamson et al., 2012). Raltegravir was originally approved as an integrase inhibitor for the treatment of HIV infections. Currently raltegravir is being evaluated for its effectiveness as an adjuvant in treating squamous cell carcinoma of the head and neck,² Sklar said.

It is typical to find new activities for existing drugs during repurposing screens and mechanism of action studies, Sklar said. Academicians can play a significant role in contributing to this type of translational research, and resources that support discovery technologies and collaborations among basic scientists and clinicians are helpful in achieving this role.

DATA MINING

Although drug development is a complex and intricate process, it generally follows one of two basic approaches, explained Lon Cardon, senior vice president of alternative discovery and development at GlaxoSmithKline. The first approach is to start with a target and then try to find chemical entities that alter the target. The second is to begin with a phenotype and work toward the identification of a target and mechanism that can be altered.

In order to repurpose a drug it is important to understand both the drug’s targets and the mechanisms of its action, Cardon said. New re-

______________________

²Pilot Study of Raltegravir and Cisplatin in Squamous Cell Carcinoma of Head and Neck (ClinicalTrials.gov identifier, NCT01275183).

search findings, animal model experiments, and modifications of existing therapies are all ways of developing information on targets, but today this approach is far from optimal. If a drug is going to fail in development, Cardon said, it should fail early instead of late in order to reduce the costs of developing drugs that do not work in the first place. However, it is often the case that not enough information is available about targets and mechanisms to predict which drugs ultimately will fail.

Pharmaceutical companies already invest a fair amount in repurposing, but it is not necessarily easier to repurpose a drug than to develop one de novo, Cardon said. A single drug target may be the focus of five to seven projects looking at different indications, but little may emerge from the efforts.

The Potential for Genomics to Guide Repurposing

The recent study of genomics has been following what Cardon referred to as the “hype cycle.”³ In the early 1990s, many geneticists predicted that genomics would revolutionize medicine by identifying new drug targets and personalizing treatments, but when these predictions did not soon become a reality, the enthusiasm surrounding the field faded. Since reaching a low point in 2006, the field has continued to make progress and is now gaining momentum in what Cardon referred to as the “enlightenment phase” of the hype cycle. For example, data from genome-wide association studies (GWAS), which not long ago were being characterized as often disappointing and lacking biological relevance, have become more valuable. A catalog of GWAS can identify genes associated with diseases, such as type 2 diabetes, that are also the targets of ongoing drug development efforts (NHGRI, 2014). By combining this type of information, drug development and repositioning efforts can become more efficacious, Cardon said (Sanseau et al., 2012).

As more data are generated and made available, the number of repurposing hypotheses that can be constructed will increase dramatically, Cardon noted. These can be integrated to reveal new targets and new drugs for repositioning that were never considered before. In particular,

______________________

³The hype cycle is a graphical representation of the interest and enthusiasm related to an emerging technology as it matures through its life cycle. The information can be used as a tool to manage the commercial viability of a product. The hype cycle was described by Gartner, Inc. (Stamford, Connecticut) (Fenn and Raskino, 2008).

Cardon said, resequencing can uncover genetic variants that might point to new indications even for well known therapeutics.

Collaboration Models

Drug repurposing is a challenging process and is neither straightforward nor guaranteed, but the chances of success can be increased by collaboration. Drug development is not the exclusive domain of either industry or academia, Cardon said. Efforts such as the Innovative Medicines Initiative in Europe and the National Center for Advancing Translational Sciences (NCATS) in the United States are demonstrating that an effective way to develop drugs is through collaborations. Rapid progress may require rethinking established practice in such areas as intellectual property, data sharing, and transparency. Industry cannot just support academic researchers and hope that something of value is generated, and academic researchers cannot just look for industrial funds to support what they were already doing. New models are needed that will play to the strengths of pharmaceutical companies, biotechnology companies, academic researchers, venture capitalists, and others, which will require give and take from all sides, Cardon said.

Pharmaceutical companies have data from high-throughput screening that academic researchers do not have, Cardon said. The barriers to the use of those data have to be broken down, so that the two groups can work collaboratively rather than on parallel tracks. For example, several years ago GlaxoSmithKline put its high-throughput screens for potential malaria drugs into the public domain. If ways could be found to make such data available for common chronic conditions, research would take a huge step forward.

Electronic Medical Records and Biobanks

EMRs can be important tools for identifying new uses for existing drugs (Hurle et al., 2013). These records could be another source of data on patient outcomes after treatment if ways could be found for researchers to access that information, Butte said. (See Chapter 6 for further discussion on EMRs.) Furthermore, Butte said, the use of data from EMRs could provide the statistical power needed to better understand drug effectiveness, but improved ways to mine the data contained in these records are needed.

The China Kadoorie Biobank⁴ study enrolled more than 500,000 adults and is collecting clinical data and biospecimens with the goal of identifying risk factors for chronic disease, Cardon said. The data collected from this study could also be a valuable tool for identifying new indications for drugs. For example, GlaxoSmithKline is evaluating cardiovascular outcomes for patients taking darapladib, an Lp-PLA2 inhibitor used to treat coronary heart disease, for the treatment of atherosclerosis.⁵ Because the prevalence of a loss-of-function mutation in phospholipase A2, group VII in some Asian populations is more than 10 percent, Cardon suggested that by examining the genotypes in the Kadoorie study and querying corresponding EMRs, useful information about safety and other indications for Lp-PLA2 may be obtained (Jang et al., 2011). Information from studies that collect biological specimens and EMRs may also provide opportunities for targeted trial recruitment and for the study of other rare loss-of-function variants.

REPURPOSING AT NCATS

The Therapeutics for Rare and Neglected Diseases (TRND) program is a collaborative drug discovery and development research program, not a grant program, in the preclinical development space, said John McKew, acting scientific director in the Division of Pre-Clinical Innovation at NCATS. Projects may enter at various stages of preclinical development, but the diseases being studied must meet FDA orphan or World Health Organization neglected tropical disease criteria, he said. Using either internal resources or government contracts, collaborative projects are taken to the stage of the development process needed to attract an external organization that can complete clinical development and registration. A wide range of small molecules and biologics have been investigated, and a wide range of collaborators are involved. The program also serves to develop new, generally applicable platform technologies and paradigms for drug discovery, including informatics, communications, and collaboration tools that could have widespread benefits for drug repurposing.

______________________

⁴China Kadoorie Biobank, http://www.ckbiobank.org (accessed January 29, 2014).

⁵In November 2013, GlaxoSmithKline announced that darapladib did not reach the primary endpoint of a statistically significant difference between treatment groups (i.e., time to a first major adverse cardiovascular event). The company indicated that it would continue to examine the role of Lp-PLA2 inhibition in heart disease and for other indications (see GSK, 2013).

Screening-Enabled Repurposing

Four projects within the TRND portfolio are focused exclusively on repurposing, McKew said: three small molecule repurposing projects and one biologics repurposing project. An important technique used for these projects is screening-based repurposing.

The screening strategy at TRND is to use phenotypic cell-based assays, such as immunofluorescence imaging, as useful tools for scanning the collection of drugs. Using a high-throughput screening assay, compounds are tested in vitro on cells containing reporter genes whose products produce light when the drug reacts in the cell. The initial screen uses primary patient-derived cells that express the disease phenotype; recently, induced pluripotent stem cells that are differentiated to the appropriate cell types have been used. When it is possible to use, McKew said, this approach is the best way to examine the potential of a compound in vitro.

Screening can have several outcomes, McKew said. It can serve as a tool to probe disease pharmacology or new targets, it can identify a compound that acts as a starting point for a chemistry optimization program, or it can yield an approved drug that can be tested directly in patients in the clinic, though existing data and intellectual property protections might need to be weighed in order to decide the path forward. For example, is it a weakly potent molecule that may require significant additional data to augment the existing drug master file? Can method-of-treatment claims or other intellectual property patents be filed? Overall, the drug screening process identifies interesting molecules and helps build a package that can attract partners to finish the development of the molecules.

As an example of new technologies that can advance repurposing, McKew mentioned matrix screening, or mechanism interrogation plates, which searches for drug synergy among compounds with known mechanisms of action. Identifying drug synergies can make it possible to use less of an individual drug, which can allay concerns about toxicity.

Niemann–Pick Type C Disease

McKew described the Niemann–Pick Type C project, which has been conducted through a collaboration among Johnson & Johnson, the Albert Einstein College of Medicine, the University of Pennsylvania, Washington University, and several NIH institutes. The repurposing screening set consists of about 4,000 molecules representing approved

drugs in Canada, Europe, and the United States (Huang et al., 2011), which has more recently been augmented with late-stage clinical compounds. Niemann–Pick Type C disease is an autosomal recessive disorder with a prevalence of about 1 in 150,000 in western Europe, and patients with this disease usually survive until the second or third decade of life (Patterson, 2013). Defects associated with Niemann–Pick disease, Type C1 or Niemann–Pick Type C2 cause cholesterol and other lipids to accumulate in lysosomes. This aberrant lipid accumulation leads to an enlargement of the spleen and liver and progressive neurological deficiencies, including cerebellar ataxia, dysarthria, dysphagia, tremor, and seizures. No FDA-approved therapies exist; miglustat has been approved in Europe, but it is not a target-specific treatment, McKew said.

Through a grant from the Ara Parseghian Medical Research Foundation, the repurposing collection was screened. This collection included 60 drugs reported in the literature to have an impact on Niemann–Pick Type C disease, McKew said. Skin biopsies from 58 different patients had been genotyped, so the underlying genetic defect was known, and multiple patient lines were used for the screening. A number of phenotypic imaging assays were used to measure accumulated cholesterol and the size of the lysosomes.

The most promising molecule that emerged from the screen, known as 2-hydroxypropyl-b-cyclodextrin (HPBCD), had been approved not as a therapeutic but as an excipient, which added to the challenge of approving the drug, McKew said. HPBCD does not cross the blood–brain barrier, so the treatment had to be adapted for a neurological disorder. This involved working with FDA to demonstrate that the drug would not be toxic if injected into the central nervous system. The drug received an orphan designation from FDA and the European Medicines Agency in 2013, and a Phase I clinical trial had just begun at the NIH Clinical Center at the time of the workshop, McKew said. The intention, he said, is to put together a package of data and incentives that will motivate a drug developer to take the project forward.

Challenges

Cost-related issues of drug repurposing can be a challenge, said McKew. The screening drug collection has been expensive to create and maintain. Because the screening plates are not distributed, researchers must work in collaboration with NIH to gain access to the plates, McKew said. Furthermore, incentives may not exist to move a generic

compound through enough clinical study to effect a label change. A robust, published study may result in the drug being prescribed off-label, and this reduces the incentive for a company to spend the money to change the label with the FDA. In addition, the costs of some repurposing candidates can be prohibitive if the originator is not willing to participate by donating the molecule, McKew said.

PATIENT-REPORTED DATA

Drug repurposing projects have uncovered dysfunction in the clinical research enterprise, said Petra Kaufmann, director of the Office of Clinical Research at the National Institute of Neurological Disorders and Stroke (NINDS). The most important stakeholders—that is, patients—can be frustrated with the lack of treatment options and therefore are often willing to bypass parts of the current drug development process.

Lithium is a mood stabilizer with presumed neuroprotective properties that is thought to promote autophagy, a process of cellular destruction, Kaufmann explained. Mouse studies suggested that lithium could increase survival in patients with amyotrophic lateral sclerosis (ALS), and a small trial in humans found a delayed progression of the disease (Fornai et al., 2008). The morning after the results of this trial were released, Kaufmann said, her phone did not stop ringing, and her patients could not understand why she was not comfortable prescribing lithium for them. But with only 44 patients in the trial—16 of whom received riluzole (a drug used to slow the progression of ALS) and lithium, and 28 of whom received riluzole only—the trial did not have enough power to establish efficacy, Kaufmann said, especially with a disease as variable as ALS.

Many of Kaufmann’s patients obtained lithium from another source and started taking it anyway. Patients using lithium would share their ALS Function Ratings Scale—a measure of a patient’s ability to complete daily living activities—on the website PatientsLikeMe. Within a few months more than 100 people on the site reported taking lithium for ALS, but the results were not interpretable, Kaufmann said. A user of PatientsLikeMe wrote this about the online study: “The study is what it is, whether you are a proponent or a critic makes no difference at the end of the day. Does it have its flaws? Sure. Does it have its redeeming points? Sure. It is a piece of evidence for people to use in their own judgments, nothing more, nothing less” (Frost et al., 2008).

At the same time NINDS, the ALS Society, and the ALS Society of Canada funded a randomized, controlled, double-blind study of 84 patients, which was stopped for futility at the first pre-planned interim analysis (Aggarwal et al., 2010). Other trials conducted in the Netherlands, the United Kingdom, and the United States did not find a benefit to treating ALS with lithium (Chio and Mora, 2013). Close to 700 patients were enrolled in these trials, with many more taking lithium outside of trials.

Patient Engagement

Having enough rare disease patients to fill a traditional large-scale, longer-term clinical trial can be difficult, Kaufmann said. Because recruitment takes a long time, patients can become frustrated and decide not to participate. Innovative approaches for evaluating drugs, such as a futility trial, which seeks to show that a drug does not work rather than to prove it does, requires fewer patients, less time, and less statistical power. This is the kind of innovative protocol that needs to be used for rare diseases, she said.

Opportunities to gain information from the compassionate use of drugs should be used, even though such data are often difficult to interpret, Kaufmann said. Instead of compassionate use, patients could be offered engagement in nontraditional trial designs. The key is to be more nimble in getting patients access to medicine while getting data that can be used as evidence to evaluate whether something works, which will require cooperation among regulators, funders, researchers, and patients. Innovative models that feature transparency and communication can connect stakeholders, including clearinghouses, crowdsourcing, and coordination through patient organizations or funders.

The critical element is public engagement, Kaufmann said. Patients and clinicians need a seat at the table during the development of the disease treatment concept and protocol, she said. For example, in some of the projects supported by the Network for Excellence in Neuroscience Clinical Trials (NeuroNEXT) program at NINDS, researchers are required to involve patients from the beginning of the drug study.⁶ This bidirectional communication makes research projects more successful and aids in information dissemination. Patients feel more comfortable and confident being part of research endeavors, including the repurpos-

______________________

⁶More information about the program is available at www.neuronext.org (accessed May 1, 2014).

ing of existing drugs. Furthermore, rare diseases have so few patients that great efficiency is needed both in enlisting subjects for studies and in conducting studies.

One way to overcome the interpretation issues of self-reported data encountered in the ALS efforts is to use a different approach to using social media to learn more about potential treatments, Kaufmann said. In general, she said, patients need more opportunities to participate in working groups, data monitoring boards, and other activities that can speed up drug development. But patients also often need clear instruction regarding what kinds of data are needed, or the quality of the data and followup may suffer. It works well when patients and clinicians work together to participate in a research opportunity so that data are more objective and curated. For example, through the use of a restricted social media group known as Ning,⁷ patients affected by ALS can ask questions about alternative off-label treatments and receive information from ALS researchers and physicians who are part of this group membership (Bedlack and Hardiman, 2009). The researchers or physicians would help assess the treatments and share the information they gather with patients.

______________________

⁷Available at ning.com (accessed January 30, 2014).

This page intentionally left blank.