On the evening of June 12, 2015, Donna Emley took two acetaminophen (Tylenol) for a slight muscle ache and went to bed.
The next day, she and her husband drove to Kentucky, where they were planning to spend a week at an organic farm. She awoke at 2 a.m. the following day and noticed that her eyes were nearly swollen shut and there were rash-like red spots on her face, chest, neck and torso. Thinking this was just an allergic reaction to something she had been exposed to earlier that day on the farm, she went back to bed. Two hours later, her symptoms had worsened, so she and her husband drove to the closest urgent care clinic.
She developed a 104-degree fever and the rash that had been dots turned into larger red spots, which became blisters, and then wounds similar to burns. She developed double vision; her eyes swelled shut and filled with mucus.
“It progressed so quickly,” Emley said. “It was such a shock, and came from out of nowhere. You don’t get up in the morning and even imagine this could happen to you. It was all such a whirlwind.”
Emley had a rare but cataclysmic reaction to the acetaminophen she’d taken.
Called Stevens-Johnson Syndrome/toxic epidermal necrolysis (SJS/TEN), it’s an extreme example of serious adverse drug reactions (ADRs) that can cause severe short- and long-term harm, are enormously expensive and difficult to treat, and are all-too-commonly fatal.
In the United States, ADRs are often cited as the fourth leading cause of death among hospitalized patients after heart disease, stroke and cancer. Treating and helping patients recover from these reactions consumes $136 billion in health care resources every year, and that’s likely an underestimate.
Researchers at Vanderbilt University Medical Center are meeting the challenge. Supported by a five-year, $12.8-million grant awarded last summer by the National Institutes of Health (NIH), they have launched a three-pronged approach, using the latest genetic, stem cell and data mining technologies, to break the ADR code.
It’s all part of a national effort in precision or personalized medicine, tailoring medical care to the unique characteristics of individual patients to improve outcomes, avoid complications and ultimately cut costs.
“Our goal is to understand the fundamental mechanisms putting patients at risk for severe adverse drug reactions,” said Dan Roden, M.D., co-principal investigator of the grant with Josh Denny, M.D., M.S., and Elizabeth Phillips, M.D.
“More broadly, we want to be able to predict how individual patients will respond to drug therapy, and to identify new drugs that carry ADR risk before they get to the market, or even before they get too far in the development process,” he said.
Roden, assistant vice chancellor for Personalized Medicine and the William Stokes Professor of Experimental Therapeutics, has spent the last 20 years investigating a potentially lethal side effect of a panoply of commonly prescribed medications—abnormal heart rhythms, or arrhythmias.
He and his colleagues are “growing” heart cells in the laboratory from patients who’ve experienced drug-induced arrhythmias, and comparing them to cells from those who haven’t. Understanding these heart problems at a cellular level will help drug companies develop safer drugs.
Phillips, the John A. Oates Professor of Clinical Research, is an internationally known expert on equally dangerous hypersensitivity and immune-mediated responses to medications, including the devastating skin reaction that sent Emley to the burn unit.
She and her colleagues have shown that careful screening using sophisticated genetic and cellular techniques in some cases can help patients avoid these catastrophic ADRs and define what drugs they can safely take in the future.
These and other tests conducted in the Vanderbilt Drug Allergy Clinic, which Phillips established in 2014, also can clear patients who haven’t been able to take useful drugs like penicillin because they either did not have a true allergy or lost the tendency to be allergic over time.
Denny, associate professor of Biomedical Informatics and Medicine and a precision medicine consultant to the federal government, is developing methods to phenome scan patients’ electronic health records for clues that can help identify and predict both adverse and beneficial drug actions.
An individual’s phenome is a unique set of physical and behavioral characteristics that are determined by the interaction of their genes and environment. Using this approach, Denny and his colleagues hope to identify additional, repurposed uses for existing drugs, thereby cutting the overall cost of drug development.
These three projects make Vanderbilt a kind of one-stop shop for cutting-edge science in pharmacogenomics and adverse drug reactions. “There are very few institutions anywhere in the world that have the capabilities to execute this broad set of studies,” Roden said.
Smoking Guns
Traditionally, ADRs have been divided into two classes: those that are predictable and related to the action of the drug, and those that are unpredictable and “idiosyncratic.”
Phillips bristles at the term idiosyncratic because it implies “we don’t know what the heck is going on.” She prefers a different classification: “on-target” and “off-target” drug effects.
“On-target” effects, which constitute 80 percent of all adverse reactions, may occur in a patient with a genetic variation that impairs the ability of a liver enzyme to break down a particular drug.
The drug may reach a toxic level—but the toxicity is predictable from what is known about how the drug works.
“Off-target” effects mean something else is going on besides the intended action of the drug. And while they make up only about 20 percent of all ADRs, “they cause more than their share of morbidity and mortality,” Phillips said.
For example, a wide variety of drugs can prolong the QT interval of the electrocardiogram and generate abnormal heart rhythms that can be fatal. Drug-induced long QT syndrome is a major reason drugs receive black-box warning labels or are pulled from the market.
Figuring out who’s at risk for this has proven to be supremely difficult. “It’s not like there’s one smoking gun or even two or three smoking guns,” Roden said. “There tend to be a lot of little smoking guns, and they’re individual per patient and they never account fully for risk.”
That’s why Roden and his colleagues are using a revolutionary new stem-cell technology to create and study heart cells in the lab, and how they react in the presence of drug.
To do that, the researchers take a skin biopsy or blood sample from a patient who’s had one of these extreme reactions. Stem cells in the skin or blood can be induced to become pluripotent, or turn into any kind of cell, including heart cells.
They are then compared to similarly reprogrammed cells from a patient who did not react to the drug. About 10 cell lines have been grown this way so far. By the end of four years, Roden hopes to have 60.
If a defect is identified that can explain the development of drug-induced arrhythmia, it might eventually be possible to edit out the genetic abnormality, and re-populate the patient’s heart with corrected cells.
“It sounds science fiction,” Roden said, “but the editing technologies are here so the scenario may be upon us soon … It’s all about precision, personalized medicine,” he said, “the idea that you really will understand on a fundamental cellular level what makes us different from each other.”
Another example of a complex off-target effect is angioedema, swelling caused by fluid buildup in the tissues which occurs in some people who are given ACE- inhibitors, drugs used to treat high blood pressure and reduce the risk of heart disease and kidney failure.
In the early 1990s, Nancy J. Brown, M.D., chair of the Department of Medicine at Vanderbilt, was among the first to show that African-Americans were at significantly greater risk than their white counterparts to experience drug-induced angioedema.
“We are honing in on some genes and some pathways,” said Brown, the Hugh J. Morgan Professor of Medicine, but more studies in large numbers of patients are needed to sort out all of the contributing factors.
Inherited Risk
Racial differences are an important part of the ADR story.
Asians, for example, are more likely than other racial groups to develop SJS/TEN when they are given carbamazepine, used to treat seizure disorders, or allopurinol, used to treat gout. Among Caucasians and African-Americans, SJS/TEN is also seen more commonly in association with sulfa antibiotics and other anticonvulsants such as lamotrigine.
SJS/TEN often begins with flu-like symptoms, followed by a painful red or purplish rash that spreads and becomes toxic epidermal necrolysis—blistering off the top layer of skin. It’s a miserable, devastating condition that, among elderly people, has a 50 percent mortality rate.
“Once the horse is out of the barn, there’s no way of stopping this,” said Phillips, who sees about one case a month through the Vanderbilt Drug Allergy Clinic, which she established in 2014. “It doesn’t matter what you use. The strongest drugs that suppress or modulate the immune system … are not going to stop this.”
But there is a way to prevent it—by screening the patient for genetic variations in the HLA system, the complex set of proteins that help regulate the immune system.
HLAs, human leukocyte antigens, are several different classes of proteins expressed on the surfaces of cells that, like nametags, help the immune system distinguish “self” tissues of the body, from “non-self,” such as invading pathogens. An individual inherits one set of HLA genes from each parent.
Racial differences in HLAs arose over thousands of years and reflect human migration, Phillips explained. “There are specific (HLA) genes that were very advantageous to have under the pressure of different infectious diseases that occurred in different continents and climates and evolved over time,” she said.
It is the inheritance of these specific HLA genes that may guard an individual against infection on the one hand, but may put them at risk of severe drug toxicity on the other.
Drugs can change the structure and/or “self” proteins that bind to certain HLA markers so they look foreign to the host immune system.
That’s apparently what happens in the case of abacavir, which is used to treat HIV infection but which is associated with a life-threatening hypersensitivity (allergic) reaction. Symptoms can include fever, rash, achiness and fatigue, nausea, vomiting, sore throat and difficulty breathing.
Between 2002-2008, Phillips and her husband Simon Mallal, MBBS, then at Murdoch University in Perth, Australia, discovered a link between this reaction and a specific HLA allele – B*57:01. They were the primary investigators leading the studies that led to the implementation of HLA-B*57:01 screening into routine clinical practice to prevent this severe toxicity.
“The PREDICT-1 study was the first clinical trial that’s ever been done to study the utility of a specific genetic screening test to prevent any drug-related toxicity,” said Phillips, who came to Vanderbilt in 2013 with Mallal, Major E.B. Stahlman Professor of Infectious Diseases and Inflammation.
“It was a major success and illustrative of how effective partnerships between industry and academia can improve drug safety.”
Today, several years after B*57:01 testing was added to the clinical guidelines, true immunologically-mediated abacavir hypersensitivity has been eliminated among patients who are screened.
Screening for another HLA allele, B*15:02, among Asians treated with carbamazepine has had the same impact—SJS/TEN related to carbamazepine “is not a disease that’s commonly seen any more in
Taiwan and other parts of Southeast Asia where screening has been implemented,” she said.
‘That’s what we’d like to see for all markers,” Phillips said. “Unfortunately not all of them have the predictive value that would facilitate their immediate and widespread application into clinical practice.”
Predicting Drug Response
Vanderbilt is uniquely suited to tease out the complexity of ADRs because of its long-standing investment in pharmacogenomics, how variations in the genetic code can affect response to drug treatment, and biomedical informatics, the use of biomedical data and knowledge in research and clinical settings to improve human health.
Roden, Denny and their colleagues have played key roles in the development of research tools and projects at Vanderbilt that are at the cutting edge of these fields. They include:
- PREDICT (Pharmacogenomic Resource for Enhanced Decisions in Care & Treatment), a project spearheaded by Roden that genotypes the DNA of Vanderbilt patients for genetic variations that may affect their response to certain drugs;
- BioVU, which, with more than 200,000 DNA samples, is one of the world’s largest leading bio-banks linked to clinical trial data; and
- The Synthetic Derivative, clinical information derived from electronic medical records that has been de-identified, or stripped of personal identifiers.
Using these resources, Denny and his colleagues have pioneered a new approach to studying specific health conditions, including ADRs, called PheWAS—phenome-wide association studies. The phenome is an individual’s collection of phenotypes, observable characteristics resulting from the interaction of their unique genetic make-up and their environment.
“Our goal is to use the wealth of clinical care for a population in their de-identified electronic medical record to learn how to pick the most efficacious drug, predict side effects, and find new uses for old drugs,” Denny said.
“This is a true use of the health care system to learn from itself to take better care of patients in the long run,” he said. “All those visits and medications in the Synthetic Derivative form a rich ‘phenome’ that is linked to DNA in BioVU—all de-identified.”
A PheWAS looks for associations between various phenotypes, from cancer and diabetes to ADRs and even the drugs that patients have been prescribed, and specific genetic variations. The goal is to find new variants that predict adverse reactions, and which also predict potential new uses for existing drugs.
“The results of this study will be to dramatically increase the catalog of genetic predictors of drug response and to create a library of potential repurposing for nearly all medications,” Denny said.
A big challenge is translating these discoveries into clinical practice. The field is moving so fast it is difficult for physicians and their patients to keep up. In Asia, for example, many physicians continue to order a different seizure drug for their patients, even when the B*15:02 screening test is negative, indicating carbamazepine is safe to prescribe, Phillips said.
Patients, too, are afraid to take certain medications because of “warnings” listed on the readouts of their genotypes they ordered from private labs. Yet it is often the case that the genetic tests which are done do not give a complete picture of the potential risk associated with taking the drugs, she said.
Phillips said her clinic spends a lot of time removing drug allergy “labels” that have been placed on patients, often from childhood.
Only one out of every 10 patients labeled with a penicillin allergy, for example, are truly allergic to the drug, and approximately 10 percent of truly penicillin allergic patients per year will lose their penicillin allergy altogether.
Skin tests, in which a small amount of drug is introduced on or just beneath the skin, can help confirm a previously determined allergy by causing a local reaction without exposing the patient to the risks of the drug.
This is important because erroneous allergy labels for drugs such as penicillin prevent patients from getting the drugs most appropriate for their medical condition, or results in them being prescribed medications that are not appropriate.
“It limits choices and may actually lead to public health consequences such as antibiotic resistance,” Phillips said.
An eventual goal is to develop pre-clinical screening that would predict the potential for drugs to cause severe immunologically-mediated drug reactions before their use in patients and before billions of dollars has been spent on drug development.
“Ultimately we may be able to alter the structure of drugs so that they still have efficacy without causing severe off-target effects,” she said. “The end of diseases such as drug-related SJS/TEN, which kills both patients and drugs, would be nice to see.”
– Donna Emley’s story reported by Ashley Culver