by Jessica Sagers
figures by Brad Wierbowski

The only word Charles Chidsey could think of to describe his situation was hairy. “Moderate hypertrichosis has been observed in five of eight patients on chronic treatment,” he noted in a comment near the conclusion of his study. Chidsey, physician and Associate Professor at the University of Colorado School of Medicine, was puzzled to note that routine administration of a new blood pressure drug resulted in hypertrichosis, or the unexpected growth of thick patches of hair on his patients. Men with bald spots invested in combs. Women were horrified to see carpets sprouting across faces and shoulders. Frustrated, the physician complained to a colleague, Dr. Guinter Kahn, whose eyes lit up with recognition of an untapped opportunity. Kahn casually offered to test the compound, known as minoxidil, in his own lab. Then, in a “renegade experiment” without Chidsey’s knowledge or institutional approval, Kahn and his medical resident applied minoxidil to their own staff members’ skin. They confirmed the hirsute results and filed a patent application within the same year. After a fifteen-year legal battle, during which Chidsey and Kahn vied for financial control, minoxidil, newly marketed as the hair-growth drug Rogaine, appeared on U.S. pharmacy shelves.

The idea of repackaging an existing drug for another, perhaps more clinically valuable use is not new. Like Rogaine in the 1980s, Viagra was originally a blood pressure drug, opportunistically repurposed when male patients started reporting predictable, long-lasting erections. Traditionally, these secondary effects have been noticed by chance, giving lucky researchers the opportunity to conduct clinical trials to determine whether a given drug can be safely repositioned to treat a different medical condition. But with the rise of next-generation sequencing technology and the age of the $1000 genome, researchers have recently acquired the tools to make smarter decisions about when, why, and how to repurpose a drug (Figure 1).

Why repurpose existing drugs?

Creating a new drug from scratch is an expensive and time-consuming process. First, researchers must identify a potential therapeutic target, or factor that contributes to the course of a given disease. Then, a drug that acts on that target must be created, purified, and tested, both on cells in a dish and in living animals. In order to be approved for clinical trials, this new drug must meet rigorous safety specifications and pass through highly controlled phases of human testing, which can take over a decade and require billions of investment dollars. Unfortunately, most drugs fail during safety or efficacy testing, and only 10 of every 100 ever make it to the clinic. Though drugs are known to act widely throughout the body, and a physician can prescribe any approved drug to a patient for any reason, a drug is generally developed and prescribed to treat one specific condition. But what if rather than designing a new medication for every disease, there was a way to make better use of the ones we already have? Perhaps a drug approved for severe asthma has the potential to work well against a rare type of tumor, or a drug in use for acute schizophrenia could do wonders for restless leg syndrome. Of course, throwing a mental dart at the Physician’s Desk Reference and prescribing unrelated chemicals to suffering patients is both irresponsible and dangerous. So, how might physicians and researchers be able to make smart repurposing decisions?

Genetics, computers, big data, oh my!

You may already know that the reason your face looks different from your mom’s, your friend’s, or your dog’s comes down to variations in your personal set of genes. Genes are the “building blocks of life,” the strings of nucleotide bases stored in your DNA. Inside your cells, DNA codes the instructions for RNA, a complementary nucleic acid that provides the instructions necessary to assemble proteins. Each person has a unique set of variations within their genetic code, which they inherit from their parents. This means that even for disorders like Huntington’s disease, which are caused by disruption of a single gene, that gene may be disrupted in different ways in different people. Though these people may end up with the same diagnosis, each will bear a slightly different combination of mistakes in their DNA, RNA, and proteins. These differences may determine whether a patient responds to a given drug, or whether the drug has no effect.

Researchers can identify the individual differences present in patients with the same disease using a technology called next-generation sequencing. From a sample of a person’s cells, which can be taken from the cheek or skin, researchers can extract and read through all the DNA and RNA inside of them. As you might imagine, sequencing generates a massive amount of information, so the problems that complicate this kind of analysis become synonymous with the problems of big data. Large-scale genomic studies, which pool genetic information from thousands of patients, can help researchers identify disease-causing mistakes that are common among members of a large population. But precision medicine for drug repositioning takes genetic analysis one step closer to the individual. Rather than pooling thousands of cases and identifying disease-causing mistakes in the aggregate, a precision medicine approach sets out to match the gene expression profiles of individual patients with known interactions between genes and drugs (Figure 2).

Within a given cell, DNA is continually being transcribed into RNA in a process called gene expression. The record that shows which genes are on or off in the cell at any given moment is known as that cell’s “RNA signature.” When cells are treated with a drug, the expression of one or more genes can change, thus altering the RNA signature of the cell. Thus, by matching changes in gene expression that are known to be caused by a given drug to the genetic mistakes observed in an individual patient, researchers can select for drugs that maximize “signature reversion,” or the functional reversal of a diseased RNA signature. For example, if the expression of gene X is reduced in disease Y, researchers would want to preferentially identify drugs that stimulate the expression of gene X, thus causing a targeted reversal of the disease’s RNA signature. Many prominent groups at Harvard Medical School and the Broad Institute are building computational tools to optimize this process and bring it to the clinic, where clinicians hope to prescribe targeted drugs to patients based on their unique genetic variations.

The future of drug repositioning

Proponents of precision medicine imagine a future doctor’s appointment to go something like this: You come into the clinic complaining of a disease that your physician suspects is genetic, so she collects some cells and sends your sample to a bioinformatician. The bioinformatician extracts your DNA and RNA, reads through your personal genetic and transcriptional (i.e. RNA) codes, and analyzes your specific data in the context of all possible interactions between your personal genetic information and every drug approved for human use. After careful computational screening, she generates a list of safe and potentially effective drugs tailored to match your individual genetic profile, and sends them back to your physician or a designated institutional board, who can help select the compound that most closely meets your needs. Though the drug you select may not be recommended for other patients with your condition, it matches your individual genetic needs, so its potential to work for you is higher than it would be if you were to take a generic drug recommended for all patients with your condition. While such an expensive, individualized approach to patient care may not show up as part of your routine doctor’s appointment anytime soon, this kind of targeted, genetics-based treatment is already in use at a few specially designated institutions. A standout example is the Michigan Oncology Sequencing Center, which provides resources to support physicians and researchers as they integrate genomic and transcriptomic data into the clinical practice of cancer medicine.

Though computational drug repositioning is still in its relative infancy, researchers have already identified new and clinically significant uses for existing drugs. A team of researchers based at Harvard Medical School and Beth Israel Deaconess Medical Center recently used a computational repositioning approach to argue for the clinical repurposing of pentamidine, a drug classically prescribed for pneumonia, to treat a metastatic form of kidney cancer. Pentamidine was shown to be effective against cancer cells in the laboratory and in a mouse model, suggesting great promise for human trials. Because the drug is already known to be safe for human use, these trials could potentially be initiated very quickly. And in February of this year, researchers in New York City used a similar type of analysis to suggest the repurposing of metformin, a drug historically prescribed to treat type 2 diabetes, for a genetically specific population of cancer patients. The promises of computational drug repurposing and precision medicine may sound like they belong to the distant future, but based on these exciting results, that future might be closer than we think. Prepare your genomes, folks; we’re heading into bioinformatics hyperdrive.

Jessica Sagers is a second-year PhD student in the Harvard-MIT Program in Speech and Hearing Bioscience and Technology. She serves as Cambridge chair of Harvard Graduate Women in Science and Engineering and studies auditory neuroscience in the Konstantina Stankovic lab at Massachusetts Eye and Ear. Learn more at scholar.harvard.edu/jsagers.

For more information:

Research papers describing computational methods for drug repositioning can be found here and here.

To read more about the use of genomics in prescribing the best cancer treatments, see the Michigan Oncology Sequencing Center.